FULL STORY

GPT-5.6 Sol Quota Bug: From Investigation to Fix

OpenAI investigated rapid quota consumption issues with GPT-5.6 Sol, subsequently fixing the underlying bug and increasing model quotas by 18%.

2026-07-28 ~ 2026-07-30 · 2 episodes · 9 posts

Episode 1 · OpenAI Resets Codex Limits and Investigates GPT-5.6 Sol Quota Drain (2026-07-28, 7 posts)

OpenAI recently reset usage limits for paid ChatGPT Work and Codex users while investigating the rapid consumption issue with the GPT-5.6 Sol model. The company clarified that this change does not reduce subscription quotas, but is rather a user experience shift resulting from a system fix.

Confirmed

  • OpenAI has reset the usage caps for paid users of Codex and ChatGPT Work.
  • The previously suspended 5-hour limit mechanism has been restored, though it remains temporarily paused while the GPT-5.6 Sol issue is under investigation.
  • The company is looking into reports of GPT-5.6 Sol draining quotas too quickly, stating that it is not a subscription downgrade but an experiential change following a fix.
  • OpenAI noted that after the fix, GPT-5.6 Sol lasts 18% longer.

Why It Matters

  • This incident directly impacts the actual available quotas and workflow continuity of paid ChatGPT Work and Codex users. By promptly resetting limits and pausing the 5-hour cap, OpenAI aims to protect the user experience while troubleshooting the abnormal model consumption, ensuring that bugs do not disrupt normal service access.

Episode 2 · OpenAI Reduces Codex Power Consumption, Boosts GPT-5.6 Sol Quota by 18% (2026-07-30, 2 posts)

OpenAI has fixed Codex power consumption issues and reset usage limits for all Codex and ChatGPT Work users. The update extends typical usage time for GPT-5.6 Sol by approximately 18%, with heavy users seeing even greater improvements.