How do I permanently change tts models?

If I run tts --list_models, I get a list of available compatible models to download. If I get the system to use one of these models, it downloads it perfectly fine. However I can't find any way to then make the container use this different model every time by default when the container is started up?

How do I switch models in this way?