| Benchmark | Claude Opus 4.6 | o1 |
|---|---|---|
| AA Agentic Index | 67.6 | 31.1 |
| AA Intelligence | 44.9 | 24 |
| AA-LCR | 74.3 | 63.3 |
| AA-Omniscience | 13.7 | -10.6 |
| AIME 2025 | 99.8 | 81.7 |
| arena_vision | 1300 | 1193 |
| Artificial Analysis Coding Index | 48.1 | 39.7 |
| critpt | 12.6 | 0.3 |
| Fortress | 20.5 | 19.4 |
| gdpval | 55.9 | 11.5 |
| GPQA Diamond | 91.3 | 75.7 |
| GSM8K | 0.6 | 97.1 |
| HLE | 53.1 | 7.7 |
| IFBench | 62.5 | 70.3 |
| livebench | 76.3 | 52.3 |
| metr_hcast | 77 | 42.6 |
| metr_re_bench | 97.4 | 0 |
| metr_swaa | 98.5 | 100 |
| mmmlu | 91.1 | 87.7 |
| OmniScience Accuracy | 47 | 34.7 |
| OmniScience Non-Hallucination | 37.2 | 30.7 |
| scicode | 52 | 35.8 |
| SEAL VISTA | 46.1 | 45.3 |
| simplebench | 67.6 | 41.7 |
| SWE-bench Verified | 80.8 | 48.9 |
| Terminal-Bench Hard | 65.4 | 12.9 |
| τ²-Bench Telecom (AA run) | 99.3 | 62.6 |
Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing.