| Benchmark | Gemini 3.1 Pro | GPT-4o mini |
|---|---|---|
| AA Agentic Index | 23 | 1 |
| AA Intelligence | 47.7 | 7 |
| ARC-AGI-2 | 77.1 | 0 |
| Artificial Analysis Coding Index | 68.8 | 11.4 |
| Fortress | 29.8 | 48.1 |
| gdpval | 23.3 | 0 |
| GPQA Diamond | 94.3 | 42.6 |
| HLE | 51.4 | 4.2 |
| IFBench | 77.1 | 31 |
| LiveCodeBench | 91.7 | 35.5 |
| mmlu | 92.6 | 81.8 |
| MMMU-Pro | 83 | 41.5 |
| scicode | 59 | 22.9 |
| simplebench | 79.6 | 10.7 |
| SWE-bench Verified | 80.6 | 8.7 |
| TauBench V3 - Banking | 21.4 | 2.9 |
| Terminal-Bench 2.1 | 74 | 5.6 |
| τ³-Bench | 67.1 | 2.9 |
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.