| Benchmark | Claude Opus 4.8 | Gemma 4 E2B |
|---|---|---|
| AA Agentic Index | 49.4 | 7.4 |
| AA Intelligence | 57.3 | 10 |
| AA-LCR | 73 | 17 |
| AA-Omniscience | 28.8 | -23.6 |
| aime_2026 | 100 | 37.5 |
| Artificial Analysis Coding Index | 74.3 | 8.3 |
| critpt | 20.9 | 0 |
| gdpval | 54.2 | 0 |
| GPQA Diamond | 93.6 | 43.4 |
| HLE | 57.9 | 4.8 |
| IFBench | 62.2 | 38 |
| mathvision | 86.7 | 52.4 |
| MMMU-Pro | 78.9 | 44.6 |
| OmniScience Accuracy | 48.8 | 8.2 |
| OmniScience Non-Hallucination | 60.7 | 67.6 |
| scicode | 53.5 | 20.9 |
| TauBench V3 - Banking | 34.2 | 4.5 |
| Terminal-Bench 2.1 | 85 | 0.4 |
| Terminal-Bench Hard | 58.3 | 3 |
| τ²-Bench Telecom (AA run) | 94.4 | 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.