Article · AI translation
IT Home reported on November 10 that Xiaomi Technology announced today on its official Weibo account that the “Thing Model Technical White Paper for IoT Devices,” prepared by Xiaomi Technology and Peking University under the leadership of the China Computer Federation (CCF) Ubiquitous Operating Systems Open Community, was officially released at the 2026 CCF China Software Conference.
According to the introduction, the thing model is a “universal digital language” for smart devices. Drawing on Xiaomi’s large-scale experience with more than 1.16 billion AIoT devices (spanning 200+ categories and 10,000+ partners), the white paper presents three core breakthroughs in dynamic architecture, AI integration, and trusted security:
First, it introduces an innovative four-layer thing model architecture and a Dynamic Service mechanism, overcoming the limitations of traditional static descriptions. It supports automatic discovery of new device functions and adaptive adjustments to scenarios, making control across categories and the expansion of complex devices more flexible and consistent;
Second, it takes the lead in advancing the thing model toward a “capability contract” for AI agents, establishing “four standards” and “three evaluations” for an “AI-friendly thing model.” It supports integration with interfaces such as MCP and Function Calling, allowing AI to precisely control devices by understanding natural language and autonomously coordinate multiple devices with a single spoken instruction;
Third, drawing on Xiaomi’s MiTEE trusted security system and combining device identity authentication, permission controls, and risk-tiered confirmation of AI actions, it builds a five-layer security barrier to make AI control of the physical world smarter, safer, and more trustworthy.
IT Home noted that Lu Weibing, partner and president of Xiaomi Group, president of its smartphone division, and general manager of the Xiaomi brand, reposted the announcement and said that Xiaomi is officially opening up to the entire industry the technology and experience it has validated across 1.16 billion devices.