| Benchmark | GLM-5.1 | o3 |
|---|---|---|
| AA Agentic Index | 66 | 36.1 |
| AA Intelligence | 41 | 31.1 |
| AA-LCR | 68 | 73.3 |
| AA-Omniscience | 0.8 | -15.3 |
| Artificial Analysis Coding Index | 55.8 | 38.4 |
| browsecomp | 79.3 | 49.7 |
| critpt | 4.6 | 1.1 |
| frontiermath_tier_4 | 12.5 | 2.1 |
| gdpval | 49.5 | 12.8 |
| GPQA Diamond | 86.8 | 83.3 |
| HLE | 52.3 | 20.3 |
| IFBench | 76.3 | 71.4 |
| MMLU-Pro | 86 | 85 |
| OmniScience Accuracy | 25.2 | 38.6 |
| OmniScience Non-Hallucination | 70.1 | 12.9 |
| scicode | 43.8 | 41 |
| simplebench | 55.1 | 53.1 |
| simpleqa_verified | 38.1 | 53 |
| Terminal-Bench Hard | 43.2 | 37.1 |
| τ²-Bench Telecom (AA run) | 97.7 | 80.7 |
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.