Rate limits
Request limits per key and per account, the 429 response and its headers.
Developer documentation
Requests are counted per API key and per account:
| Requests per key | 300 requests per 60-second window |
|---|---|
| Writes per key (non-GET) | 60 requests per 60-second window |
| Requests per account (all keys) | 1,200 requests per 60-second window |
- When a limit is exceeded you get
429 rate_limited. TheRetry-Afterheader gives the seconds to wait before retrying. - Responses carry the
RateLimit-PolicyandRateLimitheaders. These headers follow an IETF draft specification; their format can change with the draft. Build your client onRetry-Afterfirst. - The counter is bound to the API key's family: rotating the key does not reset it.
- Failed authentication attempts are counted in a separate, stricter bucket per client address. When that bucket is full, keys from that address that cannot be resolved get
429; valid keys keep working.
Recommendations
- Do bulk syncs with large pages (
limit=100) andupdated_since(Pagination). - Use webhooks instead of asking for changes often.
- On
429and503wait forRetry-After; increase the wait on repeated errors.