I thought they were saying it was more efficient, as in tokens per watt. I didn’t see a direct comparison on that metric but maybe I didn’t look well enough.