OpenAI Optimizes Systems and Adds Compute to Fix Capacity Errors

OpenAI has deployed a series of system optimizations and added computing power to significantly reduce 'model capacity limit reached' errors. This update is part of the team's ongoing 28-day continuous delivery initiative. This operational update improves user experience and reliability by addressing a frustrating issue where users encountered persistent capacity errors despite having sufficient subscription quotas. It ensures smoother interactions with ChatGPT and Codex by preventing requests from getting permanently stuck in an error state. In addition to reducing error frequencies through hardware scaling and scheduling adjustments, the system is designed to allow tasks to eventually resume automatically when errors do occur. The team has also invited users who still experience issues to provide feedback for further troubleshooting.