Skip to main content
Every error has the same shape:
code is for your code, and message is for a person: it says what to fix.

Codes

Retrying safely

Retry 5xx, 429 and idempotency_in_progress with backoff. Send an Idempotency-Key on any POST that creates something or places a call, so a retry after a timeout never does it twice. See Idempotency-Key.

Rate limits

  • Requests: 600 a minute per key unless you set another rate on it.
  • Call placements: at most 60 a minute, on POST /calls and POST /calls/ai.
  • Headers on every response: X-RateLimit-Limit, X-RateLimit-Remaining and X-RateLimit-Reset.
  • Over the limit: 429 with Retry-After.

Sizes

A field that isn’t in the documented shape is refused, not ignored, so a typo can’t silently do nothing.