The language model is inside the app. No account, no subscription, no upload. Turn on Airplane Mode and it works exactly the same.
Every other AI assistant is a thin client for someone else's server. What you type goes to a data centre, gets logged, and is often used to train the next model. This one can't do that — there is no server to send anything to. The model ships in the app and runs on the phone in your hand.
Draft, rewrite, summarise, explain, translate, plan. Conversations are saved on the device and nowhere else.
Hand it a PDF, Word or Excel file and ask about it. Text recognition runs on the phone.
Point the camera at something, or pick a photo, and ask what it is. The image never leaves the device.
Voice input and transcription, then have it summarise what was said.
Write a rough note, have the AI tidy it up, export it. Tasks with reminders, kept locally.
Connect Gmail, Outlook, Yahoo or iCloud and have it read, summarise and draft. Straight to your provider — nothing through us.
Built-in games, or describe one and watch the AI write it.
Eight accent colours, light and dark. The whole app follows instantly.
The core experience needs no connection. A few features can reach out, each one switched on by you and switched off just as easily:
| Feature | What happens |
|---|---|
| Web search | Tap the globe to let a reply draw on live results |
| In-app browser | Browse with the AI alongside to explain the page |
| Your own cloud model | Prefer ChatGPT, Claude, Gemini, Groq, Mistral, DeepSeek or Qwen? Paste your own key. You are billed by them, not by us. |
| Your own local model | Import any .gguf, or point it at Ollama or LM Studio on your network |
Private Offline AI · Built with Llama. Llama 3.2 is licensed under the Llama 3.2 Community License, © Meta Platforms, Inc. All rights reserved.