This one is very benchmaxxed, and you can tell from this page alone. Look at the huge variance in ranking per benchmark. Most models, including at that size, are much more consistent.