Still copy-pasting the same time-intelligence measures across models?