ComfyUI

ComfyUI is an open source, node-based interface for running AI image and video models. Each model, setting and processing step appears as a block on a canvas that you can connect and adjust.

ComfyUI lets you build pipelines by linking nodes together. You can start from a community template or build your own from scratch. Finished workflows can be saved, shared and, according to the site, deployed as a production API. The project is available as a desktop app, and the site also lists a cloud version and an agent that can help build workflows.

• Build pipelines on a node canvas
• Run models on your own hardware
• Start from community templates
• Save and share workflows
• Deploy a workflow as an API

Visual artists, developers and studios who want full control over each model and parameter. It suits people comfortable with a node-based layout, since every step is visible.

The project is open source under the GPL-3.0 licence. Some add-ons are sold separately; the site lists a commercial licence for MiniMax models as one example. Running models locally depends on your own hardware, so a capable graphics card is often needed for larger models.