🕊️ Libre WebUI

Your AI, Your Rules.

The privacy-first, open-source chat interface for local and cloud AI. No telemetry. No tracking. No compromises.

Why Libre WebUI?

Most AI chat tools either lock you into a cloud, harvest your data, or pull the rug on their open-source license (looking at you, BSD-3 + CLA combos 👀). We think you deserve better.

Libre WebUI is a self-hosted AI chat interface that connects to Ollama for fully local AI, or to OpenAI, Anthropic, Google, and 10+ cloud providers — all from one clean, fast UI. Your conversations never leave your machine unless you choose to send them.

Built by Kroonen AI and the open-source community. Apache 2.0 forever — we put it in our charter.

✨ Features at a Glance

| | Feature | What it does | | --- | ----------------------- | ---------------------------------------------------------------- | | 💬 | Streaming Chat | Real-time responses with dark/light themes and mobile support | | 🔌 | Plugin System | Connect any OpenAI-compatible API via simple JSON config | | 📄 | Document Chat (RAG) | Upload PDFs and chat with your documents | | 🎭 | Personas | Custom AI personalities with persistent memory | | 🎨 | Artifacts | Live HTML, SVG, and code preview right in chat | | 🖼️ | Image Generation | ComfyUI + Flux, DALL·E, and more | | 🔊 | Text-to-Speech | Qwen3-TTS, Kyutai, OpenAI voices — local or cloud | | 🤗 | HuggingFace Hub | Browse and use 1M+ models for chat, TTS, images, embeddings, STT | | 🔐 | SSO & Auth | GitHub, HuggingFace OAuth/OIDC, role-based access | | 🌍 | 25+ Languages | Full i18n — Arabic to Vietnamese | | 🖥️ | Desktop App | Native app for macOS, Windows, Linux (Electron) | | 🤖 | AI Agent Support | OpenClaw integration — persistent agents with memory and tools | | 🏢 | Enterprise Ready | GDPR, HIPAA, SOC 2 compatible • AES-256-GCM encryption |

🚀 Quick Start

One command — that's it

npx libre-webui

Opens at http://localhost:8080. Add Ollama for local AI, or plug in your API keys.

🐳 Docker

# With bundled Ollama
docker-compose up -d

# With your existing Ollama
docker-compose -f docker-compose.external-ollama.yml up -d

# GPU support (NVIDIA)
docker-compose -f docker-compose.gpu.yml up -d

🍺 Homebrew (macOS)

brew tap libre-webui/tap && brew install libre-webui
libre-webui

# Or install the desktop app
brew install --cask libre-webui

☸️ Kubernetes

helm install libre-webui oci://ghcr.io/libre-webui/charts/libre-webui

🖥️ Desktop App

Download from GitHub Releases — available for macOS, Windows, and Linux.

🛠️ From Source

git clone https://github.com/libre-webui/libre-webui
cd libre-webui
cp backend/.env.example backend/.env
npm install && npm run dev

Edit backend/.env:

# Local AI (default — just install Ollama)
OLLAMA_BASE_URL=http://localhost:11434

# Cloud providers (optional — add what you need)
OPENAI_API_KEY=sk-...
ANTHROPIC_API_KEY=sk-ant-...
HUGGINGFACE_API_KEY=hf_...

Dev builds are auto-generated from the dev branch. Not for production.

docker-compose -f docker-compose.dev.yml up -d

Uses separate data volumes (libre_webui_dev_data) so your stable install stays safe.

🏗️ Architecture

┌─────────────────────────────────────────────┐
│              Libre WebUI                    │
├──────────────────┬──────────────────────────┤
│   React + TS     │   Express + SQLite       │
│   Frontend       │   Backend                │
│   (Vite)         │   (AES-256 encryption)   │
├──────────────────┴──────────────────────────┤
│              Plugin Layer                   │
│   Ollama │ OpenAI │ Anthropic │ Google │ …  │
├─────────────────────────────────────────────┤
│   Electron (Desktop) │ Docker │ Kubernetes  │
└─────────────────────────────────────────────┘

• Frontend: React 18 + TypeScript, Vite, responsive with keyboard shortcuts
• Backend: Express 5, SQLite with AES-256-GCM encryption, WebSocket streaming
• Plugins: JSON config files — add any provider without touching code
• Desktop: Electron with native macOS/Windows/Linux builds

🔌 Plugin System

Add any AI provider with a JSON file — no code changes needed:

{
  "id": "my-provider",
  "name": "My Provider",
  "type": "completion",
  "endpoint": "https://api.example.com/v1/chat/completions",
  "auth": {
    "header": "Authorization",
    "prefix": "Bearer ",
    "key_env": "MY_API_KEY"
  },
  "model_map": ["model-a", "model-b"]
}

Built-in plugins: OpenAI, Anthropic, Google Gemini, Groq, Mistral, OpenRouter, HuggingFace, and more.

Plugins support multi-capability (chat + TTS + image gen in one config), per-user variables, and encrypted credential storage.

📖 Full plugin docs →

🤖 AI Agent Support (OpenClaw)

Libre WebUI natively supports AI agents via the OpenClaw plugin — turning your chat interface into a full agent platform.

What agents can do:

• 🧠 Persistent memory — agents remember across sessions
• 🔧 Tool use — file access, web search, code execution, device control
• 📅 Proactive actions — scheduled tasks, reminders, heartbeat monitoring
• 🎙️ Voice messages — agents generate and send audio using local TTS
• 🖼️ Image generation — agents create images via ComfyUI, DALL·E, etc.
• 💬 Multi-channel — same agent across Telegram, Discord, Signal, Nextcloud Talk
• 🔌 Plugin-powered — configure via JSON, no code required

{
  "id": "openclaw-agent",
  "name": "OpenClaw Agent",
  "type": "completion",
  "endpoint": "http://localhost:3000/v1/chat/completions",
  "auth": {
    "header": "Authorization",
    "prefix": "Bearer ",
    "key_env": "OPENCLAW_API_KEY"
  },
  "model_map": ["agent:main"]
}

Your agent, your hardware, your data. No cloud dependency. Run Claude, GPT, Gemini, or local models as the agent's brain — all through the same self-hosted interface.

📖 OpenClaw integration docs →

🤝 Contributing

We'd love your help! Libre WebUI is built by people who care about privacy and AI freedom.

Getting started is easy:

1. Fork the repo and clone it
2. npm install && npm run dev
3. Make your changes on a branch off dev
4. Open a PR — one approving review from the TSC and you're in

Ways to contribute:

• 🐛 Bug reports — open an issue
• 💡 Feature ideas — start a discussion
• 🌍 Translations — help us reach more languages
• 📖 Documentation — every improvement helps
• 🔌 Plugins — share your provider configs

All contributors follow our Code of Conduct and Community Charter.

Security issues? Email [email protected] — we follow a 30-day coordinated disclosure.

📜 Our Promise

Libre WebUI has an Ethical Charter that guarantees:

• 🔓 Apache 2.0 forever — no bait-and-switch relicensing
• 🚫 Zero telemetry — no analytics, no tracking, no phone-home. Ever.
• 🏛️ Community governance — transparent decisions, public roadmap
• 🤝 No VC capture — funded by the community, for the community
• ✊ Ethical use — we actively oppose surveillance and weapons applications

🏢 Enterprise

Need to deploy at scale? Kroonen AI provides professional services:

| Service | Description | | ---------------------- | -------------------------------------- | | Custom Deployment | On-prem, cloud, air-gapped, Kubernetes | | SSO Integration | Okta, Azure AD, SAML, LDAP | | Custom Development | Integrations, white-labeling, plugins | | Compliance | GDPR, HIPAA, SOC 2, FedRAMP | | SLA Support | Priority response, dedicated channel |

📧 [email protected] • Learn more →

🔗 Links

| | | | -------------------- | -------------------------------------------------------------------------------- | | 🌐 Website | librewebui.org | | 📖 Documentation | docs.librewebui.org | | 🐙 GitHub | github.com/libre-webui/libre-webui | | 🦊 GitLab | git.kroonen.ai/libre-webui | | 🤗 HuggingFace | huggingface.co/libre-webui | | 𝕏 Twitter | @librewebui | | 🐘 Mastodon | @[email protected] | | ❤️ Sponsor | github.com/sponsors/libre-webui |

⭐ If Libre WebUI helps you, give us a star!

It helps others discover the project and keeps us motivated.

Apache 2.0 License • Copyright © 2025–present Libre WebUI™

Built with ❤️ by Kroonen AI and the open-source community