Run Llama 3.2 3B offline
Llama 3.2 3B is an open AI model made by Meta. Quo runs it on your own computer, browser and Android phone. You download it once. After that it works with no internet, and what you type stays on your device.
- Good at: Dependable all-rounder for writing and summarising.
- Reads photos: No
- Thinks first: No: it answers straight away.
- Code mode: Basic (desktop app)
- Licence: the Llama 3.2 Community Licence (see Meta’s page for it on Hugging Face)
Where it runs in Quo
| Where | Download | Memory it needs | Cost |
|---|---|---|---|
| Desktop app (Mac, Windows, Linux) | 2 GB | 8 GB | Free |
| Quo Web (Chrome or Edge on a computer) | 2 GB | 8 GB | Free |
| Android app | 2 GB | Quo checks your phone | Premium (one-time) |
“Memory it needs” is the computer’s memory, or the graphics card’s own memory when the model runs there. A model that doesn’t fit runs slowly or not at all; Quo checks your device and suggests one that runs well.
How to run it
On a computer
- Install the desktop app: download it, or run the one-line install shown there.
- Open Models & storage → Offline and press Download next to Llama 3.2 3B.
- When it’s downloaded, start a chat. It works with Wi-Fi off.
In a browser
- Open Quo Web in Chrome or Edge on a computer.
- Open Models & storage → Offline and download Llama 3.2 3B. It stays in the browser for next time.
- Chat. Nothing you type leaves the computer.
On an Android phone
- Install Quo from Google Play.
- Open Settings → Models and choose Llama 3.2 3B. It comes with the one-time Premium upgrade.
- After the download, it answers with no signal.
Questions
Does Llama 3.2 3B work without the internet?
Yes. Quo needs the internet only to download it. After that, Llama 3.2 3B runs on your device, and your chats, documents and photos stay there.
Is it free?
Yes. In the desktop app and Quo Web it’s free, with no account and no limits. On the phone it comes with the one-time Premium upgrade.
Can I use it for my own work?
Llama 3.2 3B is released under the Llama 3.2 Community Licence. The full terms are on Meta’s page for it on Hugging Face. Quo itself is open source under Apache 2.0.