Midjourney announces MJ-UX, a UI-first image model with exported component assets

AI · 5 min read

MJ-UX departs from Midjourney's generative art focus to concentrate on interface clarity, componentization, and export hygiene. Outputs include layered PNGs, SVGs, and JSON describing component boundaries and suggested tokens for color and typography.

The model can take prompts like 'mobile checkout flow in Scandinavian style with high contrast and compact spacing' and create multiple layout variations. Midjourney added a 'dev-friendly' output mode that packages assets with naming conventions and a basic accessibility checklist.

MJ-UX is being rolled out as a subscription add-on; Midjourney is offering a developer API for integration into internal design tooling. They acknowledged the model is tuned heavily to avoid hallucinating brand-specific elements and provide an opt-in dataset pipeline for customers who want brand-consistent fine-tuning.