Skip to main content

OpenAI Researcher Tian Yonglong May Join Tencent's Horizon AI to Lead Multimodal Efforts

Another key figure from OpenAI is reportedly making the leap to Tencent. Tian Yonglong, a former researcher at the AI lab, is said to have joined Tencent's Horizon AI unit, where he will lead the development of multimodal models—think visual language models that can understand and generate content across images and text.

If confirmed, this would be the second high-profile researcher from OpenAI's core team to land at Tencent's Horizon AI, following Yao Shunyu, who joined the company last September as Chief AI Scientist. The two previously worked together at OpenAI, witnessing the rapid evolution of the GPT series and multimodal technologies. Now, they're reuniting under Tencent's umbrella.

From Tsinghua to MIT: A Stellar Academic Path

Tian Yonglong's academic journey reads like a blueprint for top AI talent. He graduated from Tsinghua University, then earned a master's degree from the Chinese University of Hong Kong, where he studied under renowned computer vision experts Tang Xiao'ou and Wang Xiaogang at MMLab. He later pursued a PhD at MIT's EECS department, advised by generative model pioneer Phillip Isola, graduating in 2022. His research focus shifted from traditional visual recognition to generative models and representation learning.

After his PhD, Tian worked as a senior research scientist at Google Research and Google DeepMind before joining OpenAI as a Member of Technical Staff. His Google Scholar profile boasts nearly 29,000 citations, with an h-index of 35. In 2026 alone, his work was cited over 3,400 times. One of his most influential contributions is supervised contrastive learning, which extended contrastive learning to supervised settings and influenced later models like SimCLR and MoCo.

Tencent's Multimodal Team: Building the Puzzle

Tian's potential addition is the latest in a series of moves by Tencent's Horizon AI to bolster its multimodal team. At the end of 2024, Liu Wei, the former head of multimodal coordination, left to start his own venture focused on video generation. The role was temporarily filled by Tencent vice president Jiang Jie. In April 2025, Tencent formally established a Large Language Model Department and a Multimodal Model Department within its TEG system, addressing earlier issues of scattered resources and virtual teams.

With Tian on board, Tencent gains not only deep expertise in visual-language models but also a researcher with a proven track record in generative AI. The company is clearly racing to assemble a world-class team capable of competing in the fast-evolving multimodal AI landscape.

Key Points

  • Tian Yonglong, former OpenAI researcher, reportedly joins Tencent's Horizon AI to lead multimodal model direction.
  • He reunites with Yao Shunyu, another OpenAI alumnus who joined Tencent in 2025.
  • Tian holds a PhD from MIT, with nearly 29,000 Google Scholar citations and key contributions to contrastive learning.
  • Tencent has been restructuring its AI teams, establishing dedicated multimodal and LLM departments in 2025.