Yep. They also still haven’t clarified where the SIC modeling happens for the stock cabs, but the lack of any answer (except “that’s not how it works”) suggests to me that all of them do work the same as the IR implementation (a linear SIC curve, basically stacking a SIC IR on top of a cab IR). (And of course, Cliff says so too, who I’m quite inclined to trust. He knows a thing or two about making accurate models.)
It’s just not a great way of doing things for many reasons. Even if they let you pick SICs, applying them after the amp is never going to be as accurate as the Line 6, Fractal, and Neural competition.