GLM-5.3 (High): benchmark results
Provider: Zhipu. Released 2026-08-14. Access: Open.
Aggregate Rating 1832 ± 15, rank #99 of 2170 rated models, from 36 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| QuantArena - One-shot - Research Period | 52.54 | Score (%) | 100 |
| o11y-bench - Pass@3 | 98.44 | Tasks passed on at least one of three attempts, Pass@3 (%) | 87.9 |
| Multi-turn Debate (Lechmazur) | 1633.9 | Bradley-Terry Rating | 85 |
| DuelLab GameBench 2 - Reasoning Variants | 69.79 | GameBench2 normalized score (0–100) | 83.8 |
| QuantArena - One-shot - Overall | 38.02 | Score (%) | 83.8 |
| Context Arena | 70.4 | Average Score (%) | 81.9 |
| AI Coding Daily (OpenCode) - Total | 54.32 | Total points (max 70) | 81 |
| Finding the Right Fit - TUA-Bench (OpenHands) | 62.86 | Task score (%), OpenHands harness, high reasoning effort, on | 75 |
| AI Coding Daily (OpenCode) - Laravel Code Quality | 17.55 | Laravel Code Quality (max 20) points, LLM-judged rubric scor | 71.4 |
| LLM Chess (Saplin) | 605.19 | Native anchored-opponent Elo (signed points) | 69.9 |
| AI Coding Daily (OpenCode) - React-TS Code Quality | 17.67 | React-TS Code Quality (max 20) points, LLM-judged rubric sco | 69 |
| DeepsecBench | 21.82 | Recall-weighted F2 score (%) | 67.9 |
Interactive version: theaggregate.ai/model?slug=glm-5-3-high · How It Works · Data refreshed daily, snapshot 2026-11-05.