Rate limits
60 requests per minute per key, best effort.
60 requests per minute, per key. This is best effort, not a guarantee: the counter lives in the memory of each server instance, and the same key can land on different instances and in practice go a little past 60 before being refused.
The error always leans toward letting more calls through, never toward refusing a legitimate one. If your use depends on an exact ceiling, do not treat this number as a hard limit. The real ceiling is your plan's credit balance.
Going over answers 429, with a Retry-After: 60 header:
{ "error": { "code": "rate_limited", "message": "too many requests: the limit is 60 requests per minute per API key" } }