| Benchmark | Claude Opus 4.7 | o1 |
|---|---|---|
| AA Agentic Index | 64.6 | 31.1 |
| AA Intelligence | 55 | 24 |
| AA-LCR | 75.3 | 63.3 |
| AA-Omniscience | 27.3 | -10.6 |
| arena_vision | 1304 | 1193 |
| Artificial Analysis Coding Index | 73.6 | 39.7 |
| critpt | 12 | 0.3 |
| gdpval | 58.6 | 11.5 |
| GPQA Diamond | 94.2 | 75.7 |
| harmbench | 17.5 | 96.3 |
| HLE | 54.7 | 7.7 |
| IFBench | 58.6 | 70.3 |
| livebench | 76.9 | 52.3 |
| mmmlu | 91.5 | 87.7 |
| OmniScience Accuracy | 48.9 | 34.7 |
| OmniScience Non-Hallucination | 57.7 | 30.7 |
| scicode | 54.5 | 35.8 |
| simplebench | 61.7 | 41.7 |
| SWE-bench Verified | 87.6 | 48.9 |
| Terminal-Bench Hard | 54.5 | 12.9 |
| τ²-Bench Telecom (AA run) | 88.6 | 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.