During the past 24 hours, the TEE-backed AI infrastructure at Phala Confidential AI successfully processed a total of 173.64 billion tokens. A closer look at the highest performing models shows that GLM-5.3 led the usage with a 51.89% share. This was followed by GLM-5.3 Flash, which accounted for 25.07% of the activity, and deepseek-v4.1-flash, making up 10.61% of the volume. For more information, please visit https://phala.com/models/z-ai/glm-5.3