| Benchmark | Gemini 3 Flash Preview | Gemma 4 E2B |
|---|---|---|
| AA Agentic Index | 49.7 | 7.4 |
| AA Intelligence | 38.7 | 10 |
| AA-LCR | 73 | 17 |
| AA-Omniscience | 10.1 | -23.6 |
| aime_2026 | 96.7 | 37.5 |
| Artificial Analysis Coding Index | 42.6 | 8.3 |
| critpt | 8.6 | 0 |
| gdpval | 35.2 | 0 |
| GPQA Diamond | 90.4 | 43.4 |
| HLE | 43.5 | 4.8 |
| IFBench | 78 | 38 |
| mmmlu | 91.8 | 67.4 |
| MMMU-Pro | 81.2 | 44.6 |
| mrcr_v2_8needle_128k_average | 67.2 | 19.1 |
| OmniScience Accuracy | 53.4 | 8.2 |
| OmniScience Non-Hallucination | 7.6 | 67.6 |
| scicode | 50.6 | 20.9 |
| TauBench V3 - Banking | 20.8 | 4.5 |
| Terminal-Bench 2.1 | 58 | 0.4 |
| Terminal-Bench Hard | 38.6 | 3 |
| τ²-Bench Telecom (AA run) | 80.4 | 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.