| Benchmark | Gemini 1.5 Pro | GPT-5.4 |
|---|---|---|
| AA Intelligence | 10 | 53.1 |
| ARC-AGI-2 | 0.8 | 74 |
| arena_vision | 1179 | 1277 |
| Artificial Analysis Coding Index | 23.6 | 71.1 |
| Chinese SimpleQA (C-SimpleQA) | 59.4 | 76.8 |
| GPQA Diamond | 59.1 | 93 |
| HLE | 4.9 | 43.7 |
| MMLU-Pro | 75.8 | 87.5 |
| MMMU-Pro | 55 | 81.2 |
| scicode | 29.5 | 56.6 |
| SEAL VISTA | 37.1 | 50.9 |
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.