Rate Limits

Every request counts towards a global limit, some endpoints have an additional, lower limit.

Limit	Endpoints	Requests	Window
Global	All endpoints	1000	1 hour
Items	/v1/items and /v1/sales/history together	8	5 minutes

How requests are counted

• With an API key, requests are counted per API key (the Authorization header).
• Without an API key, requests are counted per network: all IPv4 addresses that only differ in the last block (e.g. 203.0.113.0 to 203.0.113.255) share one limit. If you run your client on a cloud or hosting provider, other customers in the same network may use up your limit. Send your API key to get a limit of your own.
• The windows are fixed: they start with your first request and reset after 5 minutes (or 1 hour).

When you are rate limited

Requests over the limit are answered with 429 Too Many Requests:

{
  "errors": [
    {
      "id": "rate_limit_exceeded",
      "message": "Your connection is being rate limited."
    }
  ]
}

warning

Every request you send while you are rate limited extends the block by another full window. The limit only resets after a full window (5 minutes for the items limit) without any requests. Do not retry right away.

Best practices

• Send your API key with every request.
• /v1/items and /v1/sales/history are cached for 5 minutes. Requesting them more often returns the same data, so once every 5 minutes per parameter combination (e.g. tradable=0 and tradable=1) is enough.
• Don't send your requests exactly at the full 5 minutes: wait a little longer (e.g. 5.5 minutes) or add some random delay, so you never hit the end of a window.
• After a 429, wait at least a full window without any requests before you try again.