GLM-5.3
Z.ai's flagship on GLM-5.2's base, with every gain from scaled-up post-training — frontier-level coding and cyber-defense capability
GLM-5.3
Z.ai • August 2026
Training Data
Not disclosed
GLM-5.3
August 2026
Parameters
753 billion
Training Method
Same base as GLM-5.2 with scaled-up post-training; reasoning always on (low, high, max effort)
Context Window
1,000,000 tokens
Knowledge Cutoff
Not disclosed
Key Features
1M Context / 128K Output • Always-On Reasoning • Open Weights (custom licence) • Text-Only
Capabilities
Coding: Outstanding (66.9 DeepSWE v1.1)
Cybersecurity: Outstanding (84.5% CyberGym)
Agentic: Excellent
What's New in This Version
About 50% better at coding than GLM-5.2 (DeepSWE v1.1: 66.9 vs 46.2) and far stronger on cyber tasks (ExploitBench 54.4% vs 24.4%); API launched August 14, with weights held for a safety review and released August 28 under a licence requiring security review for Model-as-a-Service operators above $10B revenue; $1.40/$4.40 per 1M tokens
Z.ai's flagship on GLM-5.2's base, with every gain from scaled-up post-training — frontier-level coding and cyber-defense capability
What's New in This Version
About 50% better at coding than GLM-5.2 (DeepSWE v1.1: 66.9 vs 46.2) and far stronger on cyber tasks (ExploitBench 54.4% vs 24.4%); API launched August 14, with weights held for a safety review and released August 28 under a licence requiring security review for Model-as-a-Service operators above $10B revenue; $1.40/$4.40 per 1M tokens
Technical Specifications
Key Features
Capabilities
Other Z.ai Models
Explore more models from Z.ai
GLM-5.3-Flash
Z.ai's MIT-licensed efficient model — natively multimodal, with hybrid sparse and linear attention far cheaper to run than GLM-5.3
GLM-5.2
Zhipu AI's MIT-licensed flagship extending GLM-5.1 to a 1M-token context, the highest-ranked open-source model on long-horizon coding benchmarks
GLM-5.1
Zhipu AI's current flagship refined from GLM-5 for agentic engineering, topping the SWE-Bench Pro leaderboard with sustained 8-hour autonomous execution
