| Benchmark | o1 | Opus 5 |
|---|---|---|
| AA Agentic Index | 31.1 | 59.2 |
| AA Intelligence | 24 | 63.1 |
| AA-LCR | 63.3 | 78.7 |
| AA-Omniscience | -10.6 | 37.1 |
| Artificial Analysis Coding Index | 39.7 | 78 |
| critpt | 0.3 | 29.1 |
| gdpval | 11.5 | 67.2 |
| GPQA Diamond | 75.7 | 93.7 |
| HLE | 7.7 | 64.7 |
| OmniScience Accuracy | 34.7 | 60.9 |
| OmniScience Non-Hallucination | 30.7 | 40.5 |
| scicode | 35.8 | 55.7 |
| simplebench | 41.7 | 80.6 |
| SWE-bench Verified | 48.9 | 96 |
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.