Tencent officially luanched Hy Image3.5 preview, its latest professional image generation model on September 22. The release comes as Tencent Hy accelerates its multimodal model development and places greater emphasis on practical applications, cost-effectiveness, and real-world business requirements..
Hy Image3.5 preview achieved a 30% higher win rate in human evaluations than its predecessor, Hy Image3.0. In an in-house Good/Same/Bad (GSB) blind evaluation involving hundreds of designers, the model performed on par with Seedream 5.0 Pro and slightly better than Nano-Banana Pro and Qwen-Image-3.0 Pro..
The model is already being deployed across Tencent’s AI-native product ecosystem. Products integrating Hy Image3.5 preview include the Yuanbao AI assistant, WorkRally for professional film and television production, OnSolo for drama and game content creation, the Miora creative agent, the WorkBuddy office agent workspace, and the ima AI knowledge assistant..
Hy Image3.5 preview is also accessible through the Tencent Cloud API. International users can experience the model through the Miora and OnSolo platforms.. WorkRally is an industrial-grade AI creation platform for premium content production.
The introduction of Hy Image3.5 preview expands its capabilities in character sheet generation, precise text layouts, and single-element editing.. Miora (miora.design), Tencent’s first in-house creative agent, entered full rollout on July 22. Built on an agent architecture with the same foundation as WorkBuddy’s, Miora is positioned as a multimodal agent for creative work.
Hy Image3.5 preview will become one of the image generation models available within the platform, with a two-week free trial offered to users.. OnSolo (onsolo.ai) is an AI-native platform developed for short drama, animation, and interactive film production. Starting from a single idea, it can assist users in creating scripts, character sheets, storyboards, and finished video.
With Hy Image3.5 preview, the platform gains enhanced character sheet generation, precise text rendering, and single-element editing. Existing members will receive two weeks of complimentary access, while new users will receive 110 free images when they sign up.. Hy Image3.5 preview supports text-to-image and image-to-image generation, allowing users to upload as many as five reference images at a time.
The model’s improvements focus on three areas: text rendering and layout, realism and stylistic expressiveness, and consistency during image editing.. The expanded capabilities allow Hy Image3.5 preview to address both consumer and professional scenarios. Everyday applications include photo retouching, portrait enhancement, and educational cards.
Professional use cases include commercial posters, storyboards for television series, film and television promotional materials, UI design, creative illustration, brand visuals, and e-commerce assets.. Hy Image3.5 preview continues the design approach of Hy Image3.0 while improving its understanding of both language and visual information.
Compared with Hy Image3.0, the new preview model delivers better image generation results at a lower cost.. API access through Tencent Cloud is priced at US$0.024 per 2K image, with users charged only for output images. This pricing is intended to make professional-grade image generation more accessible across a wider range of creative and commercial workflows..
The release follows a series of model updates from Tencent Hy. Since beginning an infrastructure overhaul in February, the team has launched a major update every two months on average. Its development process increasingly connects model research with practical business use cases, while preview releases allow real-world feedback to be incorporated ahead of full model launches..
Hy Image3.5 preview was developed using this approach. Feedback from business applications helped Tencent Hy bring the visual aesthetics of the model’s output more closely in line with the preferences of professional users.. In August, Tencent Hy released and open-sourced Hy4 preview, demonstrating its capabilities in productivity tasks across coding, office work, and scientific research.
In an internal blind evaluation involving 163 in-house experts and 203 engineering tasks, Hy4 preview achieved an average score of 2.99 out of 4. This compared with 2.92 for GLM 5.3 and 2.94 for Kimi K3.. More recently, Tencent Hy merged its large language model and multimodal teams to establish a foundation models department.
The restructuring brings the two areas under a unified organization and is expected to support faster development of multimodal models. Hy Image3.5 preview is the latest model to emerge during this broader expansion.. Featured image credit.
Tags: tencenttrends
