That's useful, thank you! But it also highlights my point: Why do I have to observe minor details about how the result is being presented to me to know which model was used?

I get the intent to abstract it all behind a chat interface, but this seems a bit too much.