On July 30th, 2026, between approximately 6:20 a.m. and 7:47 a.m. PDT, some users experienced errors when starting or continuing ChatGPT conversations. The issue occurred when available capacity was insufficient for the amount of processing required by some conversations. Engineers added and reallocated capacity, restoring service by approximately 7:47 a.m. PDT.
Impact
During this period:
Some Free, Plus, Pro, and Go users experienced ChatGPT conversation failures on Web, iOS, and Android.
Public API endpoints were not affected.
Root Cause
Available capacity was insufficient for the amount of processing required by some ChatGPT conversations. As the affected systems reached their capacity limits, some conversation requests could not be completed.
Resolution
Engineers began adding capacity at approximately 7:26 a.m. PDT and reallocated additional capacity to the most affected area at approximately 7:37 a.m. PDT. As the additional capacity came online, conversation errors recovered across affected plans. Service was fully restored by approximately 7:47 a.m. PDT.
Prevention and Improvements
We have implemented or are working on several improvements to reduce the likelihood and impact of similar incidents:
Improve monitoring for usable capacity and demand so pressure is detected before it affects conversations.
Improve capacity controls so constrained systems can be detected and relieved earlier.
Maintain and validate reserve capacity across affected plans before and after capacity is moved.
Add safeguards and tests for high-load conditions and retry behavior so retries do not amplify pressure.
We apologize for the disruption and are continuing to strengthen the reliability of the systems that support ChatGPT conversations.
