Peter@peterlimg ยท Oct 7Grades each request and runs its model calls at low through max effort, lowering effort by default to save time and tokens.