ComfyUI, already set up.

Three steps to a render.

1. Attach a network volume150 GB covers one model family; 500 GB+ if you want them all. Models survive pod restarts.
2. Deploy the templateSet your tokens and the packs you want. A 48 GB card (A6000 / L40S) runs everything; 80–96 GB is faster for LTX.
3. Open ComfyUIPort 8188. Drag a workflow .json onto the canvas — the templates are also in the Workflow menu.

Environment variables

• GMANSKI_PACKS — Comma-separated packs to install on boot, e.g. krea2,captioning
• CIVITAI_TOKEN — Your Civitai API key — needed for community checkpoints and LoRAs
• HF_TOKEN — Your Hugging Face token — needed for gated models
• JUPYTER_TOKEN — Pin your own JupyterLab token (blank = a new random one each boot, printed in the pod log)
• FILEBROWSER_PASSWORD — Pin your own FileBrowser password for user admin (blank = random each boot)

Add packs later, from the Jupyter terminal

gmanski-packs list
gmanski-packs install krea2 captioning detectors
gmanski-packs status

Logging in.

ComfyUI on port 8188 opens straight away. JupyterLab (8888) and FileBrowser (8080) are protected, and their logins are printed once, at the end of the pod’s startup log.

1. Open the pod’s logIn RunPod, open your pod and click Logs. Scroll to the box printed when startup finishes.
2. JupyterLabUse the link on the JupyterLab line — the token is already in it.
3. FileBrowserUsername admin, password from the FileBrowser line.
4. Same logins every timeFill in JUPYTER_TOKEN and FILEBROWSER_PASSWORD when you deploy. Use long values — anyone with the pod URL can reach these ports.

The box at the end of the log

JupyterLab   https://<pod-id>-8888.proxy.runpod.net?token=<token>
FileBrowser  https://<pod-id>-8080.proxy.runpod.net  admin / <password>