| Benchmark | Opus 5 | Qwen2.5 Instruct 72B |
|---|---|---|
| AA Intelligence | 63.1 | 10 |
| AA-LCR | 78.7 | 20.3 |
| AA-Omniscience | 37.1 | -52.2 |
| Artificial Analysis Coding Index | 78 | 11.9 |
| critpt | 29.1 | 0 |
| GPQA Diamond | 93.7 | 49.1 |
| HLE | 64.7 | 4.2 |
| OmniScience Accuracy | 60.9 | 17.6 |
| OmniScience Non-Hallucination | 40.5 | 15.3 |
| scicode | 55.7 | 26.7 |
| SWE-bench Verified | 96 | 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.