A private AI that lives on your iPhone.
Open AI models like Gemma, Llama and Qwen, running right on your iPhone or iPad. Your chats never leave it.
Download on the App Store Requires iOS 18 or later.
Open models, downloaded once
Pick from Gemma 4 E2B and E4B, Llama 3.2 3B, and Qwen3 4B and 8B. The app ships empty; you download the models you want inside it, free, from Hugging Face.
Sized to your device
Each model shows whether it runs well, is a tight fit, or is too large for your device's memory. Models that are too large can't be downloaded, and one is recommended for you.
Private by design
Every answer is generated on your device. What you type is never sent to a server, not ours and not anyone's. No accounts, no sign-in.
Your chat, your way
Keep as many conversations as you like. Give each one its own system prompt, choose a Precise, Balanced or Creative response style, switch models mid-chat, and stop a reply at any time. Each reply shows its speed in tokens per second.
Free with ads, offline with one purchase
Offlai is free with ads, and the free version needs an internet connection to run. A one-time ad-free package removes every ad and lets you chat fully offline, even in airplane mode.
Made for iPhone and iPad
On iPad, a full split view keeps your chat list beside the open chat. The app is available in English and Spanish.
How it works
Four steps from installing the app to a private chat.
Pick a model
Start with the one recommended for your device, or compare them by size and fit.
Download it once
The model comes straight from Hugging Face and stays on your device until you delete it.
Start chatting
Ask, draft, brainstorm. Replies stream in as your device writes them.
Go offline, if you like
Get the ad-free package to drop the ads and keep chatting with no connection at all.