GLM-5.3 Flash (High): benchmark results

Provider: Zhipu. Released 2026-08-20. Access: Open.

Unified ELO 1696 ± 1, rank #149 of 3078 rated models, from 16 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
DuelLab Overall48.7DuelLab Score81.5
Korean CSAT 2026 (Easy Mode) - Physics I42Points (out of 50)78.6
Context Arena63.43Average Score (%)68.9
Korean CSAT 2026 (Easy Mode) - Society and Culture42Points (out of 50)68
Korean CSAT 2026 (Easy Mode) - Mathematics96Points (out of 100)57.8
Korean CSAT 2026 (Easy Mode) - Total409Points (out of 450)57.1
Korean CSAT 2026 (Easy Mode) - Korean93Points (out of 100)55.8
Bug Hunt Bench - VS Code Extension9Planted Bugs Fixed (out of 45)52.2
Korean CSAT 2026 (Easy Mode) - English95Points (out of 100)46.6
DeepsecBench12.98Recall-weighted F2 score (%)45.9
Korean CSAT 2026 (Easy Mode) - Chemistry I41Points (out of 50)44.6
Maze-Bench22.18Mean Score43.8

Interactive version: theaggregate.ai/model?slug=glm-5-3-flash-high · How It Works · Data refreshed daily, snapshot 2026-09-19.