Skip to main content
Ming Image 0.1 Design is a 6B open-weights text-to-image model from inclusionAI (Ant Group), released under the MIT license. It targets design work rather than photographic realism: UI screens, infographics, posters, and other layouts that carry a lot of text. Outputs are RGBA with a real alpha channel, so a transparent result can be dropped straight into a design file. The workflow puts a prompt-rewriting stage in front of the sampler. Qwen3.8-27B expands a short idea into a structured layout caption before sampling, and the enhancer can be switched off when you want your text sent exactly as typed. Related links

​
What Ming Image 0.1 Design is good at

  • Long, structured layouts: pages, posters, and infographics with many sections, composed section by section
  • Text rendering: strings quoted in the prompt are followed closely, so labels, prices, and headings stay legible
  • Transparent output: RGBA images with a real alpha channel for compositing isolated elements
  • Prompt rewriting: an instruction-following vision language model expands a short brief into a detailed layout caption
  • Open weights: MIT licensed, so the checkpoints can be used and fine-tuned freely

​
Example outputs

Text-to-image generation of a text-heavy layout, saved as a single PNG: Ming Image 0.1 Design text to image example
Make sure your ComfyUI is updated.Workflows in this guide can be found in the Workflow Templates. If you can’t find them in the template, your ComfyUI may be outdated.If nodes are missing when loading a workflow, possible reasons:
  1. You are not using the latest ComfyUI version (Nightly version)
  2. Some nodes failed to import at startup

​
Ming Image 0.1 Design text-to-image workflow

Generate a design composition from a text prompt, with the model handling layout, typography, and background transparency. Ming Image 0.1 Design text-to-image workflow preview

Download Workflow

Download JSON or search “Ming Image 0.1 Design” in Template Library

​
Steps to run the workflow

  1. Enter the Text to Image (Ming Image 0.1 Design) subgraph with the Enter subgraph button at the bottom of that node
  2. Write your prompt in the prompt field, describing the layout section by section and putting any copy you want rendered inside quotes
  3. Choose the canvas with the ResolutionSelector node: 2048 x 2048 gives the best quality, 1024 x 1024 is faster
  4. Leave the prompt enhancer on to let Qwen3.8-27B expand your prompt, or switch enable_PE off to send your text exactly as typed
  5. Queue the prompt. The decoded image appears in the PreviewImage node
  6. For a transparent background, prepend one RGBA phrase to your prompt, for example RGBA, 4-channel, transparent background

​
Model downloads

ming_image_0.1_design_int8_convrot.safetensors

Diffusion model for Ming Image 0.1 Design (loaded by the template).

ming_image_0.1_ling_mini_2.0_int8_convrot.safetensors

Text encoder for Ming Image 0.1 Design (loaded by the template).

qwen3.8_27b_w4a8.safetensors

Prompt-rewriting text encoder used by the prompt enhancer.

ming_image_vae_bf16.safetensors

VAE for Ming Image 0.1 Design.
Model Storage Location
The same repository also publishes full-precision checkpoints, ming_image_0.1_design_bf16.safetensors and ming_image_0.1_ling_mini_2.0_bf16.safetensors, plus _layer variants of both the diffusion model and the text encoder.

​
Prompt guide

​
Get started

  1. Update ComfyUI to the latest version. Ming Image 0.1 support is in the newest ComfyUI builds
  2. Go to Template Library, search for Ming Image 0.1 Design
Comfy Desktop and Comfy Cloud follow stable releases, so a model that is only supported in the newest ComfyUI build may not be available there yet.