It will be really really really useful and needed. I have problems with the automatic context window management on the Assistants API. It might work on ChatGPT that they’ve tested it on, but for Assistants, that have different and novel/weird use cases, the truncation of the messages happen in the wrong moment or it cuts in half the wrong message.
I don’t know how the devs’ needs arrive to OpenAI ears, but hopefully they look here