{"schema_version":1,"total":1,"offset":0,"limit":100,"observations":[{"id":"public:b8c2f375b9859d9649f5cd58","benchmark_id":"bullshitbench-v1::snapshot-2026-09-10","subject":{"source_id":"anthropic/claude-haiku-4.5@reasoning=none","name":"anthropic/claude-haiku-4.5@reasoning=none","model_id":"claude-haiku-4.5::non-reasoning","variant":null,"harness":null},"value":0.8182,"unit":"fraction","basis":"measured","source":{"url":"https://raw.githubusercontent.com/petergpt/bullshit-benchmark/2678ac296fe6e234d391e0f4ab339f6a777ba2a3/data/latest/leaderboard.csv","retrieved_at":"2026-09-15T01:07:18.318850+00:00","published_at":null,"file":"data/raw/benchmarks/daily-evidence/2026-09-15-bullshitbench/2605907319773e0e5858.gz","sha256":"2605907319773e0e5858a24bd8c63ec46501ce282082eff3d134283811d2af6f","locator":"csv; source row 10; anthropic/claude-haiku-4.5@reasoning=none; field green_rate"},"protocol":"BullshitBench V1 (snapshot-2026-09-10); 55 nonsense prompts; three-judge panel (Claude Sonnet 4.6, GPT-5.2, Gemini 3.1 Pro Preview) grades clear pushback / partial challenge / accepted nonsense; value = green_rate over all attempts; one row per model x reasoning label; not a Composite input.; reasoning=none; avg_score=1.7333; score_2=45; nonsense_count=55; answered_count=55; refusal_count=0; error_count=0","comparison_key":null,"join_note":"Reviewed identity map 2026-09-16: label states model claude-haiku-4.5 and setting none; exact catalog configuration claude-haiku-4.5::non-reasoning"}],"divergences":[],"cell":null,"coverage":{"model":null,"benchmark":{"total_models":844,"available":69,"unknown":775,"not_tested":0,"not_published":0,"source_unreachable":0,"contested":0,"measured":69,"self_reported":0,"observations":194,"unmatched_observations":125}},"collection":{"benchmark_id":"bullshitbench-v1::snapshot-2026-09-10","status":"collected","source_url":"https://raw.githubusercontent.com/petergpt/bullshit-benchmark/2678ac296fe6e234d391e0f4ab339f6a777ba2a3/data/latest/leaderboard.csv","reason":"194 source results parsed; configurations remain separate; unmatched model identities are retained."},"coverage_note":"Denominators are catalog model configurations × versioned registry entries; capability_available excludes Efficiency (cost) boards. Basis counts count covered cells, not runs; measured and self_reported may overlap. Unmatched source identities remain in observations and never inflate catalog coverage."}