Panel Haus Neural Forge
Run AI generation on your own GPU through a local ComfyUI, connected securely to Panel Haus. Power-user mode for unlimited, fully local generation.
ℹ The bridge architecture ComfyUI runs as a local server on your machine (default:
localhost:8188). Panel Haus connects to it and sends generation requests, the same way it talks to Gemini, except the server is your own computer. Because Panel Haus runs on HTTPS and your ComfyUI runs on HTTP, you use a Cloudflare Tunnel to bridge them securely. It is a free, one-command setup.Requirements
| Component | Minimum | Recommended |
|---|---|---|
| GPU | NVIDIA with 6GB VRAM | NVIDIA with 12GB+ VRAM (RTX 3060, 4070, etc.) |
| RAM | 16 GB | 32 GB |
| Storage | 20 GB free | 50 GB+ (models are large) |
| OS | Windows 10/11, Linux, macOS | Windows or Linux (best GPU support) |
△ Watch out AMD and Intel GPUs have limited ComfyUI support. NVIDIA GPUs with CUDA are strongly recommended. Apple Silicon (M1/M2/M3) works but is slower than equivalent NVIDIA hardware.
Setup path
There are two steps to get Neural Forge running:
- Install ComfyUI. Use the 1-click installer to get ComfyUI running with all required models and custom nodes.
- Set up the Cloudflare Tunnel. One command that creates an HTTPS bridge between Panel Haus and your local ComfyUI.