How to Deploy Ollama, Open WebUI, and AUTOMATIC1111 Stable Diffusion via Docker on a VPS (CPU‑Only)

By Pedro Martins, April 20, 2025

Running your own private AI stack in the cloud has never been easier—even if you don’t have a GPU. In this post, I’ll walk you through deploying:

• Ollama for LLM inference (e.g. LLaMA 3)
• Open WebUI as a sleek browser-based chat interface
• AUTOMATIC1111 Stable Diffusion WebUI for text-to-image generation

All on an Ubuntu VPS (22.04+) with Docker and Docker Compose, on CPU only.

Prerequisites

Before we get started, make sure you have:

• An Ubuntu 22.04 (or newer) VPS
• Docker and Docker Compose installed
• Ports 11434, 3000, and 7860 open in your firewall
• root access or a user with sudo privileges

Step 1: Install Docker & Docker Compose

sudo apt update && sudo apt install -y docker.io curl git

# If docker-compose plugin isn’t available, install manually:
sudo curl -L "https://github.com/docker/compose/releases/latest/download/docker-compose-$(uname -s)-$(uname -m)" \
  -o /usr/local/bin/docker-compose
sudo chmod +x /usr/local/bin/docker-compose

Verify both are installed:

docker --version
docker-compose --version

Step 2: Deploy Ollama + Open WebUI

1. Create a project directory

   mkdir ~/ollama-webui && cd ~/ollama-webui
   

2. Create a docker-compose.yml

   nano docker-compose.yml
   

3. Paste the following:

   version: "3.8"
   
   services:
     ollama:
       image: ollama/ollama
       container_name: ollama
       ports:
         - "11434:11434"
       networks:
         - ollama-net
       restart: always
   
     open-webui:
       image: ghcr.io/open-webui/open-webui:main
       container_name: open-webui
       ports:
         - "3000:8080"
       environment:
         - OLLAMA_API_BASE_URL=http://ollama:11434
       depends_on:
         - ollama
       networks:
         - ollama-net
       restart: always
   
   networks:
     ollama-net:
       driver: bridge
   

4. Launch the stack

   docker-compose up -d
   

5. Access the interface
   Open your browser to http://<YOUR_VPS_IP>:3000

Step 3: Download the LLaMA 3 Model

With the containers running, pull the 8‑billion‑parameter model:

docker exec -it ollama ollama pull llama3:8b

Ollama will cache it locally for inference.

Step 4: Install AUTOMATIC1111 Stable Diffusion WebUI (CPU Mode)

1. Create a new directory

   mkdir ~/sd && cd ~/sd
   

2. Add your docker-compose.yml

   nano docker-compose.yml
   

3. Use this configuration:

   version: '3.8'
   
   services:
     sd-webui:
       image: goolashe/automatic1111-sd-webui
       container_name: sd-webui
       ports:
         - "7860:7860"
       volumes:
         - ./outputs:/stable-diffusion-webui/outputs
       environment:
         - CLI_ARGS=--skip-torch-cuda-test --no-half --use-cpu all --precision full --lowvram
       restart: always
   

4. Start it up

   docker-compose up -d
   

5. Browse to http://<YOUR_VPS_IP>:7860

Step 5: Test a Prompt

In the Stable Diffusion UI, try a lightweight prompt for faster CPU performance:

a futuristic city at night, cyberpunk style, neon lights, rain, 512x512

Tip: Stick to 512×512 or smaller to keep generation times reasonable.

Step 6: Access Your Images

All generated images land in ~/sd/outputs. You can:

• Download via SFTP/SCP
• Serve them with a lightweight Nginx gallery

Conclusion

Congratulations! 🎉 You now have a self‑hosted AI playground:

• Chat with LLaMA 3 via Open WebUI
• Generate images with Stable Diffusion
• Secure, private, and fully under your control

Feel free to share this post or adapt the setup for your own projects. Happy self‑hosting!