{
  "schema": "anvil-serving.campaign-workload-manifest/v1",
  "campaign_id": "2026-09-21-mimo-v26",
  "launcher_revision": "974d38c4ad2c19f776f4d3c35f427d28be77567a",
  "generator_sources": [
    {
      "path": "anvil_serving/benchmarking/requests.py",
      "sha256": "6de56fe18a7a046b3cacfa95058d4174e056068b484cc0e798c118aba17a957e"
    },
    {
      "path": "anvil_serving/benchmarking/cli.py",
      "sha256": "9730892176675c01adc37574f43e0fbdaf12451b3a4841abe17bc078367a8154"
    },
    {
      "path": "anvil_serving/_benchmark_profiles/smoke.json",
      "sha256": "b43f6a199164c2cbb92bde24672beebd781f58851d2d1bb944fc3d4d24a8e360"
    }
  ],
  "controlled_capacity": {
    "seed": 0,
    "prompt_cache_mode": "unique",
    "request_canaries": true,
    "response_words": 16,
    "max_tokens": 2048,
    "thinking_mode": "enabled",
    "temperature": 0.0,
    "nominal_context_tokens": 32768,
    "concurrency": 4,
    "scout_requests": 8,
    "finalist_requests": 100,
    "sequence": "eight-request scout followed by100-request final; first8prefixes may remain cached; unique describes within-cell requests",
    "performance_metrics": "service visibleTTFT,E2E,requests/s; short output does not measure sustained long-answer decode"
  },
  "diagnostic_capacity": {
    "requests": 8,
    "concurrency": 1,
    "nominal_context_tokens": 8192,
    "response_words": 128,
    "max_tokens": 10240,
    "outcome": "all8completed by length limit; exact output and complete capture failed; excluded from performance"
  },
  "coding": {
    "dataset": "princeton-nlp/SWE-bench_Verified",
    "revision": "c104f840cc67f8b6eec6f759ebc8b2693d585d4a",
    "instance_ids": [
      "psf__requests-2317"
    ],
    "selection": "preselected before either model attempt; bounded single-repository smoke only",
    "grader": "official pinned SWE-bench tests; independently executable oracle"
  },
  "quality": {
    "suites_requested": [
      "chat",
      "context",
      "tool",
      "session",
      "intelligence"
    ],
    "repetitions": 3,
    "evidence_use": "diagnostic",
    "visible_answer_tokens": 1024,
    "reasoning_headroom_tokens": 8192,
    "thinking_mode": "enabled"
  }
}