Big if true.
I'll wait for the official blog with benchmark results.
I suspect that our ability to benchmark models is waning. Much more investment required in this area, but what is the play out?