A mobile app for interacting with private, local large language models that work offline and are optimized for Apple silicon.

At a Glance:

fullmoon is an iOS app optimized for Apple silicon that lets users chat with local large language models on iPhone, iPad, and Mac, with local chat history storage and customizable appearance.

Overview:

fullmoon is a native iOS application for chatting with local large language models directly on Apple devices. It is optimized for Apple silicon and runs on iPhone, iPad, and Mac. The app stores all chat history locally on the device, and users can customize the interface appearance. fullmoon currently supports Llama 3.2 1B, Llama 3.2 3B, and DeepSeek-R1-Distill-Qwen-1.5B-4bit models.

Key Decision Points:

  • Platform and optimization: Designed specifically for Apple silicon, running natively on iPhone, iPad, and Mac.

  • Local execution: Models run entirely on-device, with chat history stored locally rather than in the cloud.

  • Supported models: Currently limited to three specific models: Llama 3.2 1B, Llama 3.2 3B, and DeepSeek-R1-Distill-Qwen-1.5B-4bit.

Core Features:

  • Local model chat: Chat interface for interacting with large language models that run directly on the device.

  • Cross-device support: Works on iPhone, iPad, and Mac through a single iOS app.

  • Local history storage: All chat history is saved locally on the user's device.

  • Appearance customization: Users can customize the visual appearance of the app.

  • Apple silicon optimization: Optimized specifically for Apple silicon hardware.

Use Cases:

  • Developers and users who need offline AI access: Interact with supported large language models without requiring an internet connection once models are downloaded.

  • Apple device users focused on local data privacy: Chat with AI models while keeping all conversation history stored locally on the device.

Open-Source Alternative Value:

As an open-source iOS app, fullmoon provides a local-only AI chat experience optimized for Apple devices. It keeps all processing and data on-device, with supported models running directly on Apple silicon and chat history saved locally. Users can customize the app's appearance, and the source code is available for inspection or modification.

PartagerXLinkedInReddit

Outils associés

Statistiques du projet

Étoiles

2,258

Forks

221

Licence

MIT

Métadonnées

Alternative à
Grok