| Benchmark | GPT-4o mini | Qwen3.6 27B |
|---|---|---|
| AA Agentic Index | 1 | 27.5 |
| AA Intelligence | 7 | 37.7 |
| Artificial Analysis Coding Index | 11.4 | 53.7 |
| gdpval | 0 | 32 |
| GPQA Diamond | 42.6 | 87.8 |
| HLE | 4.2 | 24 |
| humaneval | 87.2 | 78.2 |
| IFBench | 31 | 69.1 |
| MathVista | 56.7 | 87.4 |
| MMMU | 59.4 | 82.9 |
| MMMU-Pro | 41.5 | 75.8 |
| scicode | 22.9 | 39.8 |
| simpleqa | 9.9 | 95.7 |
| SWE-bench Verified | 8.7 | 77.2 |
| TauBench V3 - Banking | 2.9 | 16.7 |
| Terminal-Bench 2.1 | 5.6 | 60.7 |
| τ³-Bench | 2.9 | 15.3 |
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.