Today Comfy API is live for everyone on a paid Comfy plan.
Turn a workflow you built in ComfyUI into the engine behind an app, tool, or service. Package the workflow with its custom nodes, LoRAs, models, and Python dependencies, then deploy it as an autoscaling API endpoint.
Until now, putting a ComfyUI workflow into production meant rebuilding its environment somewhere else. Teams had to rent GPUs, install every custom node and model again, untangle Python dependencies, and write the scaling logic around it.
Comfy API handles that work without changing the workflow or ComfyUI. Your graph stays yours, the engine stays open source, and your Builds remain portable to hardware you own.
Comfy API runs on the Developer Platform
Your Builds, deployments, usage, spend, and API keys live in one console at platform.comfy.org. You can work in the browser, from the terminal, or with a coding agent.
Turn your workflow into an API
Start with a workflow JSON or export a snapshot from Comfy Desktop. Builder reads the workflow, finds the models and custom nodes it needs, and helps resolve conflicting Python dependencies. You can override any version it picks.
The Build captures the ComfyUI version, custom nodes, models, and Python dependencies your workflow expects. Cut an immutable release from that Build, then deploy the release as a managed endpoint with its own URL.
Because the release is immutable, the environment you test is the environment you deploy. When you need to make a change, update the Build and cut a new release without changing the one that’s live.
“At Silverside, Comfy workflows are where our creative expertise becomes infrastructure. We build production pipelines then serve them through Comfy API so our clients can run the same proven workflows at scale. It lets a small team deliver enterprise-grade creative output.”
—Allie Wrubel, Head of Technology at Silverside AI
Build from the terminal
Using comfy-cli and Comfy Skills, you or your coding agent can package and deploy a local ComfyUI install:
$ comfy build init
✓ Scanned this ComfyUI install - custom nodes, models, pinned deps
$ comfy build push --release
Or skip the setup. Copy the agent prompt on the Comfy API page, paste it into your coding agent, and let it package and deploy the workflow.
The Comfy SDK supports running workflows on Comfy API and Comfy Cloud API today. To use the same SDK with a self-hosted ComfyUI deployment, run the proxy service and set one environment variable.
What you can build
Deploy your ComfyUI workflows as APIs behind customer-facing products, internal tools, and automated creative pipelines. Build the experience your users need without exposing the graph.
Give your team a tool. You’ve tuned a product photography workflow with custom nodes and a fine-tuned LoRA. Put a simple interface in front of it so the rest of your team can run it without opening ComfyUI.
Add a feature to your product. Your app lets customers restyle an image. The app sends each request to a ComfyUI workflow you control, and the deployment scales with traffic.
Automate repeatable creative work. New products land in your catalog every week. A script sends each one through your workflow and saves the outputs, without anyone watching the graph.
From one builder to a production team
The person who builds the workflow often isn’t the only person who needs to run it.
Comfy API is built for that handoff.
Share one versioned environment. On Team and Enterprise plans, teammates can work from the same Build, with the ComfyUI version, models, custom nodes, and Python dependencies pinned together.
Scale with demand. Set the minimum to zero and the deployment uses no GPU time when no workers are running. Keep workers warm when requests can’t wait for a cold start.
Track usage and spend. See workflow runs, GPU usage, and costs from the Developer Platform console.
Bring the workflow you already use. Package public or private custom nodes, model checkpoints, LoRAs, and pinned Python dependencies into the Build.
Govern what teams can use. Enterprise customers can use Managed Builds and governance controls to standardize approved ComfyUI versions, models, custom nodes, and dependencies across teams. Learn more at Comfy for Enterprise.
Why not wire it together yourself?
If you already run ComfyUI on RunPod or Modal and it works, keep it. If you need GPUs for more than ComfyUI, a general-purpose platform may be the right call.
Comfy API earns its place when ComfyUI is the part you keep managing. RunPod and Modal remove the need to manage infrastructure. Comfy API also removes the work of managing ComfyUI environments: custom nodes, models, Python dependencies, and version drift.
It’s built for ComfyUI by the team behind ComfyUI and Comfy Cloud, so the path from a local workflow to a deployed endpoint is part of the product, not an integration you maintain.
Pricing
Comfy API is available on paid Comfy plans, starting with Standard. Usage is billed separately.
GPU time is billed by the second. Storage is prorated hourly.
What this means, and what’s next
ComfyUI gave people control over every step of how an image or video gets made. Every node is visible. Every parameter is adjustable. But getting that workflow in front of a team or a customer still meant rebuilding it somewhere else and hoping it behaved the same way.
Comfy API closes that gap. You build the workflow in ComfyUI. Comfy API packages the environment and runs it behind an endpoint. Your team or product calls it without seeing the graph.
Comfy is building an open standard for visual AI. For that standard to hold up in production, the workflow can’t lose its custom nodes, models, or dependencies on the way there. That’s why Builds are versioned, the engine stays open, and your work remains portable.
A repeatable creative workflow needs the tested environment, not just the graph.