Linux · System Tools
Open WebUI
Open WebUI is a self‑hosted, offline AI platform that lets you run language models locally via a web interface.
- License
- AGPL-3.0-or-later
- Latest version
- 0.0.8 (April 11, 2026)
- Developer
- Timothy J. Baek
- Installs
- 1,851 (Flathub, 30 days)
What is Open WebUI?
Open WebUI started as a small project by Timothy J. Baek in early 2024 and has grown into a stable, feature‑rich stack for running large language models on your own hardware. It is distributed as a Flatpak on Flathub, making installation on most Linux distributions straightforward. The project is actively maintained, with the latest release 0.0.8 released in April 2026.
The tool is aimed at developers, researchers, and power users who need a private, offline AI solution without relying on cloud services. It works well for those who already use local models like Llama or GPT‑Neo.
Compared to Windows‑based Copilot or other cloud‑centric AI tools, Open WebUI feels more like a local server you control. You can tweak model parameters, add plugins, and run it behind a reverse proxy, giving you the same flexibility as a custom Windows application but with the privacy and control of Linux.
Best for: Users who need a fully private, offline AI platform and are comfortable managing local model files.
Key features of Open WebUI
- Self‑hosted and offline by default
- Extensible via plugins
- Web‑based UI accessible from any browser
- Supports multiple model backends (Llama, GPT‑Neo, etc.)
- Configurable prompt templates
- Built‑in model management and versioning
- Docker/Flatpak distribution
- OpenAPI‑compatible endpoints
Pros
- Full privacy – no data sent to the cloud
- No subscription costs once models are downloaded
- Easy to install on most Linux distros
- Active community and frequent updates
- Plugin ecosystem expands functionality
Cons
- Requires local GPU or powerful CPU for large models
- Initial setup of models can be time‑consuming
- Limited official documentation compared to commercial tools
- No native Windows installer
Windows programs Open WebUI replaces
How to install Open WebUI on Linux
flatpak install flathub com.openwebui.open-webuiTips for Windows users
If you’re moving from Microsoft Copilot, remember that Open WebUI does not use a cloud API; you’ll need to download the model weights yourself and place them in the data folder (e.g., ~/.local/share/open-webui/models). The UI is web‑based, so you open it in a browser at http://localhost:8080. Keyboard shortcuts differ: use Ctrl+Enter to submit a prompt, and the sidebar for settings. Workflow changes include manual model selection and no auto‑updates of the model.
File formats are plain text for prompts; no proprietary formats are involved. The interface is similar to Copilot’s chat window, but you’ll manage plugins via the Settings page rather than a marketplace.
Overall, the learning curve is moderate; you’ll spend a few hours setting up the environment but then the experience is stable and fast.
Recent releases
- 0.0.8 —
- 0.0.6 —
Open WebUI: frequently asked questions
Do I need a GPU to run Open WebUI?
A GPU speeds up inference significantly, especially for large models. However, a CPU can run smaller models or lower‑precision variants, though it will be slower.
Can I use Open WebUI with my own custom model?
Yes. Place the model files in the data/models folder and configure the model name in the settings. The UI will detect and load it.
Is there a Windows version?
No. Open WebUI is designed for Linux and macOS; Windows users would need to use a Linux virtual machine or WSL to run it.
How do I update the software?
If installed via Flatpak, run flatpak update open-webui. For manual installs, pull the latest release from GitHub and restart the service.
Updated