Odaily Planet Daily News: Zhipu (HK2513) officially released GLM-5.3. According to the announcement, compared with GLM-5.2, the base model of GLM-5.3 remains unchanged, but its intelligence ceiling has been greatly improved through extreme post-training Scaling. GLM-5.3 has stronger programming capabilities. In an in-house self-built embodied cognition evaluation, it is up 50% compared with GLM-5.2. It also achieved first place among open-source offerings on public benchmark tests, including Terminal Bench3.0 and Agents'Last Exam (CLI). Zhipu (HK2513) also said it will open up the model weights two weeks after the release. (First Financial)