Zhipu AI, now operating as Z.ai, a Tsinghua University-affiliated Chinese AI lab, develops the GLM model family.

GLM-4 is a series of open multilingual, multimodal chat language models, including GLM-4-9B and GLM-4.5.

Code is hosted on GitHub under the zai-org organization, with weights on Hugging Face under THUDM and zai-org.

GLM-4 released in June 2024, with GLM-4.5 following in 2025.

Built to provide an open, commercially usable alternative to ChatGLM for multilingual chat and agentic use cases.

Runs via Hugging Face Transformers, vLLM, or SGLang.

Code under Apache 2.0 with a separate model weights license permitting commercial use, no credentials required to download, self-hosted or via Zhipu's API; maintained by Zhipu AI / Z.ai.