Rate Limits

Rate limits are enforced per authenticated account, keyed off your API key — not per IP address. That means the same client hitting the API from multiple machines/IPs still shares one limit, and different customers sharing an outbound IP (behind a corporate proxy, for example) don't affect each other.

Limits by endpoint category

These are the current defaults and may be adjusted over time; the response headers described below always reflect whatever is actually in effect for the endpoint you called, so build against the headers rather than hardcoding these numbers.

Response headers

Every rate-limited response — successful or not — carries three headers describing where you stand:

Check X-RateLimit-Remaining proactively in a polling loop and back off before you hit zero, rather than waiting to react to a 429.

When you exceed a limit

Exceeding a limit returns HTTP 429 with the standard v1 error shape:

{
  "error": {
    "type": "rate_limit_error",
    "code": "rate_limit_exceeded",
    "message": "Too many requests. Slow down and retry after the window resets.",
    "doc_url": "https://imagetotable.ai/developers/guides/errors#rate_limit_exceeded"
  }
}

If you're polling for batch status, prefer registering a webhook over a tight polling loop — it eliminates the risk of tripping the status-polling limit entirely for that use case.