OpenAI

Write-up
Increased API error rates

On February 26, 2025, between approximately 3:00pm and 5:00pm PST, some OpenAI API users experienced disruptions resulting in failed API requests with 429 errors, indicating "usage exceeding quota." This disruption was traced to an unintended transition of certain user accounts' billing configurations from postpaid (billing in arrears) to prepaid (credit-based usage).

The root cause of this incident was identified as an erroneous account migration script. The script, originally intended for a limited subset of users, inadvertently impacted a broader group, incorrectly changing their billing configuration and causing unexpected quota enforcement.

Resolution efforts were initially delayed due to confusion stemming from multiple simultaneous migrations. After pinpointing the exact issue, we reverted billing configurations for all affected accounts to restore normal operations.

In response to this incident, we've implemented immediate corrective measures:

  • Reduced caching intervals in our systems to expedite the unblocking of impacted accounts.

  • Establishing enhanced transparency and communication practices to broadly surface ongoing migrations affecting user accounts to relevant internal teams.

We fully recognize that extended API disruptions significantly impact our customers' products and businesses. We deeply regret falling short in this instance and remain committed to reinforcing our processes and reliability to prevent future occurrences.

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.