| Benchmark | Claude Sonnet 4.5 | o1 |
|---|---|---|
| AA Agentic Index | 50.6 | 31.1 |
| AA Intelligence | 37.4 | 24 |
| AA-LCR | 68.3 | 63.3 |
| AA-Omniscience | -0.1 | -10.6 |
| AIME 2025 | 88 | 81.7 |
| AIR-Bench 2024 | 89.8 | 80 |
| Artificial Analysis Coding Index | 52.1 | 39.7 |
| critpt | 1.1 | 0.3 |
| Fortress | 17 | 19.4 |
| gdpval | 40.4 | 11.5 |
| GPQA Diamond | 83.4 | 75.7 |
| HLE | 19.8 | 7.7 |
| IFBench | 57.3 | 70.3 |
| MathVista | 79.8 | 71.8 |
| mmmlu | 89.1 | 87.7 |
| MMMU | 79.6 | 77.6 |
| OmniScience Accuracy | 32.9 | 34.7 |
| OmniScience Non-Hallucination | 50.9 | 30.7 |
| scicode | 45 | 35.8 |
| SEAL VISTA | 48.8 | 45.3 |
| simplebench | 54.3 | 41.7 |
| SWE-bench Verified | 82 | 48.9 |
| TAU-bench (airline) | 70 | 50 |
| TAU-bench (retail) | 86.2 | 70.8 |
| Terminal-Bench Hard | 35.6 | 12.9 |
| τ²-Bench Telecom (AA run) | 78.1 | 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.