Elevated Error Rate on GPT-3.5-Turbo-0125 Fine Tuned Models

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.

Resolved·Degraded performance

Error rates have returned to near zero and the fix has been completely deployed

Sat, Mar 2, 2024, 03:05 AM

(2 years ago)

Affected components

Mar 2, 2024, 02:39 AM

03:05 AM

Updates

Resolved

Error rates have returned to near zero and the fix has been completely deployed

Sat, Mar 2, 2024, 03:05 AM

Monitoring

A fix has been implemented and deployed. Error rates are still slightly elevated, but are dropping quickly. We expect service performance to return to normal within 30 minutes.

Sat, Mar 2, 2024, 02:56 AM

Identified

We are experiencing significantly elevated error rates and timeouts for requests made to gpt-3.5-turbo-0125 fine tuned models.

Sat, Mar 2, 2024, 02:39 AM(17 minutes earlier)