| Benchmark | Claude 3.7 Sonnet | Gemma 4 31B |
|---|---|---|
| AA Agentic Index | 37 | 14.4 |
| AA Intelligence | 27.6 | 29.7 |
| AA-LCR | 62.3 | 68.3 |
| AA-Omniscience | -0.7 | -47.9 |
| arena_vision | 1195 | 1252 |
| Artificial Analysis Coding Index | 36.4 | 43.4 |
| critpt | 0.9 | 1.4 |
| gdpval | 27.4 | 15.5 |
| GPQA Diamond | 84.8 | 85.7 |
| HLE | 9.7 | 26.5 |
| hmmt_feb_2025 | 31.7 | 88.7 |
| IFBench | 48.3 | 75.6 |
| mmmlu | 86.1 | 88.4 |
| MMMU | 75 | 80.4 |
| MMMU-Pro | 60.1 | 76.9 |
| OmniScience Accuracy | 28.2 | 20 |
| OmniScience Non-Hallucination | 60 | 18.1 |
| scicode | 40.3 | 43.4 |
| SWE-bench Verified | 70.3 | 52 |
| Terminal-Bench Hard | 21.2 | 36.4 |
| τ²-Bench Telecom (AA run) | 54.7 | 65.5 |
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.