| Benchmark | Gemma 4 E2B | o4-mini |
|---|---|---|
| AA Agentic Index | 7.4 | 36.1 |
| AA Intelligence | 10 | 26.1 |
| AA-LCR | 17 | 60 |
| AA-Omniscience | -23.6 | -35.7 |
| Artificial Analysis Coding Index | 8.3 | 25.6 |
| critpt | 0 | 0.6 |
| gdpval | 0 | 25.4 |
| GPQA Diamond | 43.4 | 81.4 |
| HLE | 4.8 | 16.5 |
| IFBench | 38 | 68.7 |
| MathVista | 62.1 | 84.3 |
| MMMU-Pro | 44.6 | 69.2 |
| OmniScience Accuracy | 8.2 | 24.8 |
| OmniScience Non-Hallucination | 67.6 | 19.5 |
| scicode | 20.9 | 46.5 |
| Terminal-Bench Hard | 3 | 15.2 |
| τ²-Bench Telecom (AA run) | 22.2 | 55.6 |
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.