| Benchmark | Gemini 2.5 Flash | Qwen3.6 27B |
|---|---|---|
| AA Agentic Index | 18.8 | 27.5 |
| AA Intelligence | 17.7 | 37.7 |
| AA-LCR | 65.7 | 73.3 |
| AA-Omniscience | -29.8 | -20 |
| Artificial Analysis Coding Index | 22.2 | 53.7 |
| critpt | 1.4 | 1.1 |
| gdpval | 11.9 | 32 |
| GPQA Diamond | 82.8 | 87.8 |
| HLE | 12.1 | 24 |
| hmmt_feb_2025 | 64.2 | 93.8 |
| IFBench | 50.3 | 69.1 |
| MMMU | 79.7 | 82.9 |
| MMMU-Pro | 69.1 | 75.8 |
| OmniScience Accuracy | 26.1 | 19.6 |
| OmniScience Non-Hallucination | 24.7 | 50.7 |
| scicode | 35.9 | 39.8 |
| simpleqa | 26.9 | 95.7 |
| SWE-bench Verified | 60.4 | 77.2 |
| Terminal-Bench Hard | 13.6 | 34.8 |
| τ²-Bench Telecom (AA run) | 31.6 | 94.2 |
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.