Skip to main content
ComfyUI Interface

What is ComfyUI?

ComfyUI is a powerful and modular visual AI engine that lets you design and execute advanced stable diffusion pipelines using a graph/nodes/flowchart based interface. Create complex workflows without writing any code. Available on Windows, Linux, and macOS with support for NVIDIA, AMD, Intel, Apple Silicon, and Ascend GPUs.

Key features

Node-based interface

Experiment with complex Stable Diffusion workflows using an intuitive graph interface. No coding required.

Extensive model support

Works with SD1.x, SD2.x, SDXL, SD3, Flux, and many more image, video, audio, and 3D models.

Smart memory management

Automatically runs large models on GPUs with as low as 1GB VRAM through intelligent offloading.

Optimized execution

Only re-executes parts of the workflow that change between runs, saving time and resources.

Get started

Installation

Install ComfyUI on Windows, Linux, or macOS with step-by-step instructions for all platforms.

Quick start

Create your first workflow and generate images in minutes with our beginner-friendly guide.

What you can do with ComfyUI

Image generation and editing

  • Generate images from text prompts with various Stable Diffusion models
  • Apply LoRAs and embeddings to customize your outputs
  • Use ControlNet and T2I-Adapter for precise control over generation
  • Inpainting and outpainting with specialized models
  • Upscale images with ESRGAN, SwinIR, and other models

Advanced workflows

  • Area composition to control different regions of your image
  • Model merging to create custom hybrid models
  • Two-pass generation for high-resolution outputs
  • GLIGEN for grounded language-to-image generation
  • Load workflows directly from PNG, WebP, and FLAC files

Video and beyond

  • Video generation with Stable Video Diffusion, Mochi, LTX-Video, and Hunyuan Video
  • Audio generation with Stable Audio and ACE Step
  • 3D model generation with Hunyuan3D 2.0

Fully offline operation

ComfyUI works completely offline by default. The core will never download anything unless you want it to. Optional API nodes for paid external models can be disabled with --disable-api-nodes.

Keyboard shortcuts

ShortcutAction
Ctrl + EnterQueue generation
Ctrl + SSave workflow
Ctrl + OLoad workflow
Ctrl + Z / Ctrl + YUndo/Redo
SpacePan canvas
Double-ClickOpen node search
.Fit view to selection
On macOS, use Cmd instead of Ctrl for all shortcuts.

Community and support

Build docs developers (and LLMs) love