| Benchmark | Claude Opus 4.7 | Gemma 4 E2B |
|---|---|---|
| AA Agentic Index | 64.6 | 7.4 |
| AA Intelligence | 55 | 10 |
| AA-LCR | 75.3 | 17 |
| AA-Omniscience | 27.3 | -23.6 |
| aime_2026 | 95.8 | 37.5 |
| Artificial Analysis Coding Index | 73.6 | 8.3 |
| critpt | 12 | 0 |
| gdpval | 58.6 | 0 |
| GPQA Diamond | 94.2 | 43.4 |
| HLE | 54.7 | 4.8 |
| IFBench | 58.6 | 38 |
| mmmlu | 91.5 | 67.4 |
| MMMU-Pro | 78.8 | 44.6 |
| mrcr_v2_8needle_128k_average | 59.3 | 19.1 |
| OmniScience Accuracy | 48.9 | 8.2 |
| OmniScience Non-Hallucination | 57.7 | 67.6 |
| scicode | 54.5 | 20.9 |
| TauBench V3 - Banking | 34.6 | 4.5 |
| Terminal-Bench 2.1 | 83.1 | 0.4 |
| Terminal-Bench Hard | 54.5 | 3 |
| τ²-Bench Telecom (AA run) | 88.6 | 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.