| Benchmark | Gemini 1.5 Pro | Gemma 4 31B |
|---|---|---|
| AA Intelligence | 10 | 29.7 |
| AI2D | 80.9 | 89 |
| arena_vision | 1179 | 1252 |
| Artificial Analysis Coding Index | 23.6 | 43.4 |
| GPQA Diamond | 59.1 | 85.7 |
| HLE | 4.9 | 26.5 |
| MathVista | 70.6 | 79.3 |
| MMLU-Pro | 75.8 | 85.2 |
| MMMU | 68.4 | 80.4 |
| MMMU-Pro | 55 | 76.9 |
| OCRBench | 800 | 86.1 |
| scicode | 29.5 | 43.4 |
| SWE-bench Verified | 34.2 | 52 |
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.