Rate limits
Limits per key, the headers that report them, and how to stay inside.
Limits apply per API key, per minute. These headers report where you stand:
| Header | Meaning |
|---|---|
X-RateLimit-Limit | Requests allowed per minute for this key |
X-Usage-Remaining | On POST /v1/extract: pages left in the project's plan this period, overage not included |
Retry-After | On 429 and 503: seconds to wait |
The limit is 60 requests per minute for every key; ask us if a batch job needs more. To stay inside it, run at most a few requests in parallel, retry 429 after Retry-After, and spread large backfills over time. A PDF counts as one request however many pages it has.
Rate limits are separate from usage: going past the plan's included units does not slow you down. It is invoiced as overage, up to the project's overage cap for the period, set on Limits; past the cap, calls answer 402 until you raise it.