Tencent Launches Hunyuan Hy Image3.5 Preview for Multi-Turn AI Image Generation
Tencent's Hunyuan team has released the Hy Image3.5 preview model, designed for professional-grade image generation supporting text-to-image, image-to-image, and multi-turn conversational editing. The model is already integrated into consumer applications like Yuanbao and ima, as well as Tencent Cloud platform services. This release reflects Tencent's rapid bi-monthly model iteration cadence, expanding generative visual capabilities across both enterprise and consumer tools. By offering low-cost API access via Tencent Cloud TokenHub, Tencent aims to capture professional workflows in media, e-commerce, and commercial graphic design. In internal GSB blind evaluations, Hy Image3.5 preview showed a 30% performance gain over Hy Image3.0 and performed on par with Seedream 5.0 pro. It supports up to 5 reference images per generation, outputs resolutions up to 2K, and is priced at 0.15 RMB ($0.024 USD globally) per 2K output image via API.
## BACKGROUND
Tencent Hunyuan is Tencent's suite of proprietary AI foundation models spanning large language models and multimodal generative systems. Image generation models synthesize visuals based on user prompts and reference photos, with recent industry trends focusing on multi-turn dialogue to enable continuous contextual editing.