ComfyUI

Node-based Stable Diffusion / Flux UI on port 8188, plus SSH.

DetailValue
Imageghcr.io/fairgpu/comfyui:latest
Modeinteractive
Ports22/ssh, 8188/http
Needs8 GB VRAM, 30 GB disk
Compatible machines online1
Used0 times

ComfyUI

Opens on port 8188 once the container has pulled the image and started. Models go in /workspace/ComfyUI/models (download them with wget/huggingface-cli over SSH; the image ships no checkpoints).

  • VRAM: 8 GB for SD 1.5/SDXL, 16 GB+ for Flux dev, 24 GB for comfortable Flux fp16.
  • Extra launch flags via the COMFYUI_ARGS env (e.g. --lowvram).

Before you stop

Everything you create lives inside the container on the host's disk and is removed when the rental ends. Push results out before stopping (scp, rsync, git, HF Hub, S3...). From inside the container you can run fairgpu extend 60 to add time, fairgpu mark "epoch 3 done" to annotate the timeline and fairgpu stop when you are done.

Step by step

1. Open ComfyUI

Wait for the ComfyUI port to turn green on the rental page (first start pulls the image; cached hosts are ready in seconds), then click it. The UI is reachable only through your rental's relay port.

2. Add models

Models go under /workspace/ComfyUI/models (checkpoints, loras, vae, controlnet, clip ...). Over the terminal:

cd /workspace/ComfyUI/models/checkpoints
wget -O sd_xl_base_1.0.safetensors "https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/resolve/main/sd_xl_base_1.0.safetensors"
huggingface-cli download black-forest-labs/FLUX.1-dev --include "flux1-dev.safetensors" --local-dir /workspace/ComfyUI/models/unet   # gated: needs HF_TOKEN

Press R in ComfyUI to refresh the model lists. Keep a persistent workspace or a checkpoint so the downloads are not repeated next time.

3. Custom nodes

cd /workspace/ComfyUI/custom_nodes && git clone https://github.com/ltdrdata/ComfyUI-Manager.git

Restart the app: Commands → pkill -f main.py (the template restarts it) or stop/start the rental.

4. Tunnel instead of the public relay port

ssh -N -L 8188:localhost:8188 -p <port> root@<relay> then open http://localhost:8188 on your machine.

5. VRAM tips

8 GB: SD 1.5/SDXL with --lowvram in COMFYUI_ARGS; 16 GB: Flux dev fp8; 24 GB: Flux fp16 comfortably. Outputs are in /workspace/ComfyUI/output.

Save your results before you stop

  • Everything outside /workspace disappears when the rental stops; /workspace survives on the same host (persistent workspace) and anywhere with a FairGPU checkpoint.
  • Save checkpoint now on the rental page (or fairgpu snapshot save "label" inside the container) archives /workspace to FairGPU cloud storage; Download workspace gets it to your computer.
  • scp -P <port> root@<relay>:/workspace/results ./ copies files out over SSH; rsync and SFTP (WinSCP/FileZilla) work the same way.
  • Stop the rental to stop billing. Set an idle auto-stop if you tend to forget, or buy the $0.99 finish alert (SMS + email).

Machines that can run it

MachineGPUVRAMPriceStatusReliabilityCountryHostCached
WhiteBobNVIDIA GeForce RTX 5060 Ti16 GB$0.10/hravailable now62 %-Aharon Sela-

Pick a machine for ComfyUI