What Is a Rate Limit? (Why Your AI Tool Says Slow Down)

Updated July 23, 2026 · Data as of 2026-07-09

A rate limit is a cap on how much you can use a service within a window of time, expressed as requests per minute, tokens per minute, or messages per hour. Hitting one pauses you; it does not usually cost you anything.

Rate limits exist because capacity is finite and shared. They protect the service from a single heavy user, and they are the mechanism behind the message you have probably seen after a burst of activity, telling you to wait and try again shortly. On consumer plans they are typically stated as messages in a rolling window. On developer plans they are stated more precisely, usually as both requests per minute and tokens per minute, and the token limit is often the one you hit first.

They are not the same as a usage cap or a credit allowance, though the two get confused constantly and vendors rarely help. A rate limit is about speed, and it resets in minutes or hours. A usage cap is about total volume, and it resets monthly or when you buy more. Running into a rate limit means slow down. Running out of an allowance means you are done until the cycle turns or you pay again.

The practical effect depends entirely on how you work. Steady interactive use rarely touches a limit. Bursty work does: an agent that fires many calls in parallel, a batch job over a hundred documents, a long automated run at midnight. That is where a plan that looks generous on paper starts stalling, and where the difference between tiers is really felt.

Why it matters when you choose a plan

Rate limits are the specification most often left out of a comparison and most likely to bite an unusual workload. Before subscribing, find the published limits for the exact tier you intend to buy, not the one in the marketing table, and compare them against your heaviest realistic hour rather than your average one. If you plan to run agents or batch work, this number can matter more than price. Also check whether limits rise automatically with account age or spend, since several providers do that quietly.

Frequently asked questions

Does hitting a rate limit cost me money?

No. The request is refused rather than charged, and you retry after a short wait. It costs you time, and in an automated pipeline it can cost you a failed run if nothing handles the retry.

Rate limit or usage cap, which one am I hitting?

If waiting a few minutes fixes it, it was a rate limit. If you are locked out until the next billing cycle or asked to buy more, it was a usage cap or a depleted allowance.

Can I get a higher limit?

Usually, by moving to a higher tier, and with several providers limits also increase automatically as an account matures or accumulates spend. Business plans commonly negotiate them directly.

Back to the glossary · Take the quiz · How we score tools