Here is the project description based on the README content, structured according to your specifications.
Overview:
fullmoon is an iOS app for chatting with local large language models, optimized for Apple silicon and compatible with iPhone, iPad, and Mac. It allows users to run supported models directly on their device without a remote server. Chat history is stored locally on the device, and users can customize the app's appearance.
Core Features:
Local model execution: Runs supported large language models directly on the device, leveraging Apple silicon optimization.
Cross-device support: Functions on iPhone, iPad, and Mac.
Local chat history: Saves all conversation history locally on the device.
Appearance customization: Users can modify the visual appearance of the application.
Use Cases:
Privacy-conscious users: Individuals who prefer to run AI chat locally on their Apple devices to keep conversations and data on-device.
Offline interaction: Developers or users who need an AI chat tool that works without an internet connection.
Model testing: Developers testing the performance of specific small-to-medium sized models, such as Meta Llama 3.2 or DeepSeek variants, directly on Apple hardware.
Why It Matters:
As a self-contained, local-first AI chat application for Apple devices, fullmoon provides a dedicated alternative to server-based or cloud-dependent AI assistants. Its explicit support for specific model families, local storage, and on-device execution directly addresses the need for a private and independent chat experience that does not depend on network connectivity or external API keys.
