Increased error rates in ChatGPT

Write-up

Approximately between April 2, 6:00 am to 10:47 am PDT, some users experienced issues initiating chat conversations.

The primary contributing factor was a new gateway layer which was being rolled out was not configured to allow enough concurrent parallel requests per node.

This led to requests which exceeded the parallel request threshold being rejected with an error. The resolution was achieved by increasing the parallel request threshold.

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.