A spend cap is a hard upper bound on spending over a period. Once hit, requests are blocked until the window resets or the cap is raised. Caps convert an unbounded cost risk into a predictable ceiling, and are the complement to token budgets and rate limits.
Why it matters
A cap converts unbounded cost risk into a predictable ceiling. It is the backstop when a budget is misconfigured or traffic spikes unexpectedly.
How it works
Once a spending window reaches the cap, further requests are blocked until it resets or the cap is raised. Caps complement token budgets and rate limits: budgets shape normal use, caps bound the worst case.