OpenAI has reset Codex's quota and will reinstate a 5-hour limit starting tomorrow.

According to Dynamic Beating monitoring, OpenAI has reset the usage limits for ChatGPT Work and Codex users. The previously paused 5-hour usage limit will also be reinstated tomorrow.

Many users have noticed that GPT-5.6 Sol is consuming the quota faster than expected. The reason is not a reduction in subscription limits, but rather the model's increased willingness to work for longer periods and make more tool and sub-agent calls.

Especially when waiting for tool execution while continuing to handle other tasks, Sol generates more tool calls, responses, and cache tokens. As a result, complex tasks and extensive web searches are more likely to quickly deplete the quota.

OpenAI has optimized the relevant mechanisms. Under normal usage, the quota is expected to be approximately 18% more efficient.