Tencent released Hy Image 3.5 Preview, its latest image-generation model, timing the launch to the same day Alibaba Group Holding kicked off its AI conference1,2.
Tencent said that during internal testing, Hy Image 3.5 Preview achieved results on par with ByteDance's Seedream 5.0 Pro. Bloomberg characterized the release as Tencent's bid to catch ByteDance and Alibaba Group Holding in AI-generated imagery.
The launch date is notable. Alibaba Group Holding's Qwen Team open-sourced Qwen-Image-2.1, a unified text-to-image generation and editing model with a 7-billion-parameter visual generation component, on September 20, 2026[2]. Alibaba Group Holding also appointed Dayiheng Liu to lead the Qwen large language model project on the same day Tencent's release landed[1].
ANALYSIS By shipping Hy Image 3.5 Preview on the opening day of Alibaba Group Holding's AI conference, Tencent forced a direct comparison cycle with two rivals simultaneously: ByteDance on benchmark parity and Alibaba Group Holding on calendar attention.
The performance claim rests on Tencent's own internal testing rather than third-party evaluation. ◆ Until independent benchmarks or community evaluations corroborate the claimed parity with Seedream 5.0 Pro, the competitive standing of Hy Image 3.5 Preview remains provisional.
Tencent described the release as its "latest and best" image-generation model.