Rate limits
Per-key limits, reported on every response.
Every key has a per-minute request limit. Limits depend on your plan — this page won't quote a number, because it can change per plan without notice; read the limit that actually applies to your key off the response headers instead of hardcoding one.
Headers#
Every response carries the standard RateLimit-* headers, whether or not you were limited:
RateLimit-Limit— the ceiling for your key's current window.RateLimit-Remaining— how many calls you have left in it.RateLimit-Reset— seconds until the window resets.
Going over#
A call over the limit gets 429 rate_limited with a Retry-After header — wait that long before retrying rather than retrying immediately. This is not one of the fail-open error codes (see errors & fail-open): a rate limit is an integration problem to fix (batch calls, cache results, request a higher limit), not a transient failure to silently paper over.
Cache hits still count#
A cached response isn't billed against your credits, but it still counts against your rate limit — otherwise the cache would double as a way around it.