GPT-5.5's Capability Degradation in Codex Investigated, OpenAI Completes Fix and Resets Usage Limit

According to Watchful AI monitoring, Tibo, the lead of OpenAI Codex, confirmed today that the team has identified and fixed two vulnerabilities. These were the root causes of the recent performance degradation in the Codex terminal programming AI, GPT-5.5, over the past two days. The system is now fully restored. To compensate developers, OpenAI has reset all user usage limits.

Over the past 48 hours, some users have reported a deterioration in the performance of GPT-5.5 in Codex. In the early stages of the official investigation, no definitive conclusion was reached. However, this afternoon, the latest monitoring metrics indicate that all services have stabilized once again.