| Benchmark | GLM-5.1 | Qwen2.5 Instruct 72B |
|---|---|---|
| AA Intelligence | 41 | 10 |
| AA-LCR | 68 | 20.3 |
| AA-Omniscience | 0.8 | -52.2 |
| Artificial Analysis Coding Index | 55.8 | 11.9 |
| critpt | 4.6 | 0 |
| GPQA Diamond | 86.8 | 49.1 |
| HLE | 52.3 | 4.2 |
| IFBench | 76.3 | 36.9 |
| livebench | 70.2 | 52.3 |
| MMLU-Pro | 86 | 71.6 |
| OmniScience Accuracy | 25.2 | 17.6 |
| OmniScience Non-Hallucination | 70.1 | 15.3 |
| scicode | 43.8 | 26.7 |
| Terminal-Bench Hard | 43.2 | 4.5 |
| τ²-Bench Telecom (AA run) | 97.7 | 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.