ComfyUI

ComfyUI is a node-graph interface and backend for building generative AI pipelines. Users connect models, processing steps and outputs on a canvas, where each parameter is visible and adjustable. It covers image, video, audio, 3D and text workflows, and supports a wide range of open models (for example Stable Diffusion, SDXL, Flux, Qwen Image, Wan, LTX-Video, HunyuanVideo, ACE-Step, Hunyuan3D). Partner nodes give access to closed models such as Nano Banana and Seedance.

The project was created by comfyanonymous in January 2023 and is now maintained by Comfy Org. The core is GPL-3.0 licensed. The README describes asynchronous queueing, partial graph re-execution, smart VRAM/RAM management, model offloading, quantized model support, and weight streaming that lets large models run on as little as 4 GB VRAM plus 8 GB RAM. It also offers subgraphs, workflow templates, an App Mode that exposes workflows through a simple UI, and API endpoints for production pipelines.

It can be used through Comfy Desktop (Windows and macOS), a manual or portable install (Windows, Linux, macOS; NVIDIA, AMD, Intel, Apple Silicon, Ascend), or Comfy Cloud, a paid hosted service with a free starter allowance. Comfy also offers a Developer Platform for deploying workflows as APIs, Enterprise plans, and Comfy Agent, which plans, builds and runs workflows from a text description. Cloud usage is credit-based and has workflow runtime limits (30 minutes, 1 hour on Pro). Cloud supports a subset of community custom nodes.