How OnDeviceAI Works

Understanding the revolutionary hybrid AI system combining Apple Intelligence with advanced MLX models

Hybrid AI. Instant. Private.

1

Install & Start

Download OnDeviceAI and start chatting instantly with Apple Intelligence. No setup, no downloads, no waiting.

2

Intelligent Routing

The app automatically chooses between Apple Intelligence (for instant responses) and MLX models (for advanced tasks).

3

Enhanced Capabilities

Optionally download specialized MLX models for advanced reasoning, coding, or domain-specific tasks.

Hybrid AI Architecture

Revolutionary combination of Apple Intelligence and advanced MLX models

🧠 Apple Intelligence

Native iOS AI integration provides instant, high-quality responses with zero latency. Available on iOS 18.1+ compatible devices.

🚀 MLX Enhancement

Advanced models running via Apple's MLX Swift framework for specialized tasks requiring deeper reasoning and domain expertise.

🧠 Intelligent Router

Smart system automatically selects the best AI model based on query complexity, device capabilities, and user preferences.

🔒 Complete Privacy

Both Apple Intelligence and MLX models process everything on-device. Zero data transmission, complete iOS ecosystem integration.

Frequently Asked Questions

How does the hybrid AI system work?

OnDeviceAI intelligently routes your queries between Apple Intelligence (for instant responses) and specialized MLX models (for advanced tasks). This gives you the best of both worlds: speed and sophistication.

What devices support Apple Intelligence?

Apple Intelligence requires iOS 18.1+ on iPhone 15 Pro or later, iPad with M1+, or Mac with M1+. On other devices, OnDeviceAI uses MLX models exclusively.

Do I need to download models to start using OnDeviceAI?

No! Apple Intelligence provides instant functionality with zero setup. MLX models are optional downloads for users who want enhanced capabilities for specific tasks.

How much storage do enhanced models require?

MLX models range from 350MB to several GB. Apple Intelligence requires no additional storage. The app recommends models based on your device's capabilities and available space.

Is my data really private?

Absolutely. OnDeviceAI has no servers, no analytics, no user accounts, and no data collection. Everything stays on your device, encrypted by iOS.

Need Help or Have Questions?

For technical support, feature requests, or general questions about OnDeviceAI, reach out directly to the developer.

🐦 DM @SimonAzoulayFr on X/Twitter