A shop handing out free samples has a rule: one each. Online services have rules like that too.
At a busy market, a stall offers free cheese samples. A sign says “One per customer, please.” Without it, a few people would eat the lot and nobody else would get any.
A rate limit is that sign for online services. It caps how many requests you can make in a set time. Go over and you’re asked to wait.
Why services use them
- to share fairly, so one heavy user doesn't slow it down for everyone
- to protect against attacks that flood a site with requests
- to control costs, especially for expensive things like AI
What happens when you hit one
Usually you get a short error, often the status code 429: “Too many requests”. Well-built apps wait a moment and try again.
Where you’ll see them
AI chat apps with message limits. APIs that allow so many requests per minute. Login pages that lock you out after several wrong passwords.
What it isn’t
It isn’t a ban. Wait and you can carry on. Paying more often raises the limit.
Where the comparison breaks down
A sample stall counts per person. Online services count per account, per API key or per internet address.
How the words fit together
APIs usually have rate limits, tied to your API key. Going over gives a 429 status code.
If an AI chat says you've hit a limit and to come back later, that's a rate limit. Paid plans usually raise the cap.

When you'll hear it
“We're hitting the rate limit.” We're sending more requests than the service allows, so some are refused.
“Back off and retry.” Wait a little, then try again, instead of hammering the service.
“It's limited to 60 requests a minute.” You can ask up to 60 times each minute. After that you have to wait.
“If our usage doubled tomorrow, would we run into any service's limits?”


