Tencent Hunyuan Changes Multimodal Leadership Amid World Model Focus

Han Hu tendered his resignation from Tencent's Hunyuan multimodal role, as 36Kr reported on July 23. Former OpenAI researcher Tian Yonglong joined Tencent's LLM department in early July and is set to take over vision-language model R&D. Hu's former research group may focus on world-model research, according to 36Kr.
Tencent's Hunyuan organization is changing leadership for multimodal research after Han Hu, head of multimodal understanding, tendered his resignation, according to a July 23 report from 36Kr. The report states that Tian Yonglong, a former OpenAI researcher, joined Tencent's Large Language Model Department in early July and is set to succeed Hu in leading research and development for vision-language models (VLMs).
Hu joined Tencent in early 2025 after serving as a principal researcher in Microsoft Research Asia's Visual Computing Group, 36Kr reported. At Tencent, he led research on vision foundation models before moving into the "Frontier" advanced technology research group within the LLM department, where he oversaw multimodal-understanding research and reported to LLM department head Shunyu Yao.
Possible world-model emphasis
According to 36Kr, Hu had also worked on world models. The outlet reported that Yao had been reviewing teams across the organization and that Hu's former research group may shift toward frontier research on world models. Tencent had not responded to the report by the time of publication, according to 36Kr.
World models are broadly used to describe systems that learn representations of an environment and predict how it may evolve, often combining perception with predictive simulation. The term covers a range of research approaches rather than a single architecture, so the reporting does not establish a specific model design, training methodology, benchmark, or product timeline for Tencent.
Broader Hunyuan personnel changes
36Kr and KR Asia describe the leadership change as part of continuing personnel adjustments around Hunyuan. Their coverage reports that Tencent dissolved its AI Lab and moved core R&D staff into the LLM department, while recruitment expanded to include personnel from ByteDance's Seed infrastructure and post-training teams.
The reporting also links these organizational changes to a concentration of talent and compute resources around foundation-model R&D in Tencent's LLM department. For ML practitioners, the immediate technical takeaway is limited: Tian's remit is reported as VLM R&D, while world-model work is described only as a potential focus for Hu's former group. Concrete evidence of the direction would require public technical releases, model documentation, benchmarks, or research papers.
Key Points
- 136Kr reports Tian Yonglong is set to lead Hunyuan vision-language R&D after Han Hu tendered his resignation.
- 2Coverage links Hu's former research group to possible world-model work, while Tencent had not responded when 36Kr published.
- 3World-model research commonly combines perception and predictive simulation, making leadership changes relevant to multimodal teams tracking the field.
Scoring Rationale
The reported change affects a major Chinese AI developer's multimodal research leadership and may shape its world-model research direction. Technical details remain sparse, with no announced model, benchmark, or product release, limiting immediate implementation impact for practitioners.
Sources
Public references used for this report.
Practice with real Ad Tech data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Ad Tech problems

