xAI multimodal lead joins Tencent's Hunyuan model team
xAI's multimodal understanding lead has decamped for Tencent, signaling a fresh push in China's race to build competitive vision-language models. The move is the latest in a pattern of senior AI researchers crossing between Western labs and Chinese tech giants — and it lands at a moment when multimodal capability is becoming the differentiator for both consumer products and enterprise agents.
Lin Xudong, who led multimodal understanding at xAI and contributed to Google DeepMind's Gemini pre-training and post-training, has joined Tencent's Hunyuan large-model team, according to reports from Chinese tech outlets Top华人科创社区 and 知潜KnowFuture. Lin's resume reads like a tour of the world's top AI labs: he was the 2014 Shaanxi provincial top scorer in the college entrance exam, earned a mathematics and physics degree from Tsinghua in 2018, and completed a Columbia CS PhD in 2023. He then joined DeepMind in January 2024 as a core contributor to Gemini's multimodal pre-training and post-training pipeline, before moving to xAI in November 2025 to lead multimodal understanding. His research spans multimodal content understanding, representation learning, video analysis, and generative models — and he is the first author of VX2TEXT, a notable work on vision-to-text grounding. Tencent reportedly plans to put Lin in charge of Hunyuan's multimodal foundation model, the component that lets models process images, video, and audio alongside text. The hire is significant because Tencent has been playing catch-up in the foundation-model race: while its Hunyuan model has shown strength in text generation and code, its multimodal capabilities have lagged behind Alibaba's Qwen-VL series and ByteDance's Doubao. Bringing in someone who helped build Gemini — one of the few models that genuinely competes with GPT-4o on multimodal benchmarks — gives Tencent a direct injection of institutional knowledge about how to train and align vision-language models at scale.
ByteDance's AI wearables product lead, Liu Rui (code name "Muràn"), has also departed the company, according to XR Vision. Liu led multiple AI hardware product lines at ByteDance, including the company's closely watched AI glasses project, and was involved in defining the product vision from its earliest stages. Before ByteDance, he worked at Huawei on HarmonyOS NEXT's AI-native intelligence features. His departure adds to a broader pattern of turnover in ByteDance's hardware ambitions: the company has been scaling back some wearable bets while doubling down on its Doubao model ecosystem. The exit is notable because AI glasses remain one of the most contested consumer form factors in China, with Xiaomi, Meta, and a cluster of startups all racing to ship viable products. Losing a key product leader mid-cycle could delay ByteDance's timeline.
What to watch: Whether Tencent announces a new multimodal model update in the coming weeks — Lin's hiring suggests one is in the pipeline.
Do you think talent poaching between Western labs and Chinese tech giants will accelerate as multimodal models become the new frontier? Tell us in the comments.
Sources: 雷峰网 Leiphone — morning roundup · Top华人科创社区 via 雷峰网 · XR Vision via 雷峰网
