| Benchmark | o1 | Qwen3.5 35B A3B |
|---|---|---|
| AA Agentic Index | 31.1 | 44.1 |
| AA Intelligence | 24 | 29.9 |
| AA-LCR | 63.3 | 68.3 |
| AA-Omniscience | -10.6 | -48.1 |
| Artificial Analysis Coding Index | 39.7 | 37 |
| critpt | 0.3 | 0.9 |
| gdpval | 11.5 | 20.3 |
| GPQA Diamond | 75.7 | 84.5 |
| GSM8K | 97.1 | 90.1 |
| HLE | 7.7 | 47.4 |
| humaneval | 88.1 | 66.5 |
| IFBench | 70.3 | 72.5 |
| MathVista | 71.8 | 86.2 |
| mmlu | 91.8 | 81.1 |
| mmmlu | 87.7 | 85.2 |
| MMMU | 77.6 | 81.4 |
| OmniScience Accuracy | 34.7 | 20.1 |
| OmniScience Non-Hallucination | 30.7 | 14.6 |
| scicode | 35.8 | 37.7 |
| SWE-bench Verified | 48.9 | 70 |
| Terminal-Bench Hard | 12.9 | 26.5 |
| τ²-Bench Telecom (AA run) | 62.6 | 89.2 |
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.