Powerful node-based UI for Stable Diffusion and other AI image models — build custom generation pipelines with a visual graph editor.
ComfyUI is the most powerful open-source UI for running Stable Diffusion and other AI image generation models. Instead of simple prompt boxes, ComfyUI gives you a node-based graph editor where every step of the generation process — model loading, sampling, encoding, upscaling, LoRA application — is a visible, connectable node. You can build any workflow imaginable: img2img pipelines, inpainting chains, ControlNet rigs, video generation, and more.
Everything runs locally on your GPU (NVIDIA, AMD, or Apple Silicon). The community has built thousands of custom nodes extending ComfyUI with new models and capabilities. Free and open-source.
Pricing model: Free
Common questions about ComfyUI
The ComfyUI team released OPENjutsu, an open-source app that swaps a person in a video for a character from a single image while keeping the scene and its sound.
Upload a clip and a character image, then choose a 5 to 15 second section to swap. You can try it on Comfy’s app hub or run your own copy from GitHub.
Running your own copy requires a Comfy Developer Platform account, an API key and a deployment with the listed models.
ComfyUI now supports MOSS-Image, OpenMOSS’s 5B open-weight image model. It accepts structured layout prompts and can render legible text.
ComfyUI's MCP server now runs locally and is open source, following the Comfy Cloud MCP release in June.
The local server gives your agent access to your actual installation, including custom nodes and models on disk. It can fetch files, start your ComfyUI instance, and build a workflow to the point where it runs.
The Krea 2 API is now available in ComfyUI, bringing style transfer, moodboard input and adjustable creativity settings to its node-based workflows.
Official videos and tutorials for ComfyUI.
Other products to explore and follow.
The people behind ComfyUI
Original creator
AI image creation and precision editing inside Google Workspace — generate from a prompt or a photo, edit individual objects, refine or translate text, and work on images with your team in Docs, Slides and Drive.
AI art platform — generate images from text, train custom models, edit with inpainting, and create consistent characters.
Node-based AI workflow platform by Figma — connect 12+ AI models with professional editing tools and share as simple apps.
AI image and video generation platform — create production-quality visuals with fine-tuned models and a generous free tier.
Generate and edit images with ChatGPT — create illustrations, logos, mockups, and photorealistic visuals from text prompts.
