Rate limited

Meaning

The caller has exhausted a burst or sustained token-bucket allowance. This is temporary and distinct from the billing-period limit represented by quota-exceeded.

When it is returned

The API returns this problem when the applicable per-key or unauthenticated request bucket has no capacity for the current request.

Client handling

Retry only after the number of seconds in the Retry-After header. The same value is also present as retry_after in the problem body. Add jitter when many requests may retry together, and reduce request concurrency or rate.