{
  "schemaVersion": 1,
  "resultVersion": "v1",
  "generatedBy": "scripts/build-backrooms-results-table.mjs",
  "generatedAt": "2026-09-21T21:43:49.599Z",
  "dataUpdatedAt": "2026-09-20",
  "canonicalUrl": "https://benchmarks.bio/benchmarks/biosecbench-surveillance/",
  "benchmark": {
    "id": "biosecbench-surveillance",
    "label": "BioSecBench-Surveillance",
    "category": "biosecurity",
    "horizon": "short",
    "evalCount": 102,
    "scoreScope": {
      "label": "Full benchmark",
      "evaluationCount": 102
    },
    "weight": 1,
    "description": "Biosecurity surveillance tasks that measure an agent's ability to analyze biological threat signals.",
    "repository": "https://github.com/latchbio/biosecbench-surveillance",
    "paper": "https://latch.bio/biosecbench-surveillance",
    "paperPdfUrl": "https://benchmarks.bio/papers/biosecbench-surveillance.pdf",
    "interactiveUrl": "https://benchmarks.bio/surveillance"
  },
  "scoreDefinition": "Published benchmark score on a 0–100 percentage scale using Full benchmark (102 evaluations); see the repository and paper for benchmark-specific grading methodology.",
  "results": [
    {
      "rank": 1,
      "id": "claude-opus-5-pi",
      "model": "Claude Opus 5",
      "harness": "Pi",
      "provider": "Anthropic",
      "scorePct": 51.5
    },
    {
      "rank": 2,
      "id": "gpt-6-astra-pi",
      "model": "GPT-6 Astra",
      "harness": "Pi",
      "provider": "OpenAI",
      "scorePct": 48.5
    },
    {
      "rank": 3,
      "id": "grok-4-6-grok-build",
      "model": "Grok 4.6",
      "harness": "Grok Build",
      "provider": "SpaceXAI",
      "scorePct": 48
    },
    {
      "rank": 4,
      "id": "grok-4-6-pi",
      "model": "Grok 4.6",
      "harness": "Pi",
      "provider": "SpaceXAI",
      "scorePct": 47.5
    },
    {
      "rank": 5,
      "id": "claude-opus-5-claude-code",
      "model": "Claude Opus 5",
      "harness": "Claude Code",
      "provider": "Anthropic",
      "scorePct": 43.6
    },
    {
      "rank": 6,
      "id": "gpt-6-astra-codex",
      "model": "GPT-6 Astra",
      "harness": "Codex",
      "provider": "OpenAI",
      "scorePct": 35.2
    }
  ]
}