Open WebUI

AI & Audio

A private ChatGPT-style interface for models running on your own GPU.

Open WebUI is a self-hosted chat interface for large language models. It works with Ollama, which runs the models on your own hardware, and with any OpenAI-compatible API. You get conversation history, multiple users, document upload with retrieval, web search, voice input and model management in a browser, and your prompts never have to leave your network. On a GPU such as an RTX 3090, a 20B model answers at over 100 tokens per second. It is a self-hosted alternative to ChatGPT-style subscriptions, as long as you are fine with the quality of open models.

License

Open WebUI License (BSD-3-Clause based, with a branding clause)

GitHub stars

156,262

Official website

https://openwebui.com

Source code

https://github.com/open-webui/open-webui

Compatibility

Architectures

amd64, arm64

Database

SQLite

Resource footprint

4096 MB RAM · 4 CPU

Measured in our LXC (6 vCPU, 6 GB RAM, RTX 3090) with Ollama 0.34.2: the Open WebUI container uses about 0.5 to 0.7 GB of RAM, and the container lives in an LXC that sat at 0.7 GB idle and about 2.9 GB with a 20B model loaded, because the model weights go to the GPU. gpt-oss:20b (13 GB download) took about 16.5 GB of VRAM at the default 32k context and generated around 122 to 125 tokens per second; the first load took about 47 seconds. The Docker image is about 6.5 GB. Without a GPU, expect much slower answers and stick to small models.

Proof of Work Gallery

Chat
New chat
Sign-in
Resource usage

Deployment

# Ollama runs on the host (or in the same LXC) and listens on 127.0.0.1:11434.
# Install it first with the official instructions at https://ollama.com/download
# For an NVIDIA GPU you also need the NVIDIA driver on the host.
# host networking lets the container reach Ollama on 127.0.0.1; the UI is then on port 8080.
services:
  open-webui:
    image: ghcr.io/open-webui/open-webui:main
    container_name: open-webui
    restart: unless-stopped
    network_mode: host
    environment:
      - OLLAMA_BASE_URL=http://127.0.0.1:11434
    volumes:
      - open-webui:/app/backend/data

volumes:
  open-webui:

Alternative to

ChatGPT, Claude.ai, Google Gemini