Skip to main content
Limits apply per key, per minute and per day, both across the whole API and per endpoint. Whichever is tighter is the one you hit first. Plan against the global daily ceiling.

Rules

  • Fixed windows. A burst across a boundary can briefly get twice the per-minute limit; never design around it.
  • The daily window resets at 00:00 UTC, not your local midnight. Place heavy nightly jobs relative to UTC.
  • Refused calls count. A 429 still uses budget, so hammering keeps you blocked.
  • Scope refusals do not count. 403 insufficient_scope is decided first.

Your limits

Your own limits are in the X-RateLimit-* headers and rate_limits on GET /api/v1/account, and on the API & Activity tab of the partner panel. The shop’s defaults follow; Mart can change them, so never hard-code one. The per-endpoint daily figures are not a budget you can add up: the global ceiling covers everything.

Headers

On every 2xx and on errors raised by an endpoint, for the tightest window. They are absent on 401, 403, 405, a route_not_found 404 and OPTIONS. Over a limit you get 429 rate_limit_exceeded with Retry-After in seconds. A daily breach advertises the daily reset, not 60 seconds: honour the header.

A schedule that fits

A catalogue of about 20,000 products at limit=1000 is about 20 pages. Polling stock every 15 minutes is 1,920 calls a day, every 5 minutes 5,760: both fit. Every 30 seconds would be 57,600, and fail partway through the day.