Frequently Asked Questions

Everything you need to know about Project AIRI - Your self-hosted AI companion

General Technical Installation Usage Community Legal

What is Project AIRI?

Project AIRI is an open-source initiative to create a self-hosted AI companion inspired by Neuro-sama. It's designed to bring cyber living beings (virtual waifus, digital companions) into our world with capabilities for real-time voice chat, gaming, and interaction across multiple platforms.

What makes AIRI different from other AI projects?

AIRI stands out with its comprehensive game integration (Minecraft, Factorio), advanced voice capabilities, cross-platform support (Web, desktop, mobile), and extensive LLM provider support. Unlike other VTuber projects, it's built with modern web technologies while maintaining native performance for critical tasks.

Is AIRI really self-hosted?

Yes! Project AIRI is fully self-hostable. You can run it locally on your own hardware with complete control over your data. The web version runs entirely in your browser, while desktop apps use native performance with optional cloud services.

What technology stack does AIRI use?

AIRI is built with TypeScript and uses modern web technologies including WebGPU, WebAudio, Web Workers, and WebAssembly. The desktop version leverages native APIs (NVIDIA CUDA, Apple Metal) for performance. The architecture is modular with separate components for brain, ears, mouth, and body.

Which LLM providers are supported?

AIRI supports numerous providers including OpenAI, Claude, Grok, DeepSeek, Qwen, Gemini, xAI, Groq, Mistral, Cloudflare Workers AI, Together.ai, Fireworks.ai, Novita, Zhipu, SiliconFlow, Stepfun, Baichuan, Minimax, Moonshot AI, ModelScope, Player2, Tencent Cloud, and many more. Full list available in documentation.

How does AIRI handle speech recognition?

AIRI uses client-side speech recognition for privacy and real-time performance. It supports multiple ASR providers and includes voice detection capabilities. The system can work entirely offline or connect to cloud services depending on your setup.

What 3D model formats are supported?

AIRI supports VRM (VRM1.0 and VRM0.0) and Live2D models. It includes advanced features like auto-blinking, gaze tracking, idle eye movements, and expression synchronization. Both static and animated models are fully supported.

What are the system requirements?

Desktop: Windows 10+, macOS 10.14+, or Linux. Recommended 8GB RAM, NVIDIA GPU (optional), stable internet connection.
Web: Modern browser with WebGPU support (Chrome/Edge 113+).
Mobile: iOS 15+ or Android 10+ with PWA support.

How do I install AIRI on Windows?

1. Download the Windows installer from our releases page.
2. Run the installer and follow the setup wizard.
3. Alternatively, use winget: winget install MoeruAI.AIRI
4. Or use Scoop: scoop bucket add airi then scoop install airi/airi

How do I install AIRI on macOS?

1. Download the DMG file from our releases page.
2. Drag the app to your Applications folder.
3. Alternatively, use Homebrew: brew install --cask airi

How do I install AIRI on Linux?

1. Download the Linux AppImage from our releases page.
2. Make it executable: chmod +x AIRI-*.AppImage
3. Run with: ./AIRI-*.AppImage
4. For Nix users: nix run github:moeru-ai/airi

How do I set up voice chat?

1. Configure your microphone input in settings
2. Select your preferred speech recognition provider
3. Choose a TTS voice from available providers
4. Enable voice detection if available
5. Connect to Discord/Telegram for voice channels

How do I make AIRI play games?

1. Install supported games (Minecraft, Factorio, etc.)
2. Enable game integration in settings
3. Configure API keys if required
4. Launch AIRI and select the game mode
5. Follow in-app instructions for setup

How do I add custom 3D models?

1. Prepare your VRM or Live2D model
2. Place models in the designated folder
3. Use the model manager in the app
4. Configure animations and expressions
5. Test with voice and gestures

Where can I get help?

Join our Discord server for community support, check the documentation site for guides, and browse GitHub issues for technical help. We also have active translation projects on Crowdin.

How can I contribute to the project?

We welcome contributions in all forms: code, art, documentation, translations, or community management. Check our GitHub discussions and contributing guidelines. Areas of interest include 3D modeling, computer vision, speech synthesis, and more.

What community platforms are available?

  • Discord: Server for support and discussions
  • GitHub: Issues, PRs, and project management
  • Twitter: Project updates and news
  • Telegram: Additional community channel
  • WeChat/QQ: Chinese community support
  • Crowdin: Translation collaboration
42,267 Stars
4,240 Forks
180 Open Issues
165 Subscribers
TypeScript Language
MIT License