Zhipu open-sources GLM-5.2 under MIT license with 1M context
Zhipu AI open-sourced GLM-5.2 on June 17 under MIT license, ranking #1 on Code Arena among available models, with 1M context and Day-0 support for 8 domestic compute platforms.
Zhipu AI officially launched and open-sourced the GLM-5.2 model on June 17. The model focuses on long-horizon task capabilities, achieving a 1M token lossless context window, targeting AI autonomously completing hours-long complex engineering projects.
In the Code Arena million-user blind test, GLM-5.2 ranks #1 among all globally available models. On FrontierSWE and Terminal-Bench 2.1 benchmarks, GLM-5.2 narrows the gap with Claude Opus 4.8 to 1-4%, the highest-ranked open-source model. Terminal-Bench 2.1 score is 81.0, up 17.5 points from GLM-5.1's 63.5.
GLM-5.2 is open-sourced under the MIT license, the most permissive license ever used for a Chinese flagship LLM. Model weights are available on Hugging Face and ModelScope.
On launch day, GLM-5.2 completed full adaptation with 8 domestic compute platforms including Huawei Ascend, Pingtouge, Moore Threads, Cambricon, Kunlunxin, MetaX, Hygon, and Biren.