| Benchmark | Qwen3.6 Plus | Qwen3 VL 4B (Reasoning) |
|---|---|---|
| AA Agentic Index | 29 | 14.4 |
| AA Intelligence | 40.5 | 7.7 |
| AA-LCR | 72.3 | 23 |
| AA-Omniscience | 2.6 | -68.4 |
| AI2D | 94.4 | 84.9 |
| Artificial Analysis Coding Index | 54.5 | 6.7 |
| critpt | 2.9 | 0 |
| ERQA | 65.7 | 47.3 |
| gdpval | 32 | 13.8 |
| GPQA Diamond | 90.4 | 64.1 |
| HLE | 28.8 | 4.6 |
| HMMT 2025 | 94.6 | 53.1 |
| IFBench | 75.2 | 36.6 |
| ifeval | 94.3 | 82.6 |
| include | 85.1 | 64.6 |
| LiveCodeBench v6 | 87.1 | 51.3 |
| mathvision | 88 | 60 |
| mmlu_prox | 84.7 | 65 |
| mmlu_redux | 94.5 | 86 |
| MMLU-Pro | 88.5 | 73.6 |
| MMMU-Pro | 78.8 | 57 |
| MMStar | 83.3 | 73.2 |
| OmniScience Accuracy | 26.4 | 12.2 |
| OmniScience Non-Hallucination | 68 | 8.2 |
| OSWorld-Verified | 62.5 | 31.4 |
| RealWorldQA | 85.4 | 73.2 |
| scicode | 40.7 | 17.1 |
| screenspot_pro_no_tools | 68.2 | 49.2 |
| supergpqa | 71.6 | 46.8 |
| Terminal-Bench Hard | 43.9 | 1.5 |
| Video-MME | 84.2 | 59.7 |
| VideoMMMU | 84 | 69.4 |
| τ²-Bench Telecom (AA run) | 97.7 | 15.5 |
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.