{"schema_version":1,"total":1,"offset":0,"limit":100,"observations":[{"id":"public:951ce6690d69cd50f95fe408","benchmark_id":"vals-index-cost::2","subject":{"source_id":"openai/gpt-6-astra","name":"openai/gpt-6-astra","model_id":"gpt-6-astra::max","variant":null,"harness":null},"value":19.088776,"unit":"USD","basis":"measured","source":{"url":"https://www.vals.ai/benchmarks/vals_index","retrieved_at":"2026-09-18T13:00:43.826192+00:00","published_at":null,"file":"data/raw/benchmarks/daily-evidence/2026-09-18T12-58-20-378Z/558f8306d2b1cf1a1874.gz","sha256":"558f8306d2b1cf1a1874ebee5c907d17b0776ccc1f9e70fbb764af4470bb7bbb","locator":"astro_props; source row 2; openai/gpt-6-astra; field cost_per_test"},"protocol":"Vals Index v2 (updated 2026-09-10), overall; published cost per test in USD per model slug; accuracy and latency retained as context, not converted.; source row: {\"task\":\"overall\",\"accuracy\":66.608,\"stderr\":1.093,\"cost_per_test\":19.088776,\"latency\":1510.529,\"compute_effort\":null,\"reasoning_effort\":\"max\",\"max_output_tokens\":128000,\"provider\":\"OpenAI\"}","comparison_key":null,"join_note":"Reviewed identity map 2026-09-16: label states model gpt-6-astra and setting max; exact catalog configuration gpt-6-astra::max"}],"divergences":[],"cell":null,"coverage":{"model":null,"benchmark":{"total_models":844,"available":31,"unknown":813,"not_tested":0,"not_published":0,"source_unreachable":0,"contested":0,"measured":31,"self_reported":0,"observations":57,"unmatched_observations":26}},"collection":{"benchmark_id":"vals-index-cost::2","status":"collected","source_url":"https://www.vals.ai/benchmarks/vals_index","reason":"57 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."}