Tencent Hunyuan multimodal understanding lead Hu Han leaves company to start venture

Hu Han, who led multimodal understanding at Tencent's Hunyuan team, recently resigned to start his own venture. He previously served as principal researcher at Microsoft Research Asia's visual computing group. After joining Tencent in early 2025, he worked on vision large models and later moved to the Frontier advanced technology research group under the LLM department, reporting to Yao Shunyu. Hu also contributed to world model development. Meanwhile, Yao Shunyu is reorganizing teams, and Hu's former group may pivot to world model research. As of press time, Tencent had not responded. Since mid-2026, personnel changes around Hunyuan have continued. In early July, former OpenAI researcher Tian Yonglong joined the LLM department to replace Hu, focusing on vision-language models. Yao Shunyu has been consolidating resources, merging AI Lab staff into the LLM department and increasing hiring. At the same time, Tencent is reassessing the returns on multimodal understanding, which has limited commercial potential compared to reasoning and agent capabilities. Given finite computing resources, Tencent is prioritizing frontier research over mature multimodal understanding.