Rate limits

Limits are per API key, not per org. Two windows apply at once: a per-minute cap and a per-second burst.

Mode	Sustained	Burst
Live (`abr_live_…`)	100 requests / minute	20 / second
Test (`abr_test_…`)	30 requests / minute	20 / second

If a header disagrees with this table, trust the header.

Headers

Every authenticated /v1 response includes these (success and errors):

Header	Meaning
`X-RateLimit-Limit`	Sustained (per-minute) limit
`X-RateLimit-Remaining`	Requests left in this minute
`X-RateLimit-Reset`	Unix timestamp (seconds) when the minute window resets

A request with no X-API-Key is not rate-limited — it fails 401 first.

429 with error.code rate_limited, plus Retry-After (seconds to wait):

{
  "error": {
    "code": "rate_limited",
    "message": "Rate limit exceeded",
    "request_id": "…"
  }
}

Wait for Retry-After (or until X-RateLimit-Reset). Tight retry loops stay at 429.