| Benchmark | Gemini 3.6 Flash | Qwen3.6 35B A3B |
|---|---|---|
| AA Agentic Index | 40.5 | 21.6 |
| AA Intelligence | 52 | 32.1 |
| AA-LCR | 79 | 66.7 |
| AA-Omniscience | 22.1 | -22.2 |
| aime_2026 | 96.7 | 92.7 |
| Artificial Analysis Coding Index | 69.2 | 41.9 |
| charxiv_rq | 85.2 | 78 |
| critpt | 10.6 | 0.3 |
| gdpval | 46.1 | 27.8 |
| GDPval-AA v2 | 1421 | 1015 |
| GPQA Diamond | 92.8 | 86 |
| HLE | 40.8 | 22.2 |
| hmmt_feb_2026 | 89.4 | 83.6 |
| LVBench | 84.2 | 71.4 |
| MMMU-Pro | 83.2 | 75.3 |
| OmniScience Accuracy | 50 | 18.8 |
| OmniScience Non-Hallucination | 44.4 | 49.5 |
| scicode | 52.7 | 35.8 |
| SWE-bench Pro | 58.7 | 49.5 |
| TauBench V3 - Banking | 29.9 | 9.3 |
| Terminal-Bench 2.1 | 78 | 44.9 |
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.