ProductFeaturesSelf-Hosted AIHow it worksPricingDocsBlog
Sign inDownload
Self-Hosted AI / Chat interface

Open WebUI

A private, self-hosted ChatGPT - talking to your own models, not OpenAI's.

Download DeployOSSee all AI templates

What is Open WebUI?

Open WebUI is a polished, actively-maintained chat interface built for exactly this: a familiar ChatGPT-style experience - conversations, model switching, file uploads, saved chats - sitting in front of a model you run yourself.

It talks to Ollama (or any OpenAI-compatible API) as its backend, so the actual reasoning happens on your infrastructure. Nothing about a conversation gets sent to OpenAI, Anthropic, or anyone else unless you explicitly configure it to.

It supports multiple users, so it works as a small team's shared AI tool as easily as a personal one.

How DeployOS deploys it

  1. 1DeployOS deploys the official ghcr.io/open-webui/open-webui image with a persistent volume at /app/backend/data for chat history, uploaded files, and settings.
  2. 2The OLLAMA_BASE_URL environment variable is pre-wired to http://ollama:11434 - Docker's internal service discovery - so if you deploy the Ollama template on the same server first, Open WebUI finds it automatically with zero manual configuration.
  3. 3It's published through your server's reverse proxy (Traefik or Caddy), so it gets a real HTTPS domain and certificate the same way any other DeployOS app does.
Imageghcr.io/open-webui/open-webui:main
Port3000
Persistent volume/app/backend/data
OLLAMA_BASE_URLhttp://ollama:11434

Pairs well with

OllamaQdrantFlowise

Common questions

Do I need Ollama running first?

Yes, or another OpenAI-compatible backend (LocalAI works too). Open WebUI is the interface, not the model runtime - deploy Ollama on the same server (or use the AI Workbench stack, which bundles both) and the connection is automatic.

Can more than one person use it?

Yes - Open WebUI has its own user accounts and admin controls, separate from your DeployOS login, so you can share one instance across a small team.

How is this different from just using ChatGPT?

The interface will feel familiar, but the model runs on your own server, so there's no per-message cost, no vendor seeing your conversations, and no dependency on a third party staying online or keeping your access.

Deploy Open WebUI on your own server in one click.

Free download, no account needed. AI templates are on paid plans.

Download DeployOS