| Benchmark | GPT-4o mini | Qwen3.5 27B |
|---|---|---|
| AA Agentic Index | 1 | 54.6 |
| AA Intelligence | 7 | 34.6 |
| AI2D | 75.2 | 92.9 |
| Artificial Analysis Coding Index | 11.4 | 34.9 |
| gdpval | 0 | 33 |
| GPQA Diamond | 42.6 | 85.8 |
| HLE | 4.2 | 48.5 |
| IFBench | 31 | 76.5 |
| MathVista | 56.7 | 87.8 |
| MMMU | 59.4 | 82.3 |
| MMMU-Pro | 41.5 | 75 |
| scicode | 22.9 | 39.5 |
| SWE-bench Verified | 8.7 | 75 |
| τ³-Bench | 2.9 | 68.4 |
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.