Reasoning models are here to stay with a reasoning parameter low medium high. So, I think this reasoning effort should be a part of the model selection or prompt. So, say someone wants to use O3 or O1 (or any incoming models) in l,m,h; one can choose based on tab requirement (reasoning effort Vs speed). This would do three things. A smoother control for users, a cleaner interface (less models options) and perhaps savings for Merlin.

This idea has been implemented by PPQ. A separate reasoning (effort) button pops up with O3 mini currently.