A lot of people have their own setups using pi with llama.cpp. I have my own so thought I'll share. This is the extension I and deepseek wrote: https://github.com/am17an/pi-llama-server . It allows you to do two very simple things: auto detect a llama-server running and list the models available. He
