| Benchmark | Claude Opus 4.5 | Gemini 1.5 Pro |
|---|---|---|
| AA Intelligence | 41.9 | 10 |
| ARC-AGI-2 | 37.6 | 0.8 |
| Artificial Analysis Coding Index | 47.8 | 23.6 |
| Fortress | 13.6 | 53.9 |
| GPQA Diamond | 87 | 59.1 |
| HLE | 30.8 | 4.9 |
| LiveCodeBench | 87 | 30.5 |
| MathVista | 80.2 | 70.6 |
| MMLU-Pro | 90 | 75.8 |
| MMMU | 80.7 | 68.4 |
| MMMU-Pro | 74 | 55 |
| OCRBench | 86.5 | 800 |
| scicode | 50 | 29.5 |
| SEAL VISTA | 46.4 | 37.1 |
| simplebench | 62 | 27.1 |
| SWE-bench Verified | 81.5 | 34.2 |
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.