Zhipu releases GLM-4.5 for agentic tasks under the MIT License

What happened

Zhipu (Z.ai) released GLM-4.5 in July 2025 together with the lighter GLM-4.5-Air. The weights were posted to Hugging Face on 20 July 2025.^[2][3] The two models were offered as open models, so developers could download them and run them on their own infrastructure.

Key details

GLM-4.5 has 355 billion total parameters with 32 billion active, while GLM-4.5-Air has 106 billion total and 12 billion active.^[1] Both models are hybrid reasoning models. The thinking mode handles complex reasoning and tool use, and the non-thinking mode gives instant answers.^[1]

In a comparison across 12 industry benchmarks, GLM-4.5 scored 63.2, ranking third among all proprietary and open models. GLM-4.5-Air scored 59.8.^[1] Wikipedia notes that GLM-4.5 can run on eight NVIDIA H20 chips.^[2] The H20 is the hardware reference Wikipedia gives for running the model, so it indicates the scale of computing needed to serve it.

Impact

The release is licensed under MIT, which permits commercial use. Wikipedia says the GLM family has been MIT-licensed since July 2025.^[1][2] For developers, this meant that a model with a large parameter count and strong benchmark results could be used, modified and redistributed under one of the most permissive open-source licences.

Reading the benchmark figures

A benchmark score is only meaningful when two models are run on the same test under the same conditions. Vendors usually publish comparison tables that set their model against selected rivals on selected tests, so a number should be read together with the name of the test and the models it is compared with. A small gap between two scores does not by itself show a real difference in ability, and one test rarely captures everything a model can do.

Agentic terms

An agent is a model that works through a task in steps, deciding what to do next and calling external tools such as search or code execution along the way. Thinking modes control whether and how a model reasons before it answers. A context window is the amount of text a model can use at one time, and a longer window lets a model keep more of a long session or document in view.