GLM-5.3 Chat Template Update Released

Algorithm Engineer zR at Zai_org posted on X that the team fixed an issue in the GLM chat template after the sgl_project community identified it. According to the post, the update optimizes tool-result reordering logic by using early exits instead of scanning every block. The post states that templates for GLM-5.3 and GLM-5.3-Flash were updated and the fix is now live. The post attached a static screenshot of the git diff for chat_template.jinja and thanked the community for the report.