Over the course of the last 24 hours, the TEE-backed AI infrastructure powering Phala Confidential AI successfully processed a total of 186.26 billion tokens. A breakdown of the leading models indicates that GLM-5.3 Flash drove the majority of this activity, accounting for 59.87% of the usage. Coming in next were Kimi K3 with a 13.74% share and DeepSeek V4.1 Flash representing 9.13% of the overall volume. You can access more details by visiting https://phala.com/models/z-ai/glm-5.3-flash.