Error codes

Handle authentication, validation, capacity, and backend errors.

Client errors#

400: malformed JSON or an invalid option. 401: missing, invalid or revoked authentication. 403: disallowed browser origin. 404: unsupported route. 405: incorrect method. 413: body too large. 415: unsupported content type. 429: account/global admission limit.

Service errors#

502: upstream transport or protocol failure. 503: service configuration or readiness unavailable. 504: inference deadline exceeded. Streaming failures may be sent as an SSE error event after headers have already been sent.

Error envelope#

Preserve X-Request-ID for diagnostics. Do not send credentials or private prompts in support material.

{
  "error": {
    "type":"rate_limit_exceeded",
    "code":"rate_limit_exceeded",
    "message":"Capacity or API key limit reached"
  }
}