| Benchmark | Claude 3.7 Sonnet | Gemini 3.1 Pro |
|---|---|---|
| AA Agentic Index | 37 | 23 |
| AA Intelligence | 27.6 | 47.7 |
| AA-LCR | 62.3 | 79 |
| AA-Omniscience | -0.7 | 32.9 |
| arc_agi_1 | 13.6 | 98 |
| ARC-AGI-2 | 0 | 77.1 |
| arena_vision | 1195 | 1280 |
| Artificial Analysis Coding Index | 36.4 | 68.8 |
| critpt | 0.9 | 17.7 |
| Fortress | 38 | 29.8 |
| gdpval | 27.4 | 23.3 |
| GPQA Diamond | 84.8 | 94.3 |
| HLE | 9.7 | 51.4 |
| IFBench | 48.3 | 77.1 |
| mmmlu | 86.1 | 92.6 |
| MMMU-Pro | 60.1 | 83 |
| multichallenge | 51.6 | 71.4 |
| MultiNRC | 27.8 | 64.7 |
| OmniScience Accuracy | 28.2 | 55.3 |
| OmniScience Non-Hallucination | 60 | 50.1 |
| scicode | 40.3 | 59 |
| simplebench | 46.4 | 79.6 |
| SWE-bench Verified | 70.3 | 80.6 |
| Terminal-Bench Hard | 21.2 | 68.5 |
| τ²-Bench Telecom (AA run) | 54.7 | 99.3 |
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.