Ming-Image UI Design-Specialized Image Generation Model: ComfyUI Paths

September 24, 2026September 30, 2026

At a Glance

Values determined by this site’s code when the information was collected. Dates are JST.

Overview

It is reported that Comfy-Org has released Comfy-Org/Ming-Image, a repackaged version of inclusionAI/Ming-Image-0.1-Design developed by inclusionAI, making it available for use in ComfyUI. This repository organizes text-to-image model files to match ComfyUI’s directory structure. Note that this information has not been officially confirmed at this time.

Specifications

The specifications of the base model inclusionAI/Ming-Image-0.1-Design and the provided file structure are as follows.

Basic Specifications

• Parameters: 6B
• Output Specifications:
  • Recommended Resolution: 2048 x 2048 – Fast Generation Resolution: 1024 x 1024 – Background: Supports RGBA (transparent background) output
• Recommended Settings:
  • Sampling Steps: 12 – CFG Scale: 1.0 – Precision: BF16
• Prompt Expansion (PE): Mentions the use of Ling-3.0-flash-VL or qwen3.8-27B
• License: MIT

File Placement in ComfyUI

The repackaged files are intended to be placed in the following ComfyUI directories.

• models/diffusion_models/: ming_image_0.1_design_bf16.safetensors, etc.
• models/text_encoders/: ming_image_0.1_ling_mini_2.0_bf16.safetensors, etc.
• models/vae/: ming_image_vae_bf16.safetensors

Performance

According to the model card, this model is specialized for generating UIs, infographics, posters, and other visual designs containing heavy text. It is reported to have the capability of generating complete visual compositions rather than just individual elements. It also supports transparent background generation in RGBA format, allowing the generation of images with transparent backgrounds by prepending specific recommended RGBA phrases to the prompt.

Specific benchmark scores or numerical comparisons with other models are not included in the provided materials.

Strengths and Use Cases

Based on the documentation for the original model inclusionAI/Ming-Image-0.1-Design, this model is specialized not only for generating general photos and illustrations but also for creating graphic designs and UI/UX designs that include text elements.

Specifically, the following use cases and features are envisioned:

• Generation of Text-Heavy Visual Designs:
  It is suitable for generating images where text and layout are crucial, such as posters, infographics, and various UI screens. The tag information also includes graphic-design and text-rendering, pointing to an expected output of complete visual compositions that balance visual and textual elements.
• Transparent Background (RGBA) Image Output:
  It supports output in RGBA format containing an alpha channel in addition to standard RGB images. It is reported that by specifying certain recommended phrases at the beginning of the prompt, users can directly generate parts or graphic assets with transparent backgrounds. This omits post-processing background removal and can be directly utilized for asset creation in web design and presentation slides.
• Integration with Design Assistance and Slide Creation Skills:
  The original model’s repository indicates the existence of recommended tools such as UI design skills (ling-ui-design) and skills to convert images into editable slides (image-to-editable-ppt), suggesting a practical workflow for prototyping presentation materials and screen layouts.

It is reported that this package, Comfy-Org/Ming-Image, repackages these features for easy handling within ComfyUI, a node-based GUI environment.

Distributed Files

Weight files published in Comfy-Org/Ming-Image, listed by this site from the Hugging Face API. Sizes are the actual file sizes.

The ComfyUI folder is shown only where the repository’s layout mirrors ComfyUI’s models/ directory.

How to Get It

Model files for Comfy-Org/Ming-Image are available from the corresponding Hugging Face repository. Since it is not a gated model, users can download it directly without prior agreement to terms of use.

Provided weight files are distributed in safetensors format, subdivided by model type and precision. Files according to the intended use should be placed in the respective folders within the ComfyUI installation directory.

The list of placement destinations and files specified in the model card is as follows:

Diffusion Model Placement Destination

ComfyUI/models/diffusion_models/
– ming_image_0.1_design_bf16.safetensors
– ming_image_0.1_design_int8_convrot.safetensors
– ming_image_0.1_design_layer_bf16.safetensors
– ming_image_0.1_design_layer_int8_convrot.safetensors

In addition to the standard BF16 precision model, quantized int8_convrot versions and layer versions supporting layer separation are available.

Text Encoder Placement Destination

ComfyUI/models/text_encoders/
– ming_image_0.1_ling_mini_2.0_bf16.safetensors
– ming_image_0.1_ling_mini_2.0_int8_convrot.safetensors
– ming_image_0.1_ling_mini_2.0_w4a8.safetensors
– ming_image_0.1_ling_mini_2.0_layer_bf16.safetensors
– ming_image_0.1_ling_mini_2.0_layer_int8_convrot.safetensors

It is reported that multiple lightweight and quantized variants, such as int8_convrot and w4a8 versions in addition to BF16 precision, are available for the text encoder side.

VAE Placement Destination

ComfyUI/models/vae/
– ming_image_vae_bf16.safetensors

Users will select the necessary precision combinations according to their environment and workflow, place them in the above folders, and load them in ComfyUI.

Can You Run It Locally?

The publisher distributes this model as safetensors.

License — mit (Commercial use allowed): Permits commercial use, modification and redistribution, provided the copyright notice and license text are retained.

Compiled by this site’s code from the published formats and the license field. License summaries are not legal advice — check the publisher’s original terms before relying on them.

Related Articles

• YuE2: 12GB+ VRAM, ComfyUI Paths
• LTX-2.5-22b-IC-LoRA-Restore Video Generation Model: File List

What to Read Next

• Engines that run this model → ComfyUI
• Formats this model is available in → Safetensors format guide and models
• Learn about the publisher → inclusionAI (Ant Group): models, licenses and articles

Sources

• Comfy-Org/Ming-Image (Hugging Face)
• inclusionAI/Ming-Image-0.1-Design (Hugging Face)
• inclusionAI/Ming-Image-0.1-Design-Layer (Hugging Face)

Update History

• 2026-09-24: Rebuilt the article: Added a “can you run it locally?" verdict from the distribution formats, converted builds, engine registries, license terms and measured compression
• 2026-09-25: Verified the content against the official primary source.
• 2026-09-25: Changed the title to show what the article covers (VRAM requirements, file list, etc.).