Skip to content

Limits and quotas

  • Requests. By default 60 successfully authenticated requests a minute per key; beyond that, 429 API_KEY_RATE_LIMITED. The counter is kept per minute and is shared by every service instance.
  • Key quota. An optional credit ceiling per key, counting used and reserved credits. A refusal is 402 API_KEY_QUOTA_EXCEEDED.
  • Owner balance. A new Podcast or News episode needs free credits only for its next paid request. If the balance is short, the created task waits with paused_insufficient_credits and resumes with the same taskId after a top-up.
  • Episode spending limit. Optional maximumCredits caps the Podcast or News episode’s cumulative charge. Omitted or null means balance-funded work without an episode cap. An explicit positive integer is not reduced by forecast or duration. If the next request cannot fit, the operation reports OPERATION_SPENDING_LIMIT; a balance top-up does not change that limit. The owner of an admissible failed Podcast may separately confirm a higher limit or its removal through Continue in the Mini App or with the creating API key; ordinary News resume does not increase it. Developer News does not accept an episode limit in its input and uses null; balance and the key’s separate quota bound its spending. Numeric News limits belong to other supported paths, such as schedules.
  • Body size. Up to 10 MiB for /api/v1/* requests; larger bodies get 413 PAYLOAD_TOO_LARGE.
  • Keys. At most 20 unrevoked keys per owner.
  • Concurrent generations. The service runs a bounded number of generations at once; the rest wait in the queue with stageCode: "queued".