Internal
Rate limits
A per key allowance, counted in a sliding minute, reported on every response.
Requests are counted per key in a sliding window of 60 seconds. The allowance comes from the plan of the workspace the key belongs to.
Allowance by plan
Basic60 a minutePlus300 a minuteMax600 a minuteThese come back on every response, not only on a refusal. A client that can see its remaining budget can slow down before it is refused.
Response headers
RateLimit-LimitintegerThe allowance for this window.
RateLimit-RemainingintegerHow many are left in it.
RateLimit-ResetintegerSeconds until the window resets.
RateLimit-PolicystringThe policy in the IETF form, for example 60;w=60.
Retry-AfterintegerOnly on a 429. Wait this many seconds before trying again.
Being refused costs a request. Back off on the number in Retry-After rather than retrying immediately.
Was this page helpful?