Fullmoon

A local LLM chat app runs fully offline on Apple silicon devices, saves chats locally and connects model output to Shortcuts.

Fullmoon is a local LLM chat app for people who want conversations to stay on their Apple devices. It runs models on-device and works fully offline, with chat history saved locally.

Model choices: Chat with Llama 3.2 1B or 3B Instruct in 4-bit form, or DeepSeek-R1-Distill-Qwen-1.5B in 4-bit or 8-bit form.
Shortcuts: Chat with a local model or pass its output to other actions in an automation.
Personalization: Change the theme and fonts, and use a custom system prompt to adjust how the model responds.

The app supports iOS, iPadOS, macOS and visionOS. Its on-device processing uses Swift MLX and Metal and is optimized for Apple silicon. Fullmoon is free and open source under the MIT license.