| Benchmark | GLM-5.2 Full Open Source | Qwen3.6 Plus |
|---|---|---|
| AA Agentic Index | 45.7 | 29 |
| AA Intelligence | 52.6 | 40.5 |
| AA-LCR | 76.7 | 72.3 |
| AA-Omniscience | 4.4 | 2.6 |
| aime_2026 | 99.2 | 95.3 |
| arena_elo | 1471 | 1444 |
| arena_text_factuality | 1460 | 1438 |
| Artificial Analysis Coding Index | 68.8 | 54.5 |
| coding_arena_elo | 1582 | 1460 |
| critpt | 20.9 | 2.9 |
| gdpval | 50.3 | 32 |
| GPQA Diamond | 91.2 | 90.4 |
| HLE | 54.7 | 28.8 |
| HLE (with tools) | 54.7 | 50.6 |
| HMMT 2025 | 94.4 | 94.6 |
| hmmt_feb_2026 | 92.5 | 87.8 |
| hmmt_nov_2025 | 94.4 | 94.6 |
| IFBench | 73.3 | 75.2 |
| imo_answer_bench | 91 | 83.8 |
| livebench_agentic_coding | 51.8 | 41.4 |
| livebench_coding | 79.7 | 78.2 |
| livebench_data_analysis | 73.7 | 69.9 |
| livebench_instruction_following | 62.3 | 58.3 |
| livebench_language | 76.2 | 75 |
| livebench_math | 89.8 | 83.7 |
| livebench_reasoning | 78.6 | 75.8 |
| MCP Atlas | 82.6 | 74.1 |
| nl2repo | 48.9 | 37.9 |
| OmniScience Accuracy | 24.3 | 26.4 |
| OmniScience Non-Hallucination | 73.7 | 68 |
| scicode | 50.5 | 40.7 |
| simpleqa_verified | 38.1 | 49.1 |
| SWE-bench Pro | 62.1 | 56.6 |
| TauBench V3 - Banking | 34.6 | 20.8 |
| Terminal-Bench 2.1 | 82.7 | 61.4 |
| Terminal-Bench Hard | 50.8 | 43.9 |
| tool_decathlon | 48.2 | 39.8 |
| toolathlon | 48.2 | 39.8 |
| τ²-Bench Telecom (AA run) | 99.1 | 97.7 |
| τ³-Bench | 26.8 | 70.7 |
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.