Open WebUI
Self-hosted chat platform that runs offline and connects to Ollama or any OpenAI-compatible API.
Connect Ollama or OpenAI-compatible APIs; install with pip, Docker or a desktop app.
Open WebUI is a self-hosted chat interface for local and cloud language models. It suits people and teams who want to run their own front end instead of paying for a hosted chatbot subscription.
What it is
The README describes a platform built to run entirely offline that connects to local models through Ollama or to OpenAI-compatible APIs. It installs with uv or pip on Python 3.12, with a Docker image, or as a desktop app for macOS, Windows and Linux.
Who it's for
It suits technical users and small teams who can run Docker or Python and supply their own models or API keys. It does not suit organisations that must remove the Open WebUI branding for more than 50 users without a separate agreement.
Key features
Offline-capable platform
Built to run entirely offline with local models.
Ollama and API backends
Connect Ollama or any OpenAI-compatible API.
Multiple Docker images
Main, slim, CUDA and Ollama-bundled image tags.
Desktop app
Native app for macOS, Windows and Linux.
Python install
Run with uvx or pip on Python 3.12.
Enterprise add-ons
Terminals orchestration is listed as an Enterprise product.
Up and running in 3 steps
Our honest verdict
What's great
- One Docker command starts a working chat interface on localhost.
- Works with local Ollama models or any OpenAI-compatible endpoint.
- Image tags cover CUDA GPUs and a slim build that uses external services.
- Can run fully offline with Hugging Face downloads disabled.
Where it falls short
- The licence is not OSI-approved: it forbids removing Open WebUI branding beyond 50 users in 30 days without permission or an enterprise licence.
- You supply the models or API keys; local models need enough RAM or an Nvidia GPU.
- Some orchestration features are listed under the Enterprise ecosystem, not the community code.
- Contributions fall under a contributor licence agreement.
Pricing & license
The community code is free to run under its own licence. The README lists Enterprise offerings but shows no price on the pages read; otherwise the only cost is the server and any model API usage.
Open remote AI-training roles
870 open remote AI-training roles, updated November 14, 2026; most pay $20–$170 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
Is Open WebUI open source?
The code is public, but its licence adds a branding clause, so it is not OSI-approved; read the LICENSE file before commercial use.
Do I need a GPU?
No for API models; local models run faster on an Nvidia GPU with the CUDA or Ollama image tags.
Can I run it offline?
The README says it is built to run entirely offline; set HF_HUB_OFFLINE=1 to stop Hugging Face downloads.
