Rate limits

AI Gen Studio applies rate limits by API key, user account, and IP address. It also applies capacity admission before reserving credits.

Create limits

create requests per hour per API key
results maximum per job
active job maximum per API key
active jobs maximum per account

A result is one requested output. For example, 5 source images with number_of_images: 2 creates 10 results.

Polling limits

120 GET /jobs or GET /jobs/{jobId} requests per hour per API key

Do not poll each result individually. Poll the job status endpoint and read all result statuses from the response.

Capacity backpressure

Background removal and Afri-backed tools have independent capacity groups. Each group limits queued results, active results, and the share one account may hold. A batch counts every requested result, so larger batches cannot bypass the per-account pending limit. When capacity is temporarily unavailable, the API returns one of:

too_many_active_jobs
too_many_pending_results
capacity_unavailable

These responses include Retry-After. No job is created and no credits are reserved. Retry the same request with the same Idempotency-Key after the indicated delay.

Too many requests

Rate limited responses use HTTP 429:

{
  "error": {
    "code": "rate_limit_exceeded",
    "message": "You have reached the allowed request limit. Please wait before trying again."
  },
  "request_id": "req_abc123"
}

Getting started

Image API guides

Guides

Create limits

Polling limits

Capacity backpressure

Too many requests

​Create limits

​Polling limits

​Capacity backpressure

​Too many requests

Create limits

Polling limits

Capacity backpressure

Too many requests