| Benchmark | o4-mini | Qwen3.6 27B |
|---|---|---|
| AA Agentic Index | 36.1 | 27.5 |
| AA Intelligence | 26.1 | 37.7 |
| AA-LCR | 60 | 73.3 |
| AA-Omniscience | -35.7 | -20 |
| Artificial Analysis Coding Index | 25.6 | 53.7 |
| critpt | 0.6 | 1.1 |
| gdpval | 25.4 | 32 |
| GPQA Diamond | 81.4 | 87.8 |
| HLE | 16.5 | 24 |
| IFBench | 68.7 | 69.1 |
| MathVista | 84.3 | 87.4 |
| MMMU | 81.6 | 82.9 |
| MMMU-Pro | 69.2 | 75.8 |
| OmniScience Accuracy | 24.8 | 19.6 |
| OmniScience Non-Hallucination | 19.5 | 50.7 |
| scicode | 46.5 | 39.8 |
| SWE-bench Verified | 68.1 | 77.2 |
| Terminal-Bench Hard | 15.2 | 34.8 |
| τ²-Bench Telecom (AA run) | 55.6 | 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.