The amount of data you cite is unlikely, as individual files can only have 2 million tokens of text extracted from each, and maximum of 20 files per assistant.
It would take a LOT of unused file overhead that can’t be extracted to approach that data usage on one assistant (but you can expect to be billed for the file size, not the effective document extraction).
If you enable retrieval for a specific Assistant, all the files attached will be automatically indexed and you will be charged the $0.20/GB per assistant per day. You can enabled/disable retrieval by using the Modify Assistant endpoint.
Note “per assistant”. Then the billing continues to per-file attached per message per-thread left undeleted.
There is no update since devday (except for the delaying the sticker shock after the free period to until developers were to invest time in naively using this)