Tencent unveils hy image 3.5 preview model for advanced image generation and multi-turn dialogue

Tencent has released the Hy Image 3.5 preview model, designed for professional image creation, supporting text-to-image, image-to-image, and multi-turn dialogue generation. The model has been tested by over 300 professional designers through a blind comparison method, showing a 30% improvement over Hy Image 3.0 and matching Seedream 5.0 Pro. It is already integrated into several apps, including Yuanbao and ima, and is available via Tencent Cloud API. The model supports up to five image references per request and offers high-resolution outputs. Tencent claims the model can handle both everyday tasks like photo editing and professional applications such as commercial posters and UI design. The preview version was rolled out two months ahead of schedule in Yuanbao, with usage increasing by over 50%. Tencent also highlighted its ongoing development cycle, with major updates every two months since February.