Offline AI – Private

关于这款应用

Offline AI – Private allows users to run real AI chat entirely on their iPhone without relying on cloud services. Built by Artem Leshchenko, the app utilizes MLX on Apple Silicon to execute inference locally, ensuring that messages, files, and downloaded models remain on the device. This design eliminates the need for Wi-Fi, cellular data, user accounts, or subscriptions for basic usage.

The application supports a variety of curated open language models, including Qwen, Llama, and SmolLM, categorized by performance tiers such as Ultra Light, Fast & Balanced, High Quality, and Vision. Users can manage their chat history locally with features like search, pinning, and streaming replies. A clear model guide helps users select compatible models based on their iPhone’s RAM requirements.

For advanced tasks, the Pro version enables on-device analysis of photos and documents using SmolVLM2. Users can attach images and files directly within the chat interface, keeping all processing local. The app is designed for privacy-focused users who want to avoid uploading data to external servers while maintaining access to functional AI capabilities in airplane mode.