Fullmoon is a free, open-source iOS app for chatting with private, local large language models directly on your iPhone. Instead of sending your conversations to a cloud service, Fullmoon downloads open models and runs them on-device, so your chats stay on your phone. The project has attracted 2,268 stars on GitHub and is released under the MIT license.
Fullmoon exists for people who want the convenience of a polished chat app without giving up privacy. Pick a local model, download it once, and you can converse entirely offline — on a plane, in a dead zone, or anywhere you would rather not stream personal questions to a remote server. Because inference happens on the device itself, there are no accounts, no subscriptions and no per-message costs, and response data never leaves your handset.
Because everything runs locally, Fullmoon is also a neat way to feel how far phone hardware has come: modern iPhones can hold a multi-billion-parameter model in memory and generate text at readable speeds. You can keep several models for different jobs — a small fast one for quick questions, a larger one when quality matters — and switch between them in the app.
Fullmoon is completely free and open source under the MIT license. There are no in-app purchases or cloud fees; the only requirement is an iPhone capable of running the local models you choose.
Fullmoon is for privacy-conscious iPhone users, offline travellers and tinkerers who want to experiment with open models on mobile hardware. Expect smaller models than cloud flagships — the trade-off for total privacy and zero running cost.