| Benchmark | Gemini 3.6 Flash | GPT-4o |
|---|---|---|
| AA Agentic Index | 40.5 | 9.7 |
| AA Intelligence | 52 | 11.2 |
| AA-LCR | 79 | 55 |
| AA-Omniscience | 22.1 | -10.5 |
| arc_agi_1 | 34.5 | 4.5 |
| ARC-AGI-2 | 2.6 | 0 |
| Artificial Analysis Coding Index | 69.2 | 24.2 |
| critpt | 10.6 | 0 |
| gdpval | 46.1 | 0 |
| GPQA Diamond | 92.8 | 70.1 |
| HLE | 40.8 | 5.3 |
| MMMU-Pro | 83.2 | 59.9 |
| OmniScience Accuracy | 50 | 23.7 |
| OmniScience Non-Hallucination | 44.4 | 62.1 |
| OSWorld-Verified | 83 | 5 |
| scicode | 52.7 | 33.4 |
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.