China's Tencent has released a preview version of its image generation AI model, "Hy Image3.5."
Model ReleasesPricing & LimitsTencentHy Image3.5
Tencent Releases Preview Version of Image Generation AI "Hy Image3.5"
This article is a translation. Read the Japanese original
This single model supports both "Text to image" generation and "Image to image" generation. It features a maximum output size of 2K and is characterized by its ability to accurately render multilingual text, symbols, and small characters. Additionally, the model is capable of performing image editing in response to user prompts.
According to Tencent's announcement, human evaluations show that it achieves 30% improved accuracy compared to the previous model, "Hy Image3.0."
The preview version of "Hy Image3.5" is available through the Tencent Cloud API. The pricing is set at 0.024 dollars per 1 images, with charging applied only to generated images.
Furthermore, the AI creative tool "Miora" and the AI video production tool "OnSolo" offer free usage for a limited period of 2 weeks. The preview version is also available on GMI Cloud, supported by NVIDIA, and on ComfyUI (use on ComfyUI requires account registration and incurs charges per process).
Sources
- Tencentが画像生成AI「Hy Image3.5」のプレビュー版をリリース （GIGAZINE、2026-09-25）