API Rate Limit Calculator
Convert request quotas into per-second, per-minute and spacing guidance.
Overview
The API rate limit calculator converts a quota — say "10,000 requests per day" — into the numbers that actually matter: requests per second, per minute, and the minimum spacing between calls. Drop in any window and a max-burst figure, and it returns a steady-state pace plus how long it would take to exhaust the budget at full throttle.
Backend engineers planning a polling loop, scrapers tuning their crawl rate, and integration teams sizing a third-party API contract all need the same math. Long-tail keywords covered: convert API requests per day to per second, calculate request spacing for rate limiting, and burst vs sustained quota planning.
How it works
Most rate limits are expressed as a quota over a sliding or fixed window — X requests per Y seconds. The sustained rate is simply X / Y. The minimum inter-request delay to never violate the limit, assuming evenly spaced calls, is Y / X seconds. A burst allowance lets you spend tokens faster than the steady rate as long as the bucket has not drained.
Token-bucket and leaky-bucket algorithms are the most common server-side implementations. A token bucket refills at the sustained rate and caps at the burst size; a leaky bucket drains at a fixed rate regardless of input. The math is the same from the client side — pace your calls at Y / X or risk a 429 Too Many Requests response.
Examples
- 10,000 requests per day → 0.115 requests/second → one call every 8.64 seconds.
- 60 requests per minute → 1 request/second, with a typical burst of 10.
- 5,000 requests per hour → 1.39 requests/second, spacing 720 milliseconds.
- 1 million requests per month → roughly 23 per minute or one every 2.6 seconds.
FAQ
Should I pace at the calculated minimum or leave headroom?
Always leave a safety margin — clock skew, retries, and bursts from concurrent workers can push you over. A 70–80% utilisation target is a common rule of thumb.
How do retries factor in?
Failed attempts that count against the quota effectively reduce your usable rate. If 5% of calls fail and you retry once, plan as if your quota were 95% of the published value.
What does the Retry-After header mean?
The server tells you how many seconds (or an HTTP date) to wait before retrying. Honour it precisely — backing off too quickly is a fast path to a temporary ban.
Is it safe to parallelise calls within the limit?
Yes, provided your aggregate throughput stays below the documented rate. Most APIs apply the limit per API key, not per connection, so concurrency does not buy you extra capacity.