| Benchmark | GLM-5.1 | GPT-4o |
|---|---|---|
| AA Agentic Index | 66 | 9.7 |
| AA Intelligence | 41 | 11.2 |
| AA-LCR | 68 | 55 |
| AA-Omniscience | 0.8 | -10.5 |
| Artificial Analysis Coding Index | 55.8 | 24.2 |
| critpt | 4.6 | 0 |
| gdpval | 49.5 | 0 |
| GPQA Diamond | 86.8 | 70.1 |
| HLE | 52.3 | 5.3 |
| IFBench | 76.3 | 36 |
| MMLU-Pro | 86 | 74.7 |
| OmniScience Accuracy | 25.2 | 23.7 |
| OmniScience Non-Hallucination | 70.1 | 62.1 |
| scicode | 43.8 | 33.4 |
| simplebench | 55.1 | 17.8 |
| Terminal-Bench Hard | 43.2 | 8.3 |
| τ²-Bench Telecom (AA run) | 97.7 | 28.9 |
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.