Local: on your device
Always free
Download open-source models like Qwen (available in 3 sizes to match your hardware) and run them directly on your computer. Everything stays on your machine: the model runs locally, and your conversations never touch a network. You don't have to trust anyone, not even PearPie. Support for connecting your own Ollama server is coming soon.
Peer-to-peer: with people you trust
Always free
If a friend has a more powerful setup, you can connect to their device over a direct, encrypted link. Your conversation goes straight from your device to theirs, with no server in between. This is great when you want to use a bigger model but don't have the hardware for it.
Cloud: leading models, European servers
Credits
For access to leading models like Claude, DeepSeek, and Mistral, PearPie connects to cloud infrastructure running in Europe. Your conversations are processed and immediately discarded, none of your data is stored. These models are provided by third parties (for example, Claude is currently accessed through AWS in Europe), but data always stays within the EU.
No account, no identity stored
When you first open PearPie, a secure cryptographic key is generated on your device. That key is your identity. It's what lets your devices recognise each other and what keeps your connections secure. There's no sign-up form, no email, no password. PearPie doesn't know who you are, and that's by design. All connections between devices are encrypted end-to-end, so even when data moves between your phone and your laptop, nobody else can read it.
Your conversations follow you, not a server
Your main PearPie app (usually on your computer) acts as the hub for your conversations. When you link your phone or tablet, conversations sync directly between those devices over a peer-to-peer connection. There's no cloud backup or central storage involved. You can even let your phone borrow your home computer's GPU, so you get fast responses on the go without sending anything to the cloud.