Meituan refutes rumors that the person in charge of the LongCat model has resigned, and continues to promote the implementation of the domestic large-scale model with trillions of computing power. On July 27, in response to market rumors that Pei Peng, the person in charge of the basic model of the Meituan LongCat team, is about to leave, Meituan officials publicly stated that the news was untrue.

Meituan refutes rumors that the person in charge of the LongCat model has resigned, and continues to promote the implementation of the domestic large-scale model with trillions of computing power. On July 27, in response to market rumors that Pei Peng, the person in charge of the basic model of the Meituan LongCat team, is about to leave, Meituan officials publicly stated that the news was untrue. Public information shows that Pei Peng graduated from Peking University. After joining Meituan in 2023, he served as the pre-training leader of the base model team and the basic model leader of the LongCat team. He comprehensively led the research and development of Meituan's large language model, multi-modal model and Agent agent model, and led the development and implementation of the trillion-parameter large model LongCat-2.0. LongCat-2.0, released by Meituan on June 30, is the industry’s first large-scale model with trillions of parameters that relies on domestic computing power to complete training and inference throughout the entire process. This model adopts MoE architecture, with a total parameter of 1.6 trillion, and a single Token activation parameter of approximately 48 billion. It natively supports 1M ultra-long context, and is deeply adapted to mainstream Harness such as Claude Code. Meituan quickly clarified the personnel rumors, demonstrating its strategic determination and the stability of its core team in the research and development of independent and controllable large models. via AI News (author: AI Base)