| Benchmark | Claude Sonnet 5 | Gemma 4 E2B |
|---|---|---|
| AA Agentic Index | 49.7 | 7.4 |
| AA Intelligence | 55.3 | 10 |
| AA-LCR | 77 | 17 |
| AA-Omniscience | 16.4 | -23.6 |
| Artificial Analysis Coding Index | 71.5 | 8.3 |
| critpt | 16.9 | 0 |
| gdpval | 54.8 | 0 |
| GPQA Diamond | 91.1 | 43.4 |
| HLE | 57.4 | 4.8 |
| MMMU-Pro | 77.3 | 44.6 |
| mrcr_v2_8needle_128k_average | 81.5 | 19.1 |
| OmniScience Accuracy | 40 | 8.2 |
| OmniScience Non-Hallucination | 60.6 | 67.6 |
| scicode | 53.6 | 20.9 |
| TauBench V3 - Banking | 37.3 | 4.5 |
| Terminal-Bench 2.1 | 80.5 | 0.4 |
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.