On March 19, at Xiaomi's Spring New Product Launch Event, Lei Jun, Founder, Chairman and CEO of Xiaomi Group, unveiled MiMo-V2-Pro, the company's first self-developed flagship large language model, and announced plans to invest more than 60 billion yuan in the AI sector over the next three years.
According to the introduction, the MiMo-V2-Pro model features a trillion-parameter scale and million-level context processing capability, demonstrating exceptional agentic intelligence with outstanding performance in response speed, accuracy and overall capability. On Artificial Analysis, a globally authoritative comprehensive large model intelligence ranking, MiMo-V2-Pro ranks eighth globally. When ranked independently by large model brand, the model secures fifth place worldwide, surpassing the well-known xAI Grok model and becoming a flagship product among domestic large models that has entered the world's first echelon.
Xiaomi stated that the model has just completed its final development and will undergo rapid iterative optimization to continuously enhance its intelligence. At present, it is available for direct use by Xiaomi smartphone users, and the supporting on-device AI agent feature "Yang Longxia" has also been updated simultaneously, delivering a more comprehensive AI experience.
In addition, the MiMo-V2 model family welcomes two new members: V2-Omni, a multimodal large model capable of vision, audio perception as well as reasoning and execution; and V2-TTS, a speech large model that understands text content and generates emotional voice output.
As introduced by Lei Jun, the MiMo model series is developed by Xiaomi's Core Team, which has an average age of only 25. More than 60 percent of the team members are graduates of Tsinghua University and Peking University.
(Reprinted from https://news.eccn.com/）