| Benchmark | Claude Sonnet 5 | Qwen3.7 Plus Preview |
|---|---|---|
| AA Agentic Index | 49.7 | 20.8 |
| AA Intelligence | 55.3 | 39.4 |
| AA-LCR | 77 | 69 |
| AA-Omniscience | 16.4 | 2.4 |
| arena_elo | 1463 | 1458 |
| arena_text_factuality | 1453 | 1454 |
| Artificial Analysis Coding Index | 71.5 | 55.9 |
| charxiv_rq | 77 | 85.9 |
| critpt | 16.9 | 9.1 |
| DeepSWE 1.1 | 54 | 14.2 |
| gdpval | 54.8 | 22.2 |
| GPQA Diamond | 91.1 | 90.3 |
| HLE | 57.4 | 35.6 |
| LVBench | 68.5 | 76.2 |
| MMMU-Pro | 77.3 | 80.5 |
| OmniScience Accuracy | 40 | 22.5 |
| OmniScience Non-Hallucination | 60.6 | 74.5 |
| OSWorld-Verified | 81.2 | 73.3 |
| scicode | 53.6 | 51.3 |
| SWE-bench Multilingual | 78.3 | 75.8 |
| SWE-bench Pro | 63.2 | 57.6 |
| SWE-bench Verified | 85.2 | 77.7 |
| TauBench V3 - Banking | 37.3 | 17.9 |
| Terminal-Bench 2.0 | 80.4 | 70.3 |
| Terminal-Bench 2.1 | 80.5 | 61 |
| τ³-Bench | 28.2 | 17.9 |
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.