I was looking for something like this. The idea is:

1. Run an open model like LLAMA or Mistral on my home server
2. Access it from a client on my phone, saving my poor phone’s processor and battery.

Does something like this exist?