What is GLM (Z.ai's LLM family)?

GLM is the large language model family from Z.ai — the Chinese lab formerly known as Zhipu AI — and its latest open-weight release competes head-on with top US frontier models. That makes it one of the most closely watched names in AI right now.

As of July 2026, the headliner is GLM-5.2, released June 13, 2026. It is a mixture-of-experts model with 744 billion total parameters (40 billion active), a one-million-token context window, and MIT-licensed weights — anyone can download, modify, and deploy it with no usage restrictions.

The benchmarks explain the buzz. GLM-5.2 scored 62.1 on SWE-bench Pro, ahead of GPT-5.5's 58.6, and it also narrowed the cybersecurity capability gap with Anthropic's restricted Mythos models. Hosted access is cheap too — roughly $1.40 per million input tokens and $4.40 per million output through routers like OpenRouter, a fraction of closed-model pricing.

The timing was pointed: the open release landed the same week US export controls restricted Anthropic's Fable 5 and Mythos 5 for foreign nationals, and Z.ai's stock jumped on the news. One caveat — the MIT weights can run anywhere, but Z.ai's own API routes data through China-hosted servers, which some companies will not accept. For how GLM compares with other open-source LLM options, check the model tracker.