| Benchmark | Claude Opus 4.7 | Qwen2.5 Instruct 72B |
|---|---|---|
| AA Intelligence | 55 | 10 |
| AA-LCR | 75.3 | 20.3 |
| AA-Omniscience | 27.3 | -52.2 |
| Artificial Analysis Coding Index | 73.6 | 11.9 |
| critpt | 12 | 0 |
| GPQA Diamond | 94.2 | 49.1 |
| harmbench | 17.5 | 72.8 |
| HLE | 54.7 | 4.2 |
| IFBench | 58.6 | 36.9 |
| livebench | 76.9 | 52.3 |
| mmmlu | 91.5 | 74.8 |
| OmniScience Accuracy | 48.9 | 17.6 |
| OmniScience Non-Hallucination | 57.7 | 15.3 |
| scicode | 54.5 | 26.7 |
| SWE-bench Verified | 87.6 | 23.8 |
| Terminal-Bench Hard | 54.5 | 4.5 |
| τ²-Bench Telecom (AA run) | 88.6 | 34.5 |
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.