| Benchmark | GLM-5.1 | Qwen3.5 27B |
|---|---|---|
| AA Agentic Index | 66 | 54.6 |
| AA Intelligence | 41 | 34.6 |
| AA-LCR | 68 | 72.3 |
| AA-Omniscience | 0.8 | -44 |
| aime_2026 | 95.8 | 92.6 |
| Artificial Analysis Coding Index | 55.8 | 34.9 |
| browsecomp | 79.3 | 61 |
| coding_arena_elo | 1510 | 1357 |
| critpt | 4.6 | 0.9 |
| gdpval | 49.5 | 33 |
| GPQA Diamond | 86.8 | 85.8 |
| HLE | 52.3 | 48.5 |
| HMMT 2025 | 94 | 89.8 |
| hmmt_feb_2026 | 89.4 | 84.3 |
| hmmt_nov_2025 | 94 | 89.8 |
| IFBench | 76.3 | 76.5 |
| imo_answer_bench | 83.8 | 79.9 |
| itbenchSre | 40.3 | 35.5 |
| MCP Atlas | 75.6 | 68.4 |
| MMLU-Pro | 86 | 86.1 |
| nl2repo | 42.7 | 27.3 |
| OmniScience Accuracy | 25.2 | 20.7 |
| OmniScience Non-Hallucination | 70.1 | 24.9 |
| scicode | 43.8 | 39.5 |
| SWE-bench Multilingual | 73.3 | 69.3 |
| SWE-bench Pro | 58.4 | 51.2 |
| Terminal-Bench 2.0 | 69 | 41.6 |
| Terminal-Bench Hard | 43.2 | 32.6 |
| tool_decathlon | 40.7 | 31.5 |
| τ²-Bench Telecom (AA run) | 97.7 | 93.9 |
| τ³-Bench | 70.6 | 68.4 |
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.