Zhipu AI (THUDM) has officially released GLM 5.2, the latest iteration of its flagship open-weights model family. Announced today by Jie Tang on Twitter, the release is already making waves on Hacker News — racking up 269 points and 146 comments within hours. For developers who have been watching the open-weight LLM race, this is a significant moment.
GLM 5.2 builds on the GLM-4 series that put Zhipu on the global map. The release focuses on three areas that matter most to production teams:
Weights, inference code, and a technical report have landed on Hugging Face under the THUDM organization, with an OpenAI-compatible API endpoint exposed by Zhipu's own platform.
The open-weights race has consolidated around a handful of serious contenders — Llama, Qwen, DeepSeek, Mistral, and now GLM. Zhipu's positioning is unique: a Chinese lab that consistently weights-and-releases frontier-class models while still maintaining a hosted commercial API. For developers, that translates to real options:
GLM 5.2 lands on the same week that U.S. regulators have reportedly cracked down on Anthropic models following Amazon CEO conversations, and state attorneys general opened an investigation into OpenAI. The open-weight ecosystem is becoming not just a technical alternative, but a strategic one. When frontier capability is available under a permissive license with a self-host path, the calculus for enterprise procurement shifts.
For indie developers and startups especially, GLM 5.2 is a reminder: you don't have to be locked into a single vendor to get frontier-class quality.
If you want to try it today:
huggingface.co/THUDM
and load with transformers
or vLLM
.https://api.zhipuai.cn
(OpenAI-compatible).GLM 5.2 is the latest signal that the open-weight frontier is alive and shipping fast. If you've been waiting for a reason to diversify away from a single API provider, today is a good day to start.
What workloads are you planning to run on GLM 5.2 — code generation, long-doc retrieval, agentic pipelines? Drop a comment with your stack and I'll share benchmark setups that have worked for me.