Error Responses
Common HTTP status codes, OpenAI-compatible errors, and async task errors.
TENSORAXIS tries to return an error shape that matches the API style you are using. OpenAI-compatible APIs usually return an error object; asynchronous task APIs usually return code, message, and data.
OpenAI-Compatible Errors
{
"error": {
"message": "invalid request",
"type": "invalid_request_error",
"param": "",
"code": "invalid_request"
}
}Fields:
| Field | Description |
|---|---|
error.message | Human-readable error message |
error.type | Error type from TENSORAXIS or the upstream provider |
error.param | Related request parameter, when available |
error.code | Error code, usually more stable than message |
Async Task Errors
Async task APIs such as video, music, and drawing may return:
{
"code": "invalid_request",
"message": "prompt is required",
"data": null
}For video generation, a common request error is a missing prompt. If you send the upstream Volcengine-style top-level content[] body directly to /v1/video/generations, TENSORAXIS cannot read a prompt field and returns 400 prompt is required.
Common HTTP Status Codes
| Status | Meaning | Common Fix |
|---|---|---|
400 | Invalid body, parameter, or model format | Check JSON, required fields, and model name |
401 | Authentication failed | Check the token and authentication header |
403 | Permission or quota issue | Check token scope, account quota, and group access |
404 | Resource not found | Check model name, task ID, or path |
429 | Rate limit exceeded | Reduce concurrency, retry later, or adjust token/group limits |
5xx | Server or upstream error | Retry later; contact support if it persists |
For automated clients, branch on HTTP status codes and code values instead of matching the full message text. Messages may vary by upstream provider, language, or deployment configuration.