Model APIs
Use Comfy Router to call hosted partner models with one Comfy API key.Generate an image
Make your first image with Flux in Python, TypeScript, or cURL.
Browse available models
Find a model, inspect its inputs, and copy its model ID.
Use a model's schema
See the request and response shape for a Router model.
Call Router over HTTP
Look up the shared Router endpoints and headers.
Comfy API
Package your own ComfyUI workflow, deploy it to managed GPU infrastructure, and run it through the Comfy API.Deploy a workflow
Create a Build, release it, and start a managed deployment.
Choose compute
Find the available regions and GPU classes before you deploy.
Run a deployed workflow
Submit an API-format workflow and download its outputs.
Use the SDK
Point the Python or TypeScript SDK at your deployment.
Full Demos
Start from a complete demo app, then adapt the workflow and interface for your own project. Choose Comfy Cloud for the fastest path, or Deploy Your Own when you need your own models or nodes.
Image-to-image web app
Upload an image, run a Comfy workflow, and show the result in a Next.js app.
Sketch to image
Draw or import a guide, then generate an image from the guide and a prompt.
Discord image bot
Add an /imagine command to a Discord server and run the bundled Comfy workflow.
Both paths need a Comfy API key. For Deploy Your Own, download the workflow, use Builds to paste it and configure the serverless endpoint, wait for its URL, then paste that URL and the API key into your Vercel or Render deployment.