| Benchmark | Claude Sonnet 4.6 | Gemma 4 E2B |
|---|---|---|
| AA Agentic Index | 61.6 | 7.4 |
| AA Intelligence | 48.4 | 10 |
| AA-LCR | 74 | 17 |
| AA-Omniscience | 12.2 | -23.6 |
| Artificial Analysis Coding Index | 63 | 8.3 |
| critpt | 3.1 | 0 |
| gdpval | 54.8 | 0 |
| GPQA Diamond | 89.9 | 43.4 |
| HLE | 49 | 4.8 |
| IFBench | 56.6 | 38 |
| mmmlu | 89.3 | 67.4 |
| MMMU-Pro | 75.6 | 44.6 |
| mrcr_v2_8needle_128k_average | 84.9 | 19.1 |
| OmniScience Accuracy | 40.9 | 8.2 |
| OmniScience Non-Hallucination | 51.6 | 67.6 |
| scicode | 47 | 20.9 |
| Tau2 retail | 91.7 | 18.9 |
| TauBench V3 - Banking | 34.4 | 4.5 |
| Terminal-Bench 2.1 | 71.2 | 0.4 |
| Terminal-Bench Hard | 59.1 | 3 |
| τ²-Bench Telecom (AA run) | 97.9 | 22.2 |
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.