Tencent
HY Image 3.5 Preview
Tencent’s image generation and editing preview

Key facts
- 24 Sep 2026
- Announced
- Tencent
- Maker
- Generation + editing
- Tasks
- Preview
- Status
Tencent’s HY Image 3.5 Preview generates and edits images through its Agent Development Platform. Tencent announced the preview on 24 September 2026.
HY Image 3.5 Preview is Tencent’s image model for generating pictures and editing existing images. Tencent announced the preview on 24 September 2026 through its Agent Development Platform.
The preview combines generation and editing
Generation starts with a written description of the desired image. Editing supplies an existing picture and an instruction about the change. Keeping those tasks in one workflow allows a user to create a draft and then adjust its subject, composition or details.
Tencent’s announcement presents the model’s intended improvements and example outputs. Those demonstrations are the maker’s evidence. For work involving a recurring subject or detailed typography, test consistency across several edits and inspect the final image at the size at which it will be published.
The release date comes from Tencent
The date used here is 24 September, the date on Tencent’s own announcement. Later listings by other services can describe another provider adding access to the same model.
Tencent’s image API documentation identifies hy-image-v3.5-preview, with text, reference-image and multi-turn workflows and output up to 4096 × 4096 pixels. The Preview name identifies the release status; check the current account requirements, available model identifier and service terms before building a workflow around it. Save the model version with important results so later changes can be compared on the same inputs.
More in Image Generation
All Image →- Black Forest LabsFLUX 3 ImageFLUX 3 Image creates and edits pictures with up to ten reference images
- IdeogramIdeogram 4.5image generation and precise editing
- AlibabaQwen-Image-2.1-TurboAn eight-step image generator with downloadable weights
- Google DeepMindNano Banana 2.1Image generation, reference pictures and conversational edits
- OpenAIGPT Image 2.5two variants, one price, the top two seats
- OpenAIGPT Image 2the leading line