How to Connect ComfyUI to WordPress

If you already run ComfyUI on a GPU, you have everything you need to generate images and video — except a clean way to put it in front of people on your website. StudioUI is that front end. On the Pro tier you point the plugin at your own ComfyUI backend and your WordPress site renders generations on your hardware, privately, with no per-generation fees. This guide is the practical wiring: prerequisites, the exact settings to fill in, and the things that usually go wrong.

Why connect ComfyUI instead of using hosted compute

Hosted compute (the Free tier) is the zero-setup path — you pay per generation in credits and never touch a GPU. You’d connect your own ComfyUI when you already have hardware and want: no usage fees, full privacy (nothing leaves your stack), and control over exactly which models and workflows run. That’s the Pro self-host case.

Prerequisites

StudioUI Pro ($149/yr) — self-hosting is a Pro capability.
A running ComfyUI instance with its API enabled and the models you intend to use already installed (for example a FLUX checkpoint and any required custom nodes for your workflow).
Network reachability — your WordPress server must be able to reach the ComfyUI machine over HTTP. That means a LAN address, a VPN, or a tunnel/reverse proxy if ComfyUI lives behind a home connection.

Step 1 — Confirm ComfyUI is reachable

ComfyUI listens on port 8188 by default. Start it bound to an address your WordPress host can reach (not just 127.0.0.1 if WordPress is on another machine) — for example launching with --listen so it accepts non-local connections. Verify by loading http://YOUR-COMFY-HOST:8188 from a browser on the same network as your WordPress server. If the ComfyUI graph loads, the endpoint is reachable.

Step 2 — Enter the endpoint in StudioUI

In StudioUI’s settings, switch the image (and/or video) backend to self-hosted ComfyUI and enter your endpoint URL, e.g. http://YOUR-COMFY-HOST:8188. Save. StudioUI will route generation jobs to that instance instead of hosted compute.

Step 3 — Match models and workflows

StudioUI sends generation requests that expect certain models/nodes to exist on your ComfyUI. Make sure the checkpoints and custom nodes your chosen workflow needs are installed in ComfyUI — a missing FLUX checkpoint or a missing custom node is the single most common reason a self-hosted job fails. Keep the model names consistent with what StudioUI is configured to request.

Step 4 — Test the real generation path

Open the studio page as a normal user and generate one image. Watch the ComfyUI console at the same time — you should see the job arrive and execute. If the image lands back in your WordPress media library, the connection is working end to end. Always test the actual generate path, not just that the settings saved.

Troubleshooting the usual failures

Job never reaches ComfyUI — reachability/firewall. Re-check that the WordPress host (not your laptop) can hit the endpoint; a localhost-only bind or a closed port is the usual culprit.
Job arrives but errors — a missing model, checkpoint or custom node on the ComfyUI side. Read the ComfyUI console error; it names what’s missing.
Works on LAN, fails in production — your ComfyUI is on a private network the live server can’t see. Use a tunnel, reverse proxy or VPN so the production host has a route.
Times out on heavy jobs — large video/animation renders can exceed default timeouts; give the backend headroom and start with lighter workflows while you validate the pipe.

FAQ

Do I need Pro to connect ComfyUI?

Yes. Self-hosting via your own ComfyUI backend is a Pro feature ($149/yr). The Free tier runs on hosted compute and needs no GPU.

Can I run hosted and self-hosted at the same time?

Pro lets you point StudioUI at your own backends and bring your own keys with failover. Most people pick one path per modality; the flexibility to mix is there.

What port and protocol does ComfyUI use?

By default ComfyUI serves on port 8188 over HTTP. Bind it to an address your WordPress server can reach and use that as the endpoint URL.

Is anything sent to StudioUI’s servers when I self-host?

When generation runs on your own ComfyUI, the generation stays on your infrastructure — that privacy is the whole reason to self-host. Your Pro licence is validated separately.

Which models can I use?

Whatever your ComfyUI has installed and your workflow supports — FLUX checkpoints, SDXL-class models and more. You manage models in ComfyUI; StudioUI is the interface in front of them.

Put a clean studio in front of your GPU

Connect your ComfyUI to StudioUI Pro and your WordPress site becomes the front end for your own hardware — private, fee-free, and brandable for your members.

Get Pro — $149/yr · Read the setup guide · Install free first