| Benchmark | GPT-4o mini | Qwen3.6 35B A3B |
|---|---|---|
| AA Agentic Index | 1 | 21.6 |
| AA Intelligence | 7 | 32.1 |
| AI2D | 75.2 | 92.7 |
| Artificial Analysis Coding Index | 11.4 | 41.9 |
| gdpval | 0 | 27.8 |
| GPQA Diamond | 42.6 | 86 |
| HLE | 4.2 | 22.2 |
| IFBench | 31 | 64.4 |
| MathVista | 56.7 | 86.4 |
| MMMU | 59.4 | 81.7 |
| MMMU-Pro | 41.5 | 75.3 |
| scicode | 22.9 | 35.8 |
| SWE-bench Verified | 8.7 | 73.4 |
| TauBench V3 - Banking | 2.9 | 9.3 |
| Terminal-Bench 2.1 | 5.6 | 44.9 |
| τ³-Bench | 2.9 | 67.2 |
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.