GLM-5.2 Announced: 1M Context, 131K Output, API & Weights Next Week
GLM-5.2
The Chinese coding flagship has been updated; for now it’s only an announcement, the actual release is next week, and it’s currently available in the coding plan.
The model will be available on all plans, with a 1M-token context window, a maximum output of 131,072 tokens, high and max reasoning, as well as improved agent sessions for coding, etc.
Next week they will roll out the API and publish the weights along with benchmarks.
If anyone tests it, share your results in the comments.