| Benchmark | Claude 3.5 Sonnet | Claude Opus 4.8 |
|---|---|---|
| AA Intelligence | 10 | 57.3 |
| arena_vision | 1146 | 1280 |
| Artificial Analysis Coding Index | 30.2 | 74.3 |
| Fortress | 13 | 18.2 |
| frontiermath_tier_4 | 0 | 31.3 |
| GPQA Diamond | 67.2 | 93.6 |
| HLE | 3.9 | 57.9 |
| longbench_v2 | 41 | 69.1 |
| MMMU-Pro | 54.7 | 78.9 |
| scicode | 36.6 | 53.5 |
| simplebench | 41.4 | 64.8 |
| SWE-bench Verified | 50.8 | 88.6 |
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.