| Benchmark | GLM-5.1 | Llama 3.1 Instruct 405B |
|---|---|---|
| AA Agentic Index | 66 | 6.3 |
| AA Intelligence | 41 | 8.5 |
| AA-LCR | 68 | 24.7 |
| AA-Omniscience | 0.8 | -17.1 |
| Artificial Analysis Coding Index | 55.8 | 14.5 |
| critpt | 4.6 | 0 |
| gdpval | 49.5 | 0 |
| GPQA Diamond | 86.8 | 51.5 |
| HLE | 52.3 | 4.2 |
| IFBench | 76.3 | 39 |
| MMLU-Pro | 86 | 73.4 |
| OmniScience Accuracy | 25.2 | 23.2 |
| OmniScience Non-Hallucination | 70.1 | 49 |
| scicode | 43.8 | 29.9 |
| simplebench | 55.1 | 23 |
| SWE-bench Pro | 58.4 | 11.2 |
| Terminal-Bench Hard | 43.2 | 6.8 |
| τ²-Bench Telecom (AA run) | 97.7 | 19 |
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.