On August 27, sources revealed that Gu Qi, the principal researcher of Meituan's LongCat Team and the head of its General Agent team, is set to step down from his current role with plans to embark on an entrepreneurial journey in the realm of multimodal models. Presently, he is actively engaging with venture capital firms.

Gu Qi stands as a pivotal contributor to the LongCat series of models. He led the development of the LongCat-Flash-Thinking-2601 model, overseeing the comprehensive construction of the Agent's overall pipeline and conducting a portion of the post-training research on the foundational model. A graduate of the Department of Mathematical Sciences at Tsinghua University, Gu Qi has a track record of publishing multiple papers at prestigious international academic conferences, including ICLR and ICML.

Prior to his tenure at Meituan, Gu Qi served as a researcher at Alibaba and Ant Group, where he played a key role in establishing the RLHF infrastructure for the Ant Bailing Large Model. In mid-2024, he transitioned to the Seed-LLM-Code team at ByteDance, focusing on enhancing model performance in Competitive Coding scenarios and delving into RL Scaling paradigms.