Home / Docs / Limits

Limits

The request-size and request-rate limits and how they are applied.

Use this page to learn how large a request may be and how fast you may send requests.

Request size

The default ceiling on a request body is 32 MiB. The plan ceiling and the route class's own limit apply too, and the smaller one wins. Over the limit is a 413.

Request rate

The rate limit is applied per workspace first and then per key, with a smooth (GCRA) window. A request over it is a 429 with Retry-After, and it does not wait in a queue. A workspace on the free endpoint is held to the free-endpoint rate whatever its other limits say.

Note The state is held in each gateway process, so with several replicas each admits up to the configured rate.

The X-RateLimit-* headers that report your remaining allowance are described in Credits and usage.

Free allowance

The request-fee allowance is a count per calendar month; see Bring your own key.