Open WebUI Desktop

Your AI, right on your desktop. Open WebUI as a native app. Run models locally or connect to any server. No Docker, no terminal, no setup. Download, launch, chat.

[!WARNING] Early Alpha. Things move fast and stuff might break. Report bugs or come hang out on Discord.

Download

Internet required on first launch. After that, everything works offline. All releases →

How It Works

🖥️ Run locally. The app runs Open WebUI on your machine. You can optionally enable the built-in llama.cpp engine to download and run models offline. Nothing leaves your computer.

☁️ Connect remotely. Point the app at any Open WebUI server. Switch between multiple connections from the sidebar.

Use both at the same time.

Highlights

• ⚡ Spotlight. Hit Shift+Cmd+I (macOS) or Shift+Ctrl+I (Windows/Linux) to summon a floating chat bar over whatever you're doing. Drag to screenshot anything on screen.
• 🎙️ Voice input. System-wide push-to-talk. Press the shortcut from any app to record, and your speech is transcribed and sent to your chat automatically.
• 🧠 Local inference. Optionally run models entirely on your hardware via the built-in llama.cpp engine. Your data never leaves your machine.
• 🎯 One-click setup. Launch and connect to a server in seconds. Local models can be enabled from the settings.
• 🔌 Multiple connections. Juggle servers and switch between them instantly.
• 🔄 Auto-updates. New releases land in the background.
• 📡 Offline-ready. No internet needed after initial setup.
• 💻 Cross-platform. macOS, Windows, and Linux.

System Requirements

[!NOTE] Local models need serious RAM (7B ≈ 8 GB, 13B ≈ 16 GB). Lighter machine? Connect to a remote server instead.

Privacy

No telemetry. No tracking. No phone-home. Your conversations stay on your machine. Period.

Community

• 💬 Discord - Come hang out
• 🐛 Issues - Report bugs or request features
• 🌐 Open WebUI - The main project
• 📖 Docs - Full documentation

Contributing

npm install
npm run dev

See CHANGELOG.md for release history. Licensed under AGPL-3.0.