I would suspect the point of the older comparisons is to be able to show that local models have caught up to where frontier models were a few months ago.