Error Codes

The API uses standard HTTP status codes. The error response body format is:

{
  "error": {
    "message": "Error description",
    "type": "invalid_request_error",
    "code": "invalid_api_key"
  }
}

Common Error Reference

HTTP Status	Common Cause	Troubleshooting
400	Request parameter error (missing required field, invalid value, unsupported resolution, etc.)	Check request body fields and values against the docs
401	Invalid API Key, malformed header, or deleted Key	Check Authorization: Bearer <key> format; confirm Key status in the console
403	Insufficient balance, Key quota exhausted, or no permission for this model	Check balance and Key configuration in the console
404	Wrong request path, or queried task ID does not exist	Check URL spelling; confirm the task ID comes from the creation endpoint
429	Request rate limit exceeded	Reduce request frequency and retry later; contact us for higher quota
500 / 503	Internal service error or upstream model service temporarily unavailable	Retry later; contact support if persistent

Video Task Errors

Video generation is asynchronous. Task failures are returned via the query endpoint's status and error fields:

Task Status	Description
queued / in_progress	Queued / in progress; keep polling
completed	Generation succeeded; result URL is in the response body
failed	Generation failed; error.message contains the reason. Failed tasks are not charged (deducted amounts are automatically refunded)

Polling Recommendation

Poll the task status every 5–10 seconds. Generation usually completes within tens of seconds to a few minutes, depending on video length and resolution.

Get Help

If you encounter an unresolved issue, record the complete request (remove the API Key), response body and occurrence time, then contact support: 控制台内联系管理员