Rate limit and usage limit
A rate limit caps how much you may send in a window of time, measured on tokens processed rather than on the number of messages sent.
Subscription plans usually enforce a session limit and a longer rolling one, both metered on usage weighted by model. Two people sending the same number of prompts can therefore land in very different places.
Limits tend to arrive mid-task rather than at the end of a slow week, because the expensive turns are the ones in the middle of a long session. Turn twenty carries nineteen turns of history plus everything read along the way.
Nothing resets a limit early and no flag raises it. What changes is the rate at which a window is consumed, which is a function of what each turn carries.