Set the provider limit first
Open your OpenAI organization or project settings and find spend limits. Choose a monthly amount and check whether hard enforcement is enabled. An alert only tells you that spending crossed a line; it does not stop requests.
OpenAI says enforcement is not instant, so usage may pass the amount slightly. Check the official spend-limit guide linked below for the current controls.
Technical details: Plain-English AI Help; provider API behavior, model costs and workflow controls. Test these choices on your own workload.
Set a limit for the job too
In Foreman, enter a Job dollar limit before launching. Check the conservative cost ceiling, then choose an amount you can afford. The job may stop before finishing if its next call will not fit.
Foreman reserves room for each call before dispatch. Calls running at the same time share that job allowance. The receipt shows recorded cost and reservations, including uncertain charges. This uses listed model prices; your provider invoice can differ.
Technical details: Plain-English AI Help; provider API behavior, model costs and workflow controls. Test these choices on your own workload.