Errors · RedRouter Docs

Check the HTTP status and JSON error. Streaming failures may arrive after HTTP 200.

HTTP statuses

Status	Action
400	Fix the request or remove unsupported fields.
401	Check your API key.
402 / 403	Check credits and key permissions. Messages uses 403 for insufficient credit; Chat Completions uses 402.
404	Check the model ID. Token counting always returns 404.
413	Reduce the request body.
429	Wait for the `Retry-After` seconds, then retry. The code names the limit: `concurrency_limit_exceeded`, `requests_per_minute_limit_exceeded` or `daily_spend_limit_exceeded`.
503	`supplier_budget_exhausted`: daily service budget reached; retry after 00:00 UTC. `model_usage_brake`: this model is paused for a usage reporting investigation; retry later or choose another model. Both are uncharged and include `Retry-After` seconds (300 for the model brake).
Other 5xx	Check request usage before retrying a generation.

Error bodies

{
  "error": {
    "message": "max_tokens must be a positive integer.",
    "type": "invalid_request_error",
    "code": "invalid_value",
    "param": "max_tokens"
  }
}

{
  "type": "error",
  "error": {
    "type": "invalid_request_error",
    "message": "max_tokens is required and must be a positive integer."
  }
}

Retry requests

Use bounded exponential backoff for temporary failures. A disconnected request may have generated tokens. Check usage before sending it again.