Repolode
Submit
Comfy-Org

Comfy-Org/ComfyUI

A node-graph GUI, API and backend for running diffusion models to generate images, video, audio and 3D locally.

Open-source alternative to:Midjourney

View on GitHubWebsite
Stars
138,477
Forks
16,490
Open issues
5,252
Last commit
8 hours ago

What is it?

ComfyUI lets you build generation pipelines visually by connecting nodes in a graph, without writing code. Workflows can be saved and loaded as JSON, and reusable subgraphs and templates make complex pipelines easier to share and run from a simpler app-style interface.

The project supports a broad range of open models, including Stable Diffusion, SDXL, Flux, Wan and LTX-Video, along with tools for inpainting, upscaling, model merging, depth estimation and segmentation. It emphasizes efficient local execution with VRAM and RAM management, model offloading and quantized model support, and runs on Windows, Linux and macOS, as well as NVIDIA, AMD, Intel, Apple Silicon and Ascend hardware.

For integration, a local API lets workflows be run from other applications. Custom nodes extend the system, and the core works offline unless optional API nodes are requested.

Key features

  • 01Visual node graph for building image, video, audio, 3D and text workflows without code.
  • 02Save and load workflows as JSON, and recover workflows and seeds from generated PNG files.
  • 03Supports many open image, video, audio and 3D models, plus text generation models.
  • 04Inpainting, outpainting, upscaling, model merging and frame interpolation are built in.
  • 05Runs locally with VRAM/RAM management and quantized model support, and offers a local API.
  • 06Extensible through custom nodes and additional model path configuration.

Use cases

  • Generating and editing images or short videos with fine control over each model and parameter.
  • Packaging complex generation pipelines as simple app-style interfaces for non-technical users.
  • Integrating diffusion model workflows into production pipelines via the local API.
  • Running large open models on consumer GPUs with limited VRAM.

Topics

#stable-diffusion#diffusion-models#image-generation#video-generation#node-editor#pytorch#local-inference#workflow#ai#comfy

FAQ

What is ComfyUI?
A node-graph GUI, API and backend for running diffusion models to generate images, video, audio and 3D locally. ComfyUI lets you build generation pipelines visually by connecting nodes in a graph, without writing code. Workflows can be saved and loaded as JSON, and reusable subgraphs and templates make complex pipelines easier to share and run from a simpler app-style interface.
Is ComfyUI free to use? What license does it have?
ComfyUI is open source under the GPL-3.0 license. Check the license terms before using it in commercial products.
Is ComfyUI still maintained?
Actively maintained. The last commit to ComfyUI was on Oct 18, 2026.
What can I use ComfyUI for?
Generating and editing images or short videos with fine control over each model and parameter. Packaging complex generation pipelines as simple app-style interfaces for non-technical users. Integrating diffusion model workflows into production pipelines via the local API. Running large open models on consumer GPUs with limited VRAM.
Where can I get ComfyUI?
The source code and installation instructions are on GitHub: https://github.com/Comfy-Org/ComfyUI

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects