API reference
Errors and rate limits
Status codes, error shapes, quota behaviour and what to retry.
{
"error": {
"code": "surface_unavailable",
"message": "doubao is temporarily unavailable in this region",
"surface": "doubao",
"retryable": true
}
}| status | code | retry? |
|---|---|---|
| 400 | invalid_request | no — fix the payload |
| 401 | invalid_key | no |
| 402 | quota_exceeded | no — raise the plan limit |
| 404 | run_not_found | no |
| 409 | duplicate_run | no — an identical run is already in flight |
| 429 | rate_limited | yes, honour Retry-After |
| 503 | surface_unavailable | yes, with backoff |
Rate limits
- 60 run creations per minute per workspace on standard plans.
- 600 reads per minute per workspace.
- Concurrency on account-gated surfaces is deliberately low; large Chinese batches are queued, not rejected.