Overview

• OpenAI has reinstated GPT-4o as a selectable legacy option for paid subscribers and reactivated the model picker in settings
• Rate limits for GPT-5’s reasoning features have been increased above pre-launch levels, offering ChatGPT Plus users up to 3,000 messages per week
• The internal autoswitcher that routes queries to the optimal AI model was repaired to resolve earlier performance degradation
• An upcoming interface update will show users exactly which AI model is responding to their prompts to enhance transparency
• OpenAI is monitoring ongoing user feedback to guide further refinements as GPT-5 remains the default model