Rate limits

Limits apply per API key, not per plan.

Limit	Value
Burst	50 requests per second
Sustained	600 requests per minute
Heavier operations	Their own per-minute limits; for example invoice writes 120 per minute and submissions 60 per minute

A refused request is answered 429 with errorCode SYS005, a Retry-After header giving the seconds to wait, and the state of the most constrained limit:

Header	Meaning
`X-RateLimit-Limit`	The size of the limit that refused you
`X-RateLimit-Remaining`	Requests left in it
`X-RateLimit-Reset`	Seconds until it refills


{
  "meta": {
    "statusCode": 429,
    "success": false,
    "message": "Too many requests",
    "errorCode": "SYS005",
    "errors": [],
    "timestamp": "2026-10-01T10:00:00.000Z",
    "requestId": "01K6P3Z0X7Q4N2M8R9V1T5B3Y7"
  },
  "data": null
}

Staying under the limits

Honour Retry-After. Do not retry a 429 sooner.
Use webhooks instead of polling where you can. Where you must poll, poll the invoices you are waiting on, not everything.
Use POST /i/v1/invoices/batch-submit to submit many invoices in one request.
Cache reference data. Code lists, plans and credit costs change rarely.
Spread scheduled jobs. Sixty submissions in one second will meet the burst limit; the same sixty over a minute will not.