Open WebUI + Ollama + Qwen3 with Docker Compose

Local stack to run Ollama (custom image with Qwen3 preloaded) and Open WebUI using Docker Compose.

This setup provides a ready-to-use local AI chat environment without downloading models at runtime.

---

Quick architecture

• ollama-qwen3 → Ollama runtime with model qwen3:4b preloaded
  API available at http://localhost:11434
• open-webui → Web-based chat interface available at http://localhost:3000

---

Requirements

• Docker (Docker Desktop on macOS)
• Docker Compose v2 (docker compose)
• Free ports:
  • 11434 (Ollama + Qwen3 API)
  • 3000 (Open WebUI)

---

Installation

Install Docker

Fedora / RHEL / CentOS

sudo dnf install -y docker
sudo systemctl enable --now docker

Ubuntu / Debian

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

macOS

brew install --cask docker

Verify installation:

docker --version
docker compose version

---

Configuration

Create a .env file in the same directory as docker-compose.yaml:

WEBUI_SECRET_KEY=change-me-to-a-long-random-secret
ENABLE_SIGNUP=true

Generate a secure secret:

openssl rand -hex 32

---

Execution

From the project directory:

docker compose up -d

Access:

• Open WebUI → http://localhost:3000
• Ollama API → http://localhost:11434

On first access, Open WebUI will ask you to create the first user (admin). This is expected behavior.

---

Testing and validation

Verify Ollama

curl http://localhost:11434

Expected response:

Ollama is running

List models

curl http://localhost:11434/api/tags

Validate Open WebUI

Open in the browser:

http://localhost:3000

Create a user, log in, and test a chat with the model.

Logs

docker logs ollama-qwen3
docker logs open-webui

---

Stop and restart

Stop

docker compose stop

Start again

docker compose start

---

Teardown

Bring down containers (keeps data)

docker compose down

Full cleanup (⚠️ removes volumes)

docker compose down -v

This removes:

• Downloaded models (Ollama)
• Users and conversations (Open WebUI)

---

Image source

ghcr.io/r3xakead0/qwen3-container/ollama-qwen3:latest

---

Security recommendations

• Change WEBUI_SECRET_KEY to a long, random secret
• After creating the first user, disable signups:

  ENABLE_SIGNUP=false

  then recreate containers:

  docker compose up -d --force-recreate

---

Expected final state

• Ollama responding on localhost:11434
• Open WebUI accessible on localhost:3000
• Model available and functional

---

Third-party components

This project uses the following open-source components:

• Ollama – MIT License
  https://github.com/ollama/ollama

• Open WebUI – MIT License
  https://github.com/open-webui/open-webui

• Buildah – Apache License 2.0
  https://github.com/containers/buildah

• Docker / Docker Compose – Apache License 2.0
  https://www.docker.com/

---

License

This project is licensed under the MIT License. See the LICENSE file for details.