> ## Documentation Index
> Fetch the complete documentation index at: https://docs.spatius.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Spatius API Errors and Limits
> Spatius API error envelope, HTTP status mapping, job failure codes, and account limits for Avatar and Video APIs.
## Error envelope
Every Spatius API error returns the same JSON envelope, regardless of which endpoint failed:
```json theme={null}
{
"error": {
"code": "rate_limit_exceeded",
"message": "Open API rate limit exceeded",
"requestId": "b3f0c2d1-4a5b-4c6d-8e9f-0a1b2c3d4e5f"
}
}
```
* `code` is a stable machine-readable string. Program against `code`, not `message`.
* `message` is a human-readable explanation specific to this occurrence.
* `requestId` uniquely identifies the request. It is also returned in the `X-Request-ID` response header. Quote it when contacting support.
## HTTP status and error code mapping
| HTTP status | `code` | Meaning |
| - | - | - |
| `400` | `invalid_request` | Malformed request: missing required fields, an invalid ID in the path, a page token that is invalid or does not match `pageSize`, or no active avatar pipeline configured. |
| `401` | `unauthorized` | Missing or mismatched `X-App-ID` / `X-API-Key` pair. See [Authentication](/api-reference/authentication). |
| `402` | `insufficient_avatar_units` | The account has no Avatar Creations left. View your balance at [app.spatius.ai/avatars/creations](https://app.spatius.ai/avatars/creations). |
| `403` | `forbidden` | API access is not configured (`Open API access is not configured`), creation is disabled for the requested API, or the account cannot access the selected Avatar. |
| `404` | `not_found` | The avatar or job does not exist, or belongs to another account. |
| `409` | `conflict` | A video `requestId` was reused with different input. Retry the original body or use a new UUID. |
| `429` | `rate_limit_exceeded` | A rate limit or the rolling 24-hour creation quota was exceeded. |
| `429` | `concurrency_limit_exceeded` | Avatar API only: too many jobs are queued or processing at once. Wait for an active job to finish. |
| `500` | `internal_error` | An unexpected server error. Retry with backoff; quote `requestId` if it persists. |
| `501` | `unsupported_input` | The request uses an input that this API version does not support yet. |
| `503` | `service_unavailable` | A transient backend failure, including a video download-link refresh failure. Retry the request; a failed link refresh preserves the successful video job. |
## Job failure codes
When an Avatar creation job reaches `failed`, the job's `error` object carries one of these codes:
| Code | Message | Meaning |
| - | - | - |
| `source_download_failed` | `The source image could not be downloaded.` | The image URL could not be fetched or stored. Host the image on fast, reliable storage and resubmit. |
| `invalid_image` | `The source image is not a supported opaque JPEG or PNG of sufficient size.` | The file is not a valid JPEG or PNG, exceeds 5 MiB, is smaller than 340 px on its shorter side, or is a non-opaque PNG. Fix the image and resubmit. |
| `image_quality_rejected` | `The source image did not pass Avatar image requirements.` | The image passed format validation but was rejected by content checks, for example because it does not contain exactly one clearly visible face. |
| `preprocess_failed` | `The source image could not be preprocessed.` | Preprocessing failed after download. Retry; contact support if it persists. |
| `avatar_generation_failed` | `Avatar generation failed.` | Avatar generation failed or could not be started. Retry; contact support if it persists. |
| `internal_error` | — | An unexpected failure. Retry; quote the job ID when contacting support. |
The `retryable` flag on the error indicates whether resubmitting the same request may succeed.
For video failures, see [Video job errors](/api-reference/video-generation#job-errors). A failed video job requires a new creation request to render again.
## Rate limits
Avatar API limits are enforced per account. Your account may be configured below these hard caps:
| Limit | Hard cap |
| - | - |
| Avatar creations per minute | 20 |
| Avatar creations per hour | 200 |
| Avatar creations per rolling 24 hours | 2,000 |
| Concurrent active creation jobs | 10 |
| Read requests per second (token bucket) | 20, burst 40 |
| Read requests per minute | 1,200 |
Read limits apply to the `GET` endpoints; creation limits apply to `POST /avatars`. When a limit is hit, the API returns `429` with `rate_limit_exceeded` (or `concurrency_limit_exceeded` when the concurrency cap is the binding constraint). Back off and retry — for job polling, an interval of 5–10 seconds stays comfortably within the read limits.
Video counters are independent, and the Console API does not impose a video active-job quota. See [Video limits](/api-reference/video-generation#limits).
## Pagination
All list endpoints (`GET /avatars`, `GET /avatar-jobs`, and `GET /video-jobs`) paginate the same way. They accept `pagination.pageSize` and `pagination.pageToken`, and return `pagination.nextPageToken` and `pagination.totalCount`.
`pageSize` defaults to 20 and has a maximum of 99.
A `pageSize` above 99 does not return 99 items — it falls back to the default of 20. Request at most 99 to avoid receiving fewer items than you expect.
To page through results, pass the previous response's `pagination.nextPageToken` as `pagination.pageToken`. An empty `nextPageToken` means there are no more pages. Treat the token as opaque and pass back exactly what you received; do not construct or increment it yourself. An invalid page token is rejected with `400` and `invalid_request`.
Keep `pageSize` the same for every request in a pagination sequence. A token is only valid for the `pageSize` it was issued with, and reusing it with a different `pageSize` is rejected with `400` and `invalid_request`. To change the page size, start again from the first page with no `pageToken`.
## Next steps
This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.