On devices with Apple Intelligence the model is already there — install, open, ask. No cloud, no account, no tracking: your conversations never leave your device.
On supported iPhones, iPads and Macs, privateSLM uses Apple's built-in system model through the Foundation Models framework. Install, open, chat — inference runs on the Neural Engine.
No Apple Intelligence? Download a small open model once and run it locally with the built-in model manager — powered by llama.cpp. Switch engines any time.
No sign-up, no analytics, no ads, no server component at all. Chats live in a local database on your device. Turn the radio off — everything keeps working.
Multiple conversations with search, edit & regenerate, voice input with spoken replies, custom system prompt and sampling. Free to download; expert model packs are optional one-time purchases with Family Sharing.
privateSLM is private AI for one person on one device. Want your own private AI for your company? That is MegaBrain — a separate product from the same team.