OpenAI restores the Codex 5-hour limit and increases usage time by 18% after optimization

In response to recent reports from many developers and enterprise users that the quota consumption of the GPT-5.6 Sol model is much faster than expected, OpenAI product manager Tibo (@thsottiaux) announced on the social platform After optimization, usage time increased by 18%. Tibo first clarified in the statement that OpenAI has not reduced the usage quota of any subscription plan. The quota has been consumed too quickly in the past few weeks. The main reason is that the GPT-5.6 Sol model tends to use deeper reasoning, initiate more tool calls (Tool Calls) and complex workflow coordination when executing tasks, resulting in a significant increase in the number of Tokens consumed in a single task. After team investigation and technical improvements, OpenAI has successfully optimized system efficiency. According to official estimates, under normal usage scenarios, users using GPT-5.6 Sol will be able to extend their quota battery life by approximately 18%, and some users will be able to experience significant improvements from now on. The 5-hour usage limit will be restored tomorrow. In addition, the "5-hour usage limit" that was temporarily suspended due to internal investigation will also be officially resumed tomorrow. For some developers (especially Plus users) it may be more disruptive to their work.