ChatGPT’s Model Picker Brings Back 4o Model Alongside New GPT-5 Controls

OpenAI has updated the model picker for ChatGPT to reintroduce the GPT-4 model for all paid users. All users of the chatbot will also see new “Auto”, “Fast”, and “Thinking” options for GPT-5, with the Auto setting still designed to pick up the best option based on the nature of the prompt.

When it launched its new flagship GPT-5 model last week, OpenAI said that it would become the new default model for all users. However, this choice received a lot of criticism from users who preferred the responses provided by GPT-4o, and this pushed the company to bring back the old model.

Over the weekend, GPT-4o became available again for ChatGPT Plus users, but it was hidden under the “show legacy models” option in the chatbot’s settings. Yesterday, the company brought back GPT-4o in the model picker for all paid users alongside the new “Auto”, “Fast”, and “Thinking” options for GPT-5. “ If we ever do deprecate it, we will give plenty of notice,” OpenAI CEO Sam Altman also posted on X.

ChatGPT paid users can also still access legacy models such as o3, o4-mini, 4.1, and GPT-5 Thinking mini from the “Show additional models” toggle in ChatGPT web settings. The company’s GPT 4.5 model, which is more resource-intensive, is now only available for ChatGPT Pro members (it used to also be available for ChatGPT Plus users).

In addition to these ChatGPT model picker updates, the company also announced yesterday the availability of new connectors for Gmail, Google Calendar, and Google Contacts. These new connectors will also become available for Plus, Team, Enterprise, and Edu plans in the coming weeks.