Over the last 24 hours, our TEE-backed AI infrastructure at Phala Confidential AI successfully processed a total of 85.07 billion tokens. We are pleased to share the breakdown of our most utilized models during this timeframe. DeepSeek V4.1 Flash led the pack, accounting for 51.54% of all activity. GLM-5.3 Flash came next with a 25.82% share, while Kimi K3 made up 5.22% of the usage. You can find more information about our leading model by navigating to https://phala.com/models/deepseek/deepseek-v4.1-flash