GLM-4.5 is an open mixture-of-experts language model from Z.ai, released in August 2025 and built for agents, reasoning and coding in one model.
In plain words
GLM-4.5 has 355 billion parameters in total but only 32 billion are active for each token, because a mixture-of-experts model sends each token to just a few expert parts. A smaller sibling, GLM-4.5-Air, has 106 billion in total. Both can switch between a thinking mode for hard problems and tool use, and a quick-answer mode. Z.ai trained it on 23 trillion tokens, then improved it with reinforcement learning, and published base and reasoning versions openly.
A quick example
Imagine a coding agent asked to build a small web page. With GLM-4.5 it could think through the plan, call tools to create and test files, and switch to fast replies for simple follow-up questions.
Where you'll meet it
- Coding agents and front-end development.
- Tool use, web browsing and reasoning tasks.
Related
Sources
- PaperGLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models, GLM-4.5 Team (arXiv) · read 29 Sept 2026
- DocsGLM-4.5 overview, Z.ai Developer Docs · read 29 Sept 2026
- DocsGLM-4.5, zai-org (GitHub) · read 29 Sept 2026
Checked by WhatIsAI against the sources above · 29 Sept 2026. Spotted an error? Report it.