| Benchmark | Claude 3.5 Sonnet | Claude Opus 4.6 |
|---|---|---|
| AA Intelligence | 10 | 44.9 |
| AIME 2025 | 7.4 | 99.8 |
| arena_vision | 1146 | 1300 |
| Artificial Analysis Coding Index | 30.2 | 48.1 |
| Chinese SimpleQA (C-SimpleQA) | 56.8 | 76.4 |
| Fortress | 13 | 20.5 |
| frontiermath_tier_4 | 0 | 22.9 |
| GPQA Diamond | 67.2 | 91.3 |
| GSM8K | 96.9 | 0.6 |
| HLE | 3.9 | 53.1 |
| LiveCodeBench | 32.8 | 88.8 |
| LiveCodeBench v6 | 37.2 | 88.8 |
| MMLU-Pro | 78 | 89.1 |
| MMMU-Pro | 54.7 | 77.3 |
| RealWorldQA | 60.1 | 73.9 |
| scicode | 36.6 | 52 |
| SEAL VISTA | 38.7 | 46.1 |
| simplebench | 41.4 | 67.6 |
| SWE-bench Verified | 50.8 | 80.8 |
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.