OpenAI

Chat completions endpoint errors
Resolved

On March 20, 2025, between 4:41 PM and 5:12 PM PT, our chat completions endpoint experienced an elevated rate of HTTP 431 errors in a single cluster. Customers whose requests were routed to the impacted cluster experienced an average error rate of 70%. The global error rate was approximately 8%. Customers routed through other clusters experienced no impact.


The root cause was a faulty code deployment that was only enabled in the affected cluster. Upon identifying the issue, we took immediate action to evacuate traffic from the impacted cluster. Full service was restored by 5:12 PM. The faulty code has since been reverted, and additional monitoring has been implemented to prevent similar incidents in the future.

Thu, Mar 20, 2025, 11:41 PM
1 week ago
Affected components

No components marked as affected

Updates

Resolved

On March 20, 2025, between 4:41 PM and 5:12 PM PT, our chat completions endpoint experienced an elevated rate of HTTP 431 errors in a single cluster. Customers whose requests were routed to the impacted cluster experienced an average error rate of 70%. The global error rate was approximately 8%. Customers routed through other clusters experienced no impact.


The root cause was a faulty code deployment that was only enabled in the affected cluster. Upon identifying the issue, we took immediate action to evacuate traffic from the impacted cluster. Full service was restored by 5:12 PM. The faulty code has since been reverted, and additional monitoring has been implemented to prevent similar incidents in the future.

Thu, Mar 20, 2025, 11:41 PM
Powered by

Availability metrics are reported at an aggregate level across all tiers, models and error types. Individual customer availability may vary depending on their subscription tier as well as the specific model and API features in use.