Why#

RunPod gives us uncensored GPU compute at ~$0.69-1.10/hr — 10-20x cheaper than per-image APIs for image gen at volume, and enables heavy LLM inference (Llama 70B), video gen, and any HuggingFace model.

Steps#

1. Sign up at https://runpod.io (free account)
2. Add $10 credits (no free tier — $10 lasts a long time)
3. Deploy a Serverless endpoint using the ComfyUI template:
   • Go to Serverless → Create Endpoint
   • Search "ComfyUI" template
   • Select RTX 3090 (cheapest) or RTX 4090
   • minWorkers: 0, maxWorkers: 2
4. Generate an API key from Account Settings → API Keys
5. Copy the endpoint ID and API key — paste them here

I'll add the keys to .env and create the bridge script.

Spec#

Full design: projects/image-gen/runpod-setup-spec.md

🕵️ Review requested — Maxime flagged this task for Robbie to review on 2026-05-10T15:35:39.636Z.

Please tell me what should be the name for this parameter I will add in .env file. Thank you

🕵️ Review requested — Maxime flagged this task for Robbie to review on 2026-05-11T00:13:53.568Z.

I'm trying to deploy comfyui, but on this page: https://console.runpod.io/hub/runpod-workers/worker-comfyui?from=endpoint-creation

After I click "Deploy 5.8.5", then the popup asks for: Comfy.org API Key Enter your Comfy.org API key to use API Nodes

Did I do something wrong? Is it normal?

🕵️ Review requested — Maxime flagged this task for Robbie to review on 2026-05-11T00:34:31.122Z.

I added the api key to our .env file, parameter name: RUNPOD_COMFYUI_API_KEY

Please review and proceed with the rest if everything is correct