Self-hosted browser chat for Ollama and OpenAI-compatible models, adding retrieval, plugins and user roles. Free software, about 154K GitHub stars.
Choose Open WebUI if you already run Ollama or a hosted API and want a shared browser interface with roles and retrieval. The vendor reports 407M+ downloads. It is not a model runner itself, so expect to set up a backend first. Single users may prefer a desktop app.
A multi-user, plugin-extensible chat platform that sits in front of Ollama and any OpenAI-compatible API with one install.
Community size, a plugin library and team features give it more lock-in than a desktop runner, but any OpenAI-compatible front end can replace it and the custom licence is a recurring friction point.
Main trade-off
It needs a separate model backend, and frequent 0.x releases mean admins must keep up with changes.
| Plan | Monthly price | What it includes |
|---|---|---|
| Self-hosted | Free | Full app under the Open WebUI License, Ollama and OpenAI-compatible connections, retrieval, roles and groups, plugins, voice and video calls |
| Enterprise | Custom | Sales-led plan with SSO, audit logs, data residency, on-premises or air-gapped deployment and permission to alter branding; no public price |
No charge for the self-hosted software when checked in October 2026. The Enterprise tier is quoted by the vendor sales team and has no published price. Your running costs are your own server and any paid model APIs you connect.
Open WebUI is a free, self-hosted chat interface for Ollama and OpenAI-compatible models, with about 154,000 GitHub stars on 2026-10-06. Its latest release, v0.11.4, shipped on 2026-09-21. It adds document retrieval, user roles, plugins and MCP tool servers, and installs with pip or Docker.
Ollama and OpenAI-compatible connections
Connects to local Ollama models and to any OpenAI-compatible endpoint, so one window can switch between a laptop model and a hosted one.
Retrieval over your documents
Uploads and knowledge collections let a model answer from your own files, with web search available as another source.
Roles, groups and permissions
Administrators set per-user and per-group access to models, tools and knowledge, which is what lets a team share one install.
Plugin system with MCP support
Filters, Actions, Pipes, Tools and Skills extend the app, and MCP, MCPO and OpenAPI tool servers connect outside services.
Custom models and agents
Wrap a base model with instructions, tools and knowledge, and share the preset with specific users or groups.
Voice and video calls
Hands-free calls support several speech-to-text providers, including local Whisper, OpenAI, Deepgram and Azure.
Most users expect these and they are not currently supported.
| Tool | Price from | Free plan | Runs on | Rating |
|---|---|---|---|---|
| -- | Yes | Web | -- | |
| $20/mo | Yes | Web · Mac · Windows | 4.0 | |
| $20/mo | Yes | Mac · Windows | -- | |
| Free | Yes | Mac · Windows | -- |
Open WebUI License (BSD-style with a branding clause); downloaded models carry their own licences
The self-hosted software is free as of October 2026. You pay for your own server and for any hosted model APIs you connect. The vendor also sells a sales-led Enterprise tier, with SSO and air-gapped options, but does not publish a price.
Only in limited cases. The Open WebUI License bars altering or removing the branding once a deployment has more than 50 end users in any rolling 30 days, unless you have written permission or an enterprise licence. Smaller deployments are exempt from that clause.
LM Studio, Jan and GPT4All bundle their own chat screen in a desktop app, which suits one person on one computer. Ollama is the engine many people put behind Open WebUI. If you do not want to host anything, a hosted assistant such as ChatGPT or Claude avoids the setup.
No. It connects to Ollama or to any OpenAI-compatible API, so a model backend has to be running somewhere, whether that is a local server on your own network or a hosted endpoint you pay for.
Run pip install open-webui or use the Docker image from the GitHub README, then open the page in your browser and create the first account, which becomes the administrator. Add an Ollama address or an API key under the connection settings and pick a model to chat.
Open WebUI is a self-hosted web interface for talking to language models, built by Timothy Jaeryang Baek and published at github.com/open-webui/open-webui. On 2026-10-06 the repository showed about 154,000 stars and 22,500 forks, and the vendor's site reports 407M+ downloads and 507K+ community members. You install it on your own machine or server, open it in a browser and point it at the models you want to use.
It does not run models itself. Instead it connects to Ollama and to any OpenAI-compatible API, and the README names LM Studio, GroqCloud, Mistral, OpenRouter and vLLM as examples. That makes it a front end rather than a runner: the typical local stack pairs Ollama as the engine with Open WebUI as the chat window. If you want a runner that bundles its own chat screen, LM Studio, Jan and GPT4All each ship a desktop app, and Open WebUI can also talk to hosted services such as OpenRouter or models from OpenAI and Anthropic.
What sets it apart from a plain chat box is the platform around the conversation. Administrators define roles, groups and permissions, so a team can share one deployment. Models can be wrapped with custom instructions, tools and knowledge to make specialised agents. Retrieval over uploaded documents, web search, voice and video calls, notes, channels, a calendar and scheduled prompts are all listed in the README. Plugins come as Filters, Actions, Pipes, Tools and Skills, and external services connect through MCP, MCPO and OpenAPI tool servers. An optional Open Terminal gives an agent a filesystem and shell for multi-step work.
Setup is a single command: pip install open-webui, or a Docker image, with Kubernetes manifests and Helm available for larger installs. The software is free to run. The licence is its own BSD-style Open WebUI License, not an OSI-approved one, and it forbids removing or replacing the Open WebUI branding once a deployment passes 50 end users in 30 days unless you hold written permission or an enterprise licence. An Enterprise plan with SSO, audit logs and air-gapped deployment is sold through the vendor's sales team, with no public price. The latest tagged release on 2026-10-06 was v0.11.4, published 2026-09-21.
Product facts on this page (pricing, plans, integrations, compliance) are read from Open WebUI's own pages and documentation. The verdict is Hokai's reading of them, not the vendor's claim.