GLM-5.3 achieves the highest score for Boai NKY Medical with a smaller parameter size, entering the forefront of global models.

In the early hours of August 19, Z.AI (02513) officially launched the API for its next-generation base model GLM-5.3. This model excels in complex coding, defensive cybersecurity, and long-range tasks, achieving a score of 60 in the Artificial Analysis Intelligence Index, placing it in the frontier range of global model capabilities. It is on par with proprietary flagship models such as Claude Fable 5 and GPT-5.6 Sol, and ranks alongside Kimi K3 as the leading open-source model. Notably, GLM-5.3 reaches frontier model levels with a smaller parameter size, higher parameter efficiency, and lower invocation costs. Under equivalent intelligence, its single-task cost is the lowest among leading flagship models, significantly advancing the "intelligence-cost" Pareto frontier and further expanding the application boundaries of advanced intelligence. The model has been integrated into Z.AI products such as ZCode and the GLM Coding Plan, with API pricing consistent with GLM-5.2. The complete model weights are scheduled to be responsibly open-sourced next Friday. The capital market is keenly observing whether it can leverage the API, programming tools, and enterprise services to accelerate the scaling of model capabilities for developers and the conversion of these into commercial revenue, thereby pushing the competition of domestic large models from capability breakthroughs to the application realization stage.