{
  "schema_version": "anvil-serving-benchmark-publication/v1",
  "observed_at": "2026-08-03",
  "run_id": "deepseek-agentic-recovery-smoke-20260803-01",
  "benchmark_kind": "agentic",
  "profile": "smoke",
  "endpoint": "http://100.64.0.10:8000/v1",
  "model": {
    "checkpoint": "deepseek-ai/DeepSeek-V4-Flash-0731",
    "revision": "9e165c30e2704aec5d9d593cce3eebd58bbef1cb",
    "served_name": "deepseek-v4-flash-0731-r16-b12x-dspark5-maxseq16-tp2-650k"
  },
  "topology": {
    "worker": "AI-MBP25",
    "serve_host": "Primary Node",
    "gpu": "2x NVIDIA RTX PRO 6000 Blackwell Max-Q",
    "tensor_parallel_size": 2
  },
  "result": {
    "status": "completed",
    "scenario": "tool-recovery-error",
    "score": 0,
    "maximum_score": 1,
    "failure_class": "reasoning_failure",
    "checks": {
      "protocol": "pass",
      "result_incorporation": "pass",
      "recovery": "pass",
      "history": "pass",
      "reasoning": "fail",
      "final_answer": "fail"
    },
    "tool_attempts": 2,
    "interpretation": "The model followed the tool protocol and retried after an injected tool error, but did not produce the required final answer."
  },
  "integrity": {
    "spec_sha256": "ec61ff097b94e770480360d2a2616bdca1be067fb0243b9da06594b88cba241d",
    "artifact_sha256": "b9d336e8709db2660c03d060a9f942afd609e585af0c98f74d9fad650f68693b",
    "stage_sha256": "6659ae77717e3535911820cf674bd69c9878d3e9649d4d9fc4c6dc627e68dac6"
  }
}