Docs
Reference

Errors and limits

Errors follow the OpenAI shape. Quota and concurrency limits are per account and announced with Retry-After.

HTTP
HTTP/1.1 429 Too Many Requests
Retry-After: 5

{
  "error": {
    "message": "Monthly quota exceeded.",
    "type": "rate_limit_error",
    "code": "quota_exceeded"
  }
}

Error codes

StatuscodeMeaning
400invalid_requestThe body could not be parsed.
401missing_api_key · invalid_api_key · token_expiredNo key, an invalid key, or an expired session token.
413file_too_large · audio_too_longThe 100 MB or 2 hour limit was exceeded.
422invalid_textAlania: empty text, control characters or the character limit; unknown style.
429quota_exceeded · rate_limit_exceeded · rate_limitedMonthly quota, per-key request limit or a full generation queue. Retry-After is in seconds.
503server_busy · model_loading · service_disabled · demo_unavailableTemporary; retry after Retry-After.

Limits

LimitValue
Duyu · file size100 MB
Duyu · file length2 hours
Duyu · concurrent requests / streams (per key)3 / 2
Alania · text length5000 characters
Session token lifetime10 minutes
During the launch campaign quotas are not enforced for three months on any account; the limits above exist only to prevent abuse.