AntLing Open-Sources Ming-Image-0.1-Design Model Family for UI and Slide Generation
AntLing (Ant Group's inclusionAI team) has open-sourced the 6.15-billion-parameter Ming-Image-0.1-Design model family under an MIT license, including specialized Agent skills for UI design and image-to-editable-PPT creation. The flagship model currently holds the #1 ranking among open-weight models on Artificial Analysis’s UI/UX Design leaderboard. This release significantly advances open-source generative AI capabilities in graphic design, UI/UX prototyping, and document conversion. By supporting native RGBA transparency and layer separation, it equips developers with powerful, customizable open-weight tools for production design workflows. The 6.15B parameter text-to-image models natively output RGBA formats with transparent backgrounds and layered visual assets. The accompanying open-source skills enable direct automated generation of structured UI layouts and editable presentation slides.
## BACKGROUND
Standard image generation models usually produce flattened, single-layer raster images, making it challenging for designers to extract individual UI elements or transparent assets. Dedicated design models solve this by supporting layered compositions and RGBA channels, allowing direct integration into tools like Figma or PowerPoint.