| Benchmark | GLM-5.1 | Opus 5 |
|---|---|---|
| AA Agentic Index | 66 | 59.2 |
| AA Intelligence | 41 | 63.1 |
| AA-LCR | 68 | 78.7 |
| AA-Omniscience | 0.8 | 37.1 |
| arena_elo | 1468 | 1487 |
| arena_text_factuality | 1458 | 1481 |
| Artificial Analysis Coding Index | 55.8 | 78 |
| browsecomp | 79.3 | 90.8 |
| coding_arena_elo | 1510 | 1662 |
| critpt | 4.6 | 29.1 |
| Finance Agent v2 | 44.8 | 58.6 |
| gdpval | 49.5 | 67.2 |
| GDPval-AA (Elo) | 1535 | 1827 |
| GPQA Diamond | 86.8 | 93.7 |
| HLE | 52.3 | 64.7 |
| HLE (with tools) | 52.3 | 64.7 |
| MCP Atlas | 75.6 | 85.8 |
| OmniScience Accuracy | 25.2 | 60.9 |
| OmniScience Non-Hallucination | 70.1 | 40.5 |
| scicode | 43.8 | 55.7 |
| simplebench | 55.1 | 80.6 |
| simpleqa_verified | 38.1 | 56.7 |
| SWE-bench Multilingual | 73.3 | 89.5 |
| SWE-bench Pro | 58.4 | 79.2 |
| TauBench V3 - Banking | 13.6 | 44.7 |
| Terminal-Bench 2.1 | 63.5 | 89.1 |
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.