GLM-5.2 (Max): benchmark results
GLM-5.2 evaluated at the max reasoning-effort setting. Provider: Zhipu. Released 2026-06-13. Access: Open.
Aggregate Rating 1763 ± 7, rank #209 of 2168 rated models, from 341 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OmniaBench - DAG-S | 69 | Pass@1 (%; 200 single-turn tasks refined from the DAG route; | 100 |
| SQBench - L3 Business Scenarios | 28.3 | Strict Pass@1 (%; the 60 L3 domain business-scenario tasks o | 100 |
| AA TAU-2 Bench | 99.12 | Accuracy (%) | 99.9 |
| Nejumi 4 - jaster (2-shot) - MAWPS | 99 | Exact match (%) | 96.9 |
| OmniaBench - ToE | 57.55 | Pass@1 (%; tasks from the employee-oriented (ToE) domain spl | 95.2 |
| AA Terminal-Bench Hard | 50.76 | Accuracy (%) | 95 |
| AA-Omniscience Hallucination Rate | 26.3 | Hallucination Rate (%) | 94.8 |
| Chatbot Arena (Text - Life, Physical & Social Science) | 1499 | Arena Score | 94.8 |
| Chatbot Arena (Text - Spanish) | 1478 | Arena Score | 94.3 |
| AA Omniscience - Software Engineering (SWE) - TypeScript | 48.89 | Accuracy (%) | 93.3 |
| Chatbot Arena (Text - Legal & Government) | 1490 | Arena Score | 93.3 |
| Nejumi 4 - jaster (0-shot) - MGSM | 93 | Exact match (%) | 93 |
Interactive version: theaggregate.ai/model?slug=glm-5-2-max · How It Works · Data refreshed daily, snapshot 2026-10-20.