Zhiyu released GLM-5.3

On August 14, Zhiyu officially released GLM-5.3. According to the introduction, the base model of GLM-5.3 remains unchanged compared to GLM-5.2, but its intelligent upper limit has been significantly enhanced through extreme post-training scaling. GLM-5.3 boasts stronger programming capabilities, achieving a 50% improvement over GLM-5.2 in internal self-built evaluation metrics, and it secured the top position in open benchmark tests, including Terminal Bench 3.0 and Agents' Last Exam. Zhiyu also stated that the model weights will be made available two weeks after the release.