Zhipu AI open-sourced GLM-5.2 on June 17, focusing on coding and long-term task execution. It ranks second globally on Code Arena and first among open-source models. The model features 1M lossless context, processes up to 880k tokens, and achieves performance between Claude Opus 4.7 and 4.8 on FrontierSWE. It uses IndexShare architecture, reducing FLOPs to 2.9x under 1M context, and is released under MIT license on Hugging Face and ModelScope, with inference adapted to Huawei Ascend and other domestic platforms.