| Benchmark | Claude 3.5 Sonnet | Gemini 2.5 Flash |
|---|---|---|
| AA Intelligence | 10 | 17.7 |
| aider_polyglot | 45.3 | 61.9 |
| AIME 2024 | 16 | 82.3 |
| AIME 2025 | 7.4 | 72 |
| AIR-Bench 2024 | 85.9 | 68.7 |
| anthropic_red_team | 99.8 | 98.8 |
| arena_vision | 1146 | 1214 |
| Artificial Analysis Coding Index | 30.2 | 22.2 |
| bbq | 94.9 | 97.7 |
| frontiermath_tier_4 | 0 | 4.2 |
| GPQA Diamond | 67.2 | 82.8 |
| harmbench | 98.1 | 62.6 |
| HLE | 3.9 | 12.1 |
| hmmt_feb_2025 | 1.7 | 64.2 |
| LiveCodeBench | 32.8 | 76.2 |
| MMMU | 72 | 79.7 |
| MMMU-Pro | 54.7 | 69.1 |
| scicode | 36.6 | 35.9 |
| SEAL VISTA | 38.7 | 49.1 |
| simple_safety_tests | 100 | 98 |
| simplebench | 41.4 | 41.2 |
| simpleqa | 28.4 | 26.9 |
| SWE-bench Verified | 50.8 | 60.4 |
| xstest | 95.6 | 98.8 |
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.