Skip to main content
FreeModel may apply per-key fairness limits to protect service availability. Limits can vary by model and current capacity. FreeModel does not publish fixed limits at this time. When a request exceeds a limit, the API returns HTTP 429 with the standard error envelope.

​
Handle 429 responses

  • Queue work instead of sending large bursts of parallel requests.
  • Retry with exponential backoff and jitter.
  • Set a maximum retry count and surface a useful error when it is reached.
  • Stream or batch work only when it suits your workload and the selected endpoint.
Do not retry immediately in a tight loop. That can extend the throttling period. See Errors for the error response format.