Set a hard dollar ceiling per key, team, and org. A runaway agent or a traffic spike can never turn into a surprise bill — requests stop at the limit, not after the money is gone.
Both return 429, but they solve different problems. Here is a decision guide for when to use a budget cap, when to use a rate limit, and why you almost always want both.
Auto-topup keeps your gateway from running dry mid-traffic — but done wrong it's an uncapped spend loop. Here is how to configure it with a ceiling so it helps without becoming a runaway.
A runaway agent or leaked key can burn thousands in hours. Here is how hard budget caps, soft alerts, and reserve-and-settle stop the bleeding on an LLM gateway — before the invoice arrives.
Coding agents burst into hundreds of model calls per task across many tools. Here is how a gateway gives them budgets, fallback, and per-task attribution so an autonomous loop can't run up a surprise bill.