Is Cowork a token burner ?
I have been running some tasks through cowork, document and data summarising and creating reports, powerpoints or pdfs depending on the task.
Been using Opus 4.7 for this and I am in the Pro plan. Basically every query takes up 40%+ of my limit.
Is this normal ? Is this because of the way Cowork reads the whole directory and files or something ?
I am not sure if I could achieve the same results with Code CLI.
Do you guys have any ideas how to improve this insane token burn ? The documents I was digesting were not even huge. Small excel files and some 10-20 page slide-style reports (text heavy and not much of it)