| Benchmark | Claude Sonnet 4 | Qwen3.7 Max |
|---|---|---|
| AA Agentic Index | 39.2 | 30.9 |
| AA Intelligence | 29.8 | 46.7 |
| AA-LCR | 69.7 | 74.7 |
| AA-Omniscience | 0.2 | 13.5 |
| Artificial Analysis Coding Index | 37.6 | 66 |
| critpt | 1.1 | 13.4 |
| gdpval | 31.2 | 38.5 |
| GPQA Diamond | 78 | 92.4 |
| HLE | 10.7 | 41.4 |
| HLE (with tools) | 20.3 | 53.5 |
| IFBench | 55 | 80.5 |
| ifeval | 87.6 | 94.3 |
| livebench | 74.8 | 74.3 |
| LiveCodeBench v6 | 48.5 | 91.6 |
| mmlu_redux | 93.6 | 95 |
| MMLU-Pro | 84 | 89.6 |
| mmmlu | 86.5 | 90.3 |
| OmniScience Accuracy | 22.7 | 31.1 |
| OmniScience Non-Hallucination | 70.9 | 74.4 |
| scicode | 40 | 53.5 |
| simplebench | 45.5 | 70.4 |
| supergpqa | 55.7 | 73.6 |
| SWE-bench Multilingual | 56.9 | 78.3 |
| SWE-bench Pro | 42.7 | 60.6 |
| SWE-bench Verified | 80.2 | 80.4 |
| TauBench V3 - Banking | 16.7 | 11.8 |
| Terminal-Bench 2.1 | 36.3 | 75 |
| Terminal-Bench Hard | 31.1 | 50.8 |
| τ²-Bench Telecom (AA run) | 64.6 | 94.7 |
| τ³-Bench | 13.8 | 10.9 |
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.