OpenWebUI is a web interface for interacting with local LLMs such as Llama, Qwen, DeepSeek, Phi, and others.
It connects to Ollama through a simple API endpoint and provides chat, RAG, knowledge bases, file uploads, and model management.

• Windows 11
• WSL2 with Ubuntu
• Docker installed inside WSL2
• Ollama running in Docker
• Both containers on the same Docker network
• Basic familiarity with Docker Compose

If you do not already have a shared network for your containers, create one.

docker network create mediaserver

If you already have one, skip this step.

Below is a generic working example of an OpenWebUI service.

services:
  openwebui:
    container_name: openwebui.<HOST_NAME>
    hostname: openwebui.<HOST_NAME>.lan
    image: ghcr.io/open-webui/open-webui:latest
    environment:
      TZ: <TZ>
      WEBUI_AUTH: "false"
      OLLAMA_API_BASE_URL: "http://ollama.<HOST_NAME>:11434"
    networks:
      - mediaserver
    ports:
      - "<OPENWEBUI_PORT>:8080"
    volumes:
      - <DOCKERDIR>/openwebui:/app/backend/data
    restart: always
    security_opt:
      - no-new-privileges:true
    labels:
      - "com.centurylinklabs.watchtower.enable=true"
      - "homepage.group=AI"
      - "homepage.name=OpenWebUI"
      - "homepage.icon=openwebui.png"
      - "homepage.href=https://openwebui.<DOMAIN_NAME>/"
      - "homepage.description=Local AI chat interface"

Start OpenWebUI.

docker compose up -d

Check that the container is running.

docker ps

Open your browser and go to:

http://<HOST_NAME>:<OPENWEBUI_PORT>

If WEBUI_AUTH is disabled, you will be logged in automatically.

OpenWebUI connects to Ollama using the API endpoint defined in the environment variable.

Example endpoint
http://ollama.<HOST_NAME>:11434

Test connectivity from inside the OpenWebUI container.

docker exec -it openwebui.<HOST_NAME> curl http://ollama.<HOST_NAME>:11434/api/tags

If you see your model list, the connection is working.

Inside the OpenWebUI interface:

1. Open Settings
2. Select Models
3. Ensure Ollama is listed as a provider
4. Select your default model (example: llama3.1:8b)
5. Test a prompt to confirm GPU inference through Ollama

OpenWebUI supports RAG through its built‑in knowledge base system.

To add a knowledge source:

1. Go to Knowledge
2. Click Add Source
3. Choose URL, File, or Text
4. Allow OpenWebUI to embed the content
5. Ask questions referencing the new knowledge

This requires Ollama to support embeddings.

Backup the OpenWebUI data directory.

<DOCKERDIR>/openwebui

This contains chat history, knowledge bases, and settings.

Copy the directory into a new OpenWebUI container volume and restart the stack.

Verify the endpoint.

curl http://ollama.<HOST_NAME>:11434/api/tags

Ensure both containers are on the same Docker network.

Check the environment variable.

OLLAMA_API_BASE_URL="http://ollama.<HOST_NAME>:11434"

Ensure Ollama has embeddings enabled and the model supports them.

Check logs.

docker logs openwebui.<HOST_NAME>

• OpenWebUI requires Ollama for model execution
• Both containers must share a Docker network
• Disable WEBUI_AUTH only on trusted networks
• Knowledge bases require embedding‑capable models
• GPU inference happens inside Ollama, not OpenWebUI