| Benchmark | Claude 3.7 Sonnet | Qwen3.7 Max |
|---|---|---|
| AA Agentic Index | 37 | 30.9 |
| AA Intelligence | 27.6 | 46.7 |
| AA-LCR | 62.3 | 74.7 |
| AA-Omniscience | -0.7 | 13.5 |
| Artificial Analysis Coding Index | 36.4 | 66 |
| critpt | 0.9 | 13.4 |
| gdpval | 27.4 | 38.5 |
| GPQA Diamond | 84.8 | 92.4 |
| HLE | 9.7 | 41.4 |
| IFBench | 48.3 | 80.5 |
| ifeval | 93.2 | 94.3 |
| mmmlu | 86.1 | 90.3 |
| OmniScience Accuracy | 28.2 | 31.1 |
| OmniScience Non-Hallucination | 60 | 74.4 |
| scicode | 40.3 | 53.5 |
| simplebench | 46.4 | 70.4 |
| SWE-bench Verified | 70.3 | 80.4 |
| Terminal-Bench Hard | 21.2 | 50.8 |
| τ²-Bench Telecom (AA run) | 54.7 | 94.7 |
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.