Operate
Rate limits and quotas
Handle quotas and request limits without interrupting a publishing workflow.
Account and key limits
API capacity is attached to the account entitlement and may also be narrowed per key. A successful billable publication consumes capacity once, even when an idempotent response is replayed.
429 is not 402
A 429 response is transient and includes Retry-After. A 402 quota_exhausted response requires additional entitlement and should not be placed into an automatic retry loop.
Plan batches defensively
Surface remaining capacity before a long batch begins and stop admitting new final publications when the available allowance cannot cover the batch.