| Benchmark | Claude Opus 4.5 | o1 |
|---|---|---|
| AA Agentic Index | 59.6 | 31.1 |
| AA Intelligence | 41.9 | 24 |
| AA-LCR | 76 | 63.3 |
| AA-Omniscience | 14 | -10.6 |
| AIME 2025 | 92.8 | 81.7 |
| Artificial Analysis Coding Index | 47.8 | 39.7 |
| critpt | 4.6 | 0.3 |
| Fortress | 13.6 | 19.4 |
| gdpval | 47.3 | 11.5 |
| GPQA Diamond | 87 | 75.7 |
| HLE | 30.8 | 7.7 |
| IFBench | 58 | 70.3 |
| livebench | 76 | 52.3 |
| MathVista | 80.2 | 71.8 |
| metr_hcast | 73.4 | 42.6 |
| metr_re_bench | 40 | 0 |
| metr_swaa | 99.5 | 100 |
| mmmlu | 90.8 | 87.7 |
| MMMU | 80.7 | 77.6 |
| OmniScience Accuracy | 46.6 | 34.7 |
| OmniScience Non-Hallucination | 39 | 30.7 |
| scicode | 50 | 35.8 |
| SEAL VISTA | 46.4 | 45.3 |
| simplebench | 62 | 41.7 |
| SWE-bench Verified | 81.5 | 48.9 |
| Terminal-Bench Hard | 47 | 12.9 |
| τ²-Bench Telecom (AA run) | 89.5 | 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.