vllm/docs/deployment/frameworks/open-webui.md

Open WebUI

Open WebUI is an extensible, feature-rich, and user-friendly self-hosted AI platform designed to operate entirely offline. It supports various LLM runners like Ollama and OpenAI-compatible APIs, with built-in RAG capabilities, making it a powerful AI deployment solution.

To get started with Open WebUI using vLLM, follow these steps:

1. Install the Docker.

2. Start the vLLM server with a supported chat completion model:

   vllm serve Qwen/Qwen3-0.6B-Chat
   

   !!! note When starting the vLLM server, be sure to specify the host and port using the --host and --port flags. For example:

    ```console
    vllm serve <model> --host 0.0.0.0 --port 8000
    ```
   

3. Start the Open WebUI Docker container:

   docker run -d \
       --name open-webui \
       -p 3000:8080 \
       -v open-webui:/app/backend/data \
       -e OPENAI_API_BASE_URL=http://0.0.0.0:8000/v1 \
       --restart always \
       ghcr.io/open-webui/open-webui:main
   

4. Open it in the browser: http://open-webui-host:3000/

   At the top of the page, you should see the model Qwen/Qwen3-0.6B-Chat.