# ComfyUI

> Node-graph app for generating images, video, audio and 3D with local AI models

- Repository: https://github.com/Comfy-Org/ComfyUI
- Category: Image, video & audio
- Stars: 142,050 · Forks: 16,750 · Language: Python · License: GPL-3.0
- Experience: Some experience
- Page: https://ir.adyog.com/repos/comfy-org/comfyui/

Running diffusion models usually means writing scripts or using a fixed web form. ComfyUI replaces that with a canvas where each step (load a model, encode a prompt, sample, upscale, save) is a node, and you connect nodes with wires to build a workflow.

Because the workflow is a graph, you can reuse parts, swap models, and re-run only the pieces that changed. The README says it manages VRAM and RAM carefully, offloads models, and supports quantized ones, so large open models can run on modest GPUs; it claims as little as 4GB VRAM with 8GB RAM. That is the project's own claim, so expect speed to vary.

It supports many open models natively, covering Stable Diffusion, SDXL, Flux, Wan, LTX-Video and others. Optional paid partner nodes reach closed models, and an --offline flag turns those off. Workflows save as JSON, can be exposed as simple apps through App Mode, and can be called through a local API. It runs on Windows, Linux and macOS, with a desktop app, a manual install, and a paid cloud option.

## What it does

- **Visual node graph**: Build and reuse image, video, audio, 3D and text workflows by wiring nodes together. Workflows save and load as JSON, and can be recovered from supported generated media.
- **Efficient local execution**: Asynchronous queueing, partial graph re-execution, smart VRAM and RAM management, model offloading and quantized model support are all described in the README.
- **Wide native model support**: Covers image generation and editing, video, audio, 3D and vision models, and text models like Gemma and Qwen. It loads checkpoints, LoRAs, ControlNets, VAEs and upscalers.
- **Built-in editing tools**: Includes inpainting, outpainting, masks and compositing, model merging, upscaling, frame interpolation, segmentation and depth estimation.
- **API, App Mode and offline use**: Workflows can be exposed through a simple UI or called via API endpoints. Core runs fully offline, and custom nodes extend it.

## What you could build

- Generate images from text prompts with Stable Diffusion, SDXL or Flux on your own GPU.
- Build a video generation pipeline with models such as Wan or LTX-Video.
- Edit photos with inpainting, outpainting, upscaling and background removal.
- Wrap a finished workflow in App Mode so non-technical teammates can use it.
- Call workflows through the API from a production pipeline or application.

## Who it's for

Suits creators, hobbyists and developers who want hands-on control over generative models; you should be comfortable downloading model files and have a reasonably capable GPU.

## Get started

The easiest route is the desktop app for Windows or macOS, downloaded from comfy.org/download. A Windows portable build and a manual install for other systems and GPU types are also available. The README excerpt gives no commands.



Tags: stable-diffusion, node-graph, image-generation, video-generation, local-inference