| Benchmark | Gemini 1.5 Pro | Gemini 3.6 Flash |
|---|---|---|
| AA Intelligence | 10 | 52 |
| ARC-AGI-2 | 0.8 | 2.6 |
| Artificial Analysis Coding Index | 23.6 | 69.2 |
| GPQA Diamond | 59.1 | 92.8 |
| HLE | 4.9 | 40.8 |
| MMMU-Pro | 55 | 83.2 |
| scicode | 29.5 | 52.7 |
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.