ComfyUI

Open-source node-based engine for building image, video and audio generation workflows, run locally or on Comfy Cloud.

AI-poweredUnclaimed

Pricing: Open source · see prices
Platforms: Windows, macOS, Linux, Web, API
Categories: Design & photo, Video

About ComfyUI

ComfyUI lets you build generation pipelines by connecting nodes on a canvas, exposing every model, parameter and processing step. It runs open models such as FLUX, Wan, Qwen Image and LTX, plus paid partner models like Nano Banana and GPT Image through Partner Nodes. Over 5,000 community extensions add nodes and integrations with tools like Photoshop, Blender, Nuke and Houdini. Comfy Desktop runs offline on Windows, macOS and Linux; Comfy Cloud provides hosted GPUs and pre-installed models; a developer platform deploys workflows as APIs; and Comfy Agent can build workflows from a description.

Good for

Technical artists, VFX and production teams who want precise, reproducible control over every step of image and video generation, and who want to adopt new open models quickly.

Watch out for

The node graph has a real learning curve compared with prompt-box tools. Local use needs a dedicated GPU, and model files take significant disk space. Partner models and Cloud usage cost credits even though the core app is free.

Pricing notes

The desktop app is free and open source under GPL-3.0. As of October 2026 the pricing page lists Comfy Cloud Standard at $20/month ($16 billed yearly), Creator $35 ($28), Pro $100 ($80) and Team $700 ($630), with new users getting 5 free cloud runs. Cloud charges only for active GPU time.