API reference
Errors
Errors use HTTP status codes and one JSON shape.
{
"error": {
"type": "rate_limit_error",
"code": "rate_limited",
"message": "Rate limit: 10 requests a minute per key. Retry after 41 s.",
"request_id": "req_Lr0d1sY2rE9q",
"doc_url": "https://vitalsbio-console.pages.dev/docs/errors/#rate_limited"
}
}
Every error has type, code, message, request_id and doc_url. The request id is also in the X-Request-Id header of every response, errors or not. param is added when one option is at fault. Failed calls are not metered.
| Status | Code | What to do |
|---|---|---|
| 400 | missing_video, invalid_video_b64, unsupported_source | Send the video bytes, or video_b64 in JSON. Paths and URLs are not accepted. |
| 400 | invalid_species, invalid_parameter, invalid_json, unknown_sample, invalid_scope | Fix the option named in param. |
| 401 | missing_api_key, invalid_api_key, api_key_revoked, api_key_expired | See authentication. Expired and revoked keys can't be brought back; create a new one. |
| 403 | insufficient_scope | This key is limited to other endpoints. Use another key, or widen its scopes in the console. |
| 403 | forbidden_origin | Console sessions only work inside the console. Use a key. |
| 404 | not_found | No such route. Check the path and the /v1 prefix. |
| 405 | method_not_allowed | Model endpoints take POST. The Allow header says which method to use. |
| 413 | file_too_large, body_too_large | Keep videos under 25 MB. Cut or re-encode the clip. |
| 415 | unsupported_media_type | Use video/*, application/octet-stream or application/json. |
| 422 | video_rejected | The model could not read the video (for example, not a video file). The message says why. |
| 429 | rate_limited | Wait for Retry-After seconds. The clients do this for you. |
| 429 | monthly_quota_exceeded | The plan's video minutes for this month are used up. They reset on the 1st, UTC. |
| 429 | key_cap_exceeded | This key reached the monthly cap you set on it. Raise the cap in the console. |
| 502 | inference_error | The model failed on this video. Try again; if it repeats, send us the request id. |
| 503 | inference_unavailable | No inference backend is reachable. Nothing ran and nothing was counted. GET /v1/health tells you when it is up. Sample calls still work. |
| 504 | inference_timeout | The call took more than 5 minutes. Use a smaller max_seconds. |
| 500 | internal_error | Our fault. Retry with backoff. |