Tencent has launched Hy Image 3.5 Preview, its latest image-generation model, and said it will integrate the model across Tencent products including the Yuanbao chatbot and film-editing and design tools, the company said in a statement on Tuesday (Sep 22).
Hy Image 3.5 and product integration
The Hy Image 3.5 Preview model delivers a performance boost over its predecessor and is being rolled into Tencent services as the company expands its AI use beyond text. Tencent said the move supports its vision of building AI that performs tasks for the billion-plus users of its WeChat ecosystem.
The launch coincided with rival Alibaba Group Holding opening its AI conference, underlining Tencent’s intent to catch up with industry leaders such as ByteDance. Tencent said Hy Image 3.5 was tested among hundreds of in-house designers and achieved results on par with ByteDance’s Seedream 5.0 Pro.
Testing, comparisons and product claims
Tencent also stated that the new model is slightly better than Nano Banana Pro from Alphabet’s Google and Qwen-Image-3.0 Pro from Alibaba, although the company did not quantify those quality claims. Tencent employs thousands of game designers and artists who the company said may benefit from the tool.
Market context and talent strategy
The release comes as Tencent seeks to narrow a gap with Big Tech peers in producing boundary-pushing large language models. The company recently hired former OpenAI researcher Yao Shunyu, who now serves as chief AI scientist and has criticised training models solely for leaderboard scores, saying Tencent will focus on product integration and solving real-world problems. In July, former OpenAI researcher Tian Yonglong joined Tencent to lead vision-language models under the Hunyuan team.
Tencent’s shares gained more than 7 percent in Hong Kong after the announcement. Market observers have drawn comparisons between Tencent and Meta, noting that WeChat’s social ecosystem could support a personal AI assistant at scale, and that agent-style AIs such as Meta’s Muse have attracted strong user interest on mobile app stores.
