| Benchmark | Claude Sonnet 5 | Qwen2.5 Instruct 72B |
|---|---|---|
| AA Intelligence | 55.3 | 10 |
| AA-LCR | 77 | 20.3 |
| AA-Omniscience | 16.4 | -52.2 |
| Artificial Analysis Coding Index | 71.5 | 11.9 |
| critpt | 16.9 | 0 |
| GPQA Diamond | 91.1 | 49.1 |
| HLE | 57.4 | 4.2 |
| OmniScience Accuracy | 40 | 17.6 |
| OmniScience Non-Hallucination | 60.6 | 15.3 |
| scicode | 53.6 | 26.7 |
| SWE-bench Verified | 85.2 | 23.8 |
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.