AnythingLLM Mobile: Your private AI assistant, on the go.
Run LLMs on-device, connect to your AnythingLLM instance, or use any compatible provider — all from your phone.

| | Docs | Download

👉 AnythingLLM for desktop Android! Download Now

---

Product Overview

AnythingLLM Mobile brings the power of AnythingLLM to your Android device. Run large language models directly on your phone, connect to a self-hosted AnythingLLM instance over your local network, or chat with any OpenAI-compatible provider — all with a beautiful, native experience.

The design thesis of a "Local AI First" of AnythingLLM applies to the mobile app as well. So anything that you can do in the mobile app will always be local first.

Features

• On-device LLM inference — Run GGUF models locally on your phone using Cactus Compute (llama.cpp for React Native).
• Connect to AnythingLLM — Pair with your self-hosted AnythingLLM desktop or server instance for full workspace access.
• Multiple AI providers — Use Ollama, LM Studio, OpenRouter, or any OpenAI-compatible API.
• Workspace chat — Organize conversations into workspaces, just like the desktop app.
• Document chat — Upload and chat with your documents via a connected AnythingLLM instance.
• Beautiful native UI — Built with React Native, NativeWind, and a polished dark-mode-first design.
• Privacy-first — Your data stays on your device or your own server. No third-party telemetry.

Supported Providers

Provider	Type
On-device (GGUF via cactus)	Local inference
AnythingLLM Instance	Delegated (LAN/remote)
Ollama	Local/remote
LM Studio	Local/remote
OpenRouter	Cloud
Any OpenAI-compatible API	Cloud/local

Supported Operating Systems

• Android
• iOS

How to setup for development

Prerequisites

• Node.js >= 18
• Yarn
• React Native development environment (setup guide)
• Android SDK (for Android builds)
• Xcode (for iOS builds, macOS only)

Getting Started

# Install dependencies
yarn install

# Start Metro bundler
yarn start

# Run on Android
yarn android

# Run on iOS (macOS only)
cd ios && pod install && cd ..
npx react-native run-ios

Building for Release

1. Place your anythingllm-upload-key.keystore in android/app/
2. Set APP_RELEASE_STORE_PASSWORD and APP_RELEASE_KEY_PASSWORD environment variables
3. Run yarn build:android:release
4. Output will be in the release/ folder

Technical Overview

This is a React Native application targeting Android (with iOS support planned). Key technologies:

• React Native 0.76 — Core framework
• NativeWind / Tailwind CSS — Styling
• MobX — State management
• WatermelonDB — Local database
• cactus-react-native — On-device LLM inference (llama.cpp bindings)
• React Navigation — Routing and navigation

Project Structure

src/
├── components/     # Reusable UI components
├── contexts/       # React contexts
├── database/       # WatermelonDB models & schema
├── hooks/          # Custom React hooks
├── screens/        # App screens
├── services/       # Business logic & API services
├── store/          # MobX stores
└── utils/          # Utilities & AI provider implementations

👋 Contributing

• Create issues for bugs or feature requests
• PRs are welcome — please follow existing code style and run yarn lint before submitting

🔗 Related Projects

• AnythingLLM: The all-in-one AI app for desktop and self-hosting.
• AnythingLLM Embed: Embeddable chat widget for websites.
• Cactus Compute: The runtime for running LLMs on-device.

---

Copyright © 2025 Mintplex Labs.
This project is MIT licensed.