| Benchmark | GPT-4o | Qwen3.7 Plus Preview |
|---|---|---|
| AA Agentic Index | 9.7 | 20.8 |
| AA Intelligence | 11.2 | 39.4 |
| AA-LCR | 55 | 69 |
| AA-Omniscience | -10.5 | 2.4 |
| arena_vision | 1162 | 1263 |
| Artificial Analysis Coding Index | 24.2 | 55.9 |
| critpt | 0 | 9.1 |
| ERQA | 35.2 | 69.8 |
| gdpval | 0 | 22.2 |
| GPQA Diamond | 70.1 | 90.3 |
| HLE | 5.3 | 35.6 |
| IFBench | 36 | 79.1 |
| ifeval | 84.3 | 94.6 |
| LiveCodeBench v6 | 30.9 | 89.6 |
| mathvision | 30.4 | 90.3 |
| mmlu_prox | 61.1 | 85.4 |
| mmlu_redux | 88 | 94.5 |
| MMLU-Pro | 74.7 | 88.5 |
| mmmlu | 81.4 | 89 |
| MMMU-Pro | 59.9 | 80.5 |
| OmniScience Accuracy | 23.7 | 22.5 |
| OmniScience Non-Hallucination | 62.1 | 74.5 |
| OSWorld-Verified | 5 | 73.3 |
| RealWorldQA | 75.4 | 86.9 |
| scicode | 33.4 | 51.3 |
| supergpqa | 42.4 | 71.4 |
| SWE-bench Verified | 38.8 | 77.7 |
| Terminal-Bench Hard | 8.3 | 47 |
| Video-MME | 77.2 | 88 |
| VideoMMMU | 61.2 | 85.4 |
| τ²-Bench Telecom (AA run) | 28.9 | 93 |
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.