Top Researcher Jiahui Yu Announces Departure from Meta Despite High-Profile Recruitment Efforts

Jiahui Yu, a core member of Meta's Superintelligence Labs (MSL) and head of multimodal research, officially announced his departure on Friday (the 14th), revealing plans to found a new company.

Since joining Meta in June 2025, Yu has led his team in launching the Muse series of native multimodal models, marking a significant advancement in Meta's recent efforts in generative artificial intelligence (Generative AI).

### From Academic Prodigy to Top Engineer

Yu's academic and professional journey has consistently revolved around 'multimodal' technology. A graduate of the Young Scholars Program at the University of Science and Technology of China, he later earned his Ph.D. from the University of Illinois Urbana-Champaign (UIUC), studying under renowned computer vision expert Thomas Huang.

Before joining Meta, Yu was already recognized as a top-tier talent in the AI community. He contributed to Google's Gemini project and served as head of OpenAI's Perception team, leading the development of GPT-4o and the o-series reasoning models.

At Meta, Yu served as the technical lead for multimodal systems within TBD Lab, part of the Superintelligence Labs. In his departure announcement, he described his past year working with CEO Mark Zuckerberg and Chief AI Officer Alexandr Wang to build TBD Lab as "highly inspiring and fulfilling."

### Comprehensive Product Line for the Muse Series

During Yu's 14-month tenure, TBD Lab executed a rapid sequence of product launches. Starting with the foundational model Muse Spark in April 2026, the shift signaled Meta's strategic pivot from the Llama series to the more capable multimodal Muse series. The team subsequently released Voice Mode, Muse Image, and Muse Video, comprehensively covering speech, image, and video generation.

Just eight days before his resignation announcement, Meta released Muse Spark 1.2, further enhancing agent capabilities and long-horizon task execution potential.

### The Controversy Over Nine-Figure Compensation

Market speculation has widely circulated that Meta lured Yu from OpenAI with a record-breaking annual compensation package of up to "$100 million." While Meta has not confirmed the exact figure, CTO Andrew Bosworth clarified that the high compensation is not a one-time signing bonus but a four-year total compensation plan including stock, bonuses, and performance-based incentives.

Regarding this high-stakes talent acquisition, OpenAI CEO Sam Altman commented: "Missionaries will eventually beat the mercenaries."

With Yu departing shortly after completing key milestones to launch his own venture, tech writer Gergely Orosz pointed out that internal reorganizations and efficiency adjustments at Meta have created unease among engineers, prompting many senior experts to remain open to external opportunities.

### Meta's Talent Retention Challenge

Regarding his entrepreneurial direction, Yu has not disclosed details about his new company, only stating that he is fully focused on a problem that is "extremely important for humanity's future but currently underexplored."

Data from research platform alphaXiv shows that Meta has lost over 200 prominent researchers, including Yu, with a total of 929 researchers now categorized as having previously worked at Meta but since departed. While Meta can close technological gaps through short-term, high-value compensation, repairing organizational trust and retaining top-tier talent remains a critical challenge moving forward.