{
  "created_at": "2026-09-23T09:13:21.129Z",
  "models": [
    "nvidia/nemotron-3-super-120b-a12b",
    "nvidia/Nemotron-3_5-Lightning"
  ],
  "reps": 3,
  "cases": [
    "T1",
    "T2",
    "T3",
    "T4",
    "T5",
    "T6"
  ],
  "dev": false,
  "provider": "Nebius Token Factory (OpenAI-compatible chat.completions)",
  "base_url": "https://api.tokenfactory.nebius.com/v1",
  "prompt_version": "agent-prompt-v1.1",
  "adapter_version": "tf-adapter-v1",
  "system_prompt_sha256": "06989b3ca70d692100b0a20f6f60435821e0f858a95a059cbfabcd472003dfab",
  "schema_sha256": "97c7786afdda13d710e29afe811783c413cd0edb7f8e269fcdc57a5e83b3b1da",
  "corpus_sha256": "dfc5a112a711c32aade2b41d1f11fbad8372c2836a808fdd9f01a3f64fad8189",
  "cases_sha256": "5d62376536c17b53858370cfdce47a723787178764ce18b2fe49ebbe9c8ca925",
  "sampling": {
    "temperature": 0.2,
    "max_tokens": 8000,
    "json_mode": true
  },
  "timeout_ms": 90000,
  "retry_policy": "max 1 retry on network/429/5xx with 1.5 s backoff; no retry on 4xx or timeout; no repair pass",
  "concurrency": 1,
  "order": "alternating model order per case and repetition",
  "caching": "provider defaults, not disabled; cached tokens reported only if the API returns them (else N/A)",
  "pricing": {
    "Qwen/Qwen3-235B-A22B-Instruct-2507": {
      "input": 0.19999999999999998,
      "output": 0.6,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "google/gemma-3-27b-it": {
      "input": 0.09999999999999999,
      "output": 0.3,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "Qwen/Qwen3-Embedding-8B": {
      "input": 0.01,
      "output": 0,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "openai/gpt-oss-120b": {
      "input": 0.15,
      "output": 0.6,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "Qwen/Qwen3-30B-A3B-Instruct-2507": {
      "input": 0.09999999999999999,
      "output": 0.3,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "NousResearch/Hermes-4-405B": {
      "input": 1,
      "output": 3,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B": {
      "input": 0.06,
      "output": 0.24,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "zai-org/GLM-5.1": {
      "input": 1.4,
      "output": 4.4,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "nvidia/Nemotron-3_5-Lightning": {
      "input": 0.06,
      "output": 0.24,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "openbmb/MiniCPM-V-4_5": {
      "input": 0.658,
      "output": 1.1099999999999999,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "deepseek-ai/DeepSeek-V4-Flash-0731": {
      "input": 0.14,
      "output": 0.28,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "MiniMaxAI/MiniMax-M3": {
      "input": 0.3,
      "output": 1.2,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "moonshotai/Kimi-K2.7-Code": {
      "input": 0.95,
      "output": 4,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "nvidia/Nemotron-3-Ultra-550b-a55b": {
      "input": 1,
      "output": 3,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "moonshotai/Kimi-K2.6": {
      "input": 0.95,
      "output": 4,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "nvidia/nemotron-3-super-120b-a12b": {
      "input": 0.3,
      "output": 0.8999999999999999,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "deepseek-ai/DeepSeek-V4-Pro": {
      "input": 1.75,
      "output": 3.5,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "zai-org/GLM-5.2": {
      "input": 1.4,
      "output": 4.4,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "moonshotai/Kimi-K3": {
      "input": 3,
      "output": 15,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "zai-org/GLM-5.3-Flash": {
      "input": 0.15,
      "output": 0.5,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "deepseek-ai/DeepSeek-V4.1-Flash": {
      "input": 0.3,
      "output": 1.2,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "zai-org/GLM-5.3": {
      "input": 1.4,
      "output": 4.4,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    },
    "deepseek-ai/DeepSeek-V4-Pro-0813": {
      "input": 1.32,
      "output": 3.9600000000000004,
      "cached_input": null,
      "source": "GET /v1/models?verbose=true at 2026-09-23T09:09:44.251Z (models_2026-09-23T09-09-44.json); API unit USD/token, multiplied by 1e6 once -> USD per 1M tokens; no cached-token price in the API response"
    }
  },
  "run_location": "Accel AI Innovate Amsterdam venue network, MacBook (arm64), sequential calls"
}