[jasondclinton]

If you use context cacheing, it saves quite a lot on the costs/budgets. You can cache 900k tokens if you want.