Z.aiGLM-4
GLM-4.5
A hybrid reasoning model with sparse experts.
Practical use
Practical guidance. Keep a mode choice tied to the task. A direct answer may suit a simple extraction, while an extended tool sequence needs review of intermediate results. Confirm each endpoint's settings before transferring a local inference configuration.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | API, Downloadable |
| Model identifiers | glm-4.5glm-4.5-xglm-4.5-flashzai-org/GLM-4.5zai-org/GLM-4.5-FP8 |
| Size | 355 billion total parameters; 32 billion active.GLM-4.5 model card |
| Modes | Thinking and non-thinking; full 128,000-token context is documented.GLM-4.5 model card |
| Licence | MIT.GLM-4.5 model card |
| API variants | GLM-4.5, faster GLM-4.5-X and GLM-4.5-Flash are listed.Current Z.ai API catalogue |
Related models
A reasoning and coding release with preserved thinking across turns.
Launch date not confirmedAPI · DownloadableA GLM text release with a 200,000-token context.
Launch date not confirmedAPI · Downloadable- GLM-4-32B-0414
Z.ai · GLM-4
A 32-billion-parameter GLM release from April 2025.
Launch date API · Downloadable