Self-host Open WebUI

updated Sep 2026deployed by usprices checked · Jul 2026Install verified on Ubuntu 26.04 · Sep 2026 · how we test

We earn commissions when you shop through the links below. Full disclosure →

Feature-rich, self-hosted ChatGPT-style interface for Ollama and any OpenAI-compatible API, with built-in RAG, model management, and multi-user support. The web UI itself is lightweight; pair it with a GPU-backed Ollama backend for local model inference.

Key facts

CategoryAI Chat Interfaces

LicenseOpen WebUI License (modified BSD-3-Clause with a branding clause)

StackPython, Svelte, TypeScript

Min RAM2048 MB

Official imageyes

Measured idle RAM928 MB

Difficulty

Sourcegithub.com/open-webui/open-webui

Websiteopenwebui.com

Our recommendation

Pick Open WebUI when you want the polished, ChatGPT-style front end for a local model stack — it's the most popular self-hosted chat UI, with built-in RAG, model management, and multi-user accounts, and the web tier itself is light (2 GB). It's a UI, not an inference engine: point it at an Ollama or any OpenAI-compatible backend, which is where the real hardware goes. If you'd rather have multi-provider routing and agents in one app, LibreChat is the broader tool.

Follow the Open WebUI deploy guide →

What you need

Any VPS with at least 2048 MB of RAM
A domain you control — most self-hosted setups need HTTPS in front of them
Under an hour if you've used Docker before

Where to host itaffiliate disclosure

Hetzner Cloudrun it on

From $23.59/mo · 2 vCPU / 4 GB / 80 GB · EU + US

Get Hetzner Cloud

Kamaterafree trial

From $4/mo · 1 vCPU / 1 GB / 20 GB · US + EU + Asia

Start free on Kamatera →

DigitalOceanalso works on

From $6/mo · 1 vCPU / 1 GB / 25 GB · US + EU + Asia

Deploy on DigitalOcean →

Paid link — we earn a commission if you shop through it.

Install

Run these commands on your server:

# Open WebUI — official quick-start container
docker run -d -p 3000:8080 \
  --add-host=host.docker.internal:host-gateway \
  -v open-webui:/app/backend/data \
  --name open-webui --restart always \
  ghcr.io/open-webui/open-webui:main  #  →  http://SERVER_IP:3000
# :main is upstream's documented tag and moves with every release; pin a release tag (e.g. :vX.Y.Z) for production

deploy guideHow to Deploy Open WebUI on a VPS deploy guideInstall Open WebUI with Ollama vps picksBest VPS for self-hostingbenchmarked quarterly · re-priced monthly

What you take on

Open WebUI is the friendly face on a self-hosted LLM — but the friendly face carries a licence quirk, and the real cost lives in the backend:

non-negotiableIt isn't plain open source. The Open WebUI License is a modified BSD-3-Clause with a branding-retention clause — you can't remove or alter the "Open WebUI" branding unless you have fewer than 50 users, written permission, or an enterprise licence. Fine for personal and small-team use; read the terms before you white-label it.

non-negotiableThe UI is light; the model isn't. Open WebUI on its own sips RAM, but the Ollama (or other) backend it drives wants lots of memory and is far happier with a GPU — that's the honest cost of local inference, and it's a separate service you also run and size.

non-negotiableModel licences are their own thing. The app being BSD-derived says nothing about the weights you pull — each open model (Llama, Mistral, Qwen) carries its own licence and use restrictions to check before you rely on it commercially.