Looks like it checks your device specs and downloads whatever the best model that will work? On mine it’s using a 3.5b version of llama