| Benchmark | Gemini 3.6 Flash | o4-mini |
|---|---|---|
| AA Agentic Index | 40.5 | 36.1 |
| AA Intelligence | 52 | 26.1 |
| AA-LCR | 79 | 60 |
| AA-Omniscience | 22.1 | -35.7 |
| arc_agi_1 | 34.5 | 58.7 |
| ARC-AGI-2 | 2.6 | 6.1 |
| Artificial Analysis Coding Index | 69.2 | 25.6 |
| critpt | 10.6 | 0.6 |
| gdpval | 46.1 | 25.4 |
| GPQA Diamond | 92.8 | 81.4 |
| HLE | 40.8 | 16.5 |
| MMMU-Pro | 83.2 | 69.2 |
| OmniScience Accuracy | 50 | 24.8 |
| OmniScience Non-Hallucination | 44.4 | 19.5 |
| scicode | 52.7 | 46.5 |
| simpleqa_verified | 68.7 | 23.9 |
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.