OpenAI has rolled out new options in ChatGPT’s model picker, giving Plus subscribers more control over which variant of GPT-5 or previous models they want to use. In a post on X, CEO Sam Altman revealed that users can now choose between three GPT-5 variants; namely “Auto”, “Fast” and “Thinking”, as well as the recently reinstated GPT-4o.
Altman explained that most users will likely prefer Auto, the standard reasoning model. Fast prioritises quick responses, while Thinking is designed to deliver longer and more comprehensive answers. For GPT-5 Thinking, OpenAI has introduced a rate limit of 3,000 messages per week, after which users will be switched to GPT-5 Thinking mini. Altman also confirmed that work is underway to adjust GPT-5’s personality to make it “warmer.”
GPT-4o, which was removed during the GPT-5 launch, is now back in the model picker for all paying users by default. OpenAI initially removed the option to choose between models in an effort to simplify the ChatGPT experience, with its newer model meant to unify performance across all queries. However, the decision faced backlash from users who said they preferred GPT-4o’s personality and weren’t given prior notice of its removal.
OpenAI restored GPT-4o access for Plus subscribers during the weekend, just days after replacing it with its newer GPT-5 model. Altman has since promised that, should the older model be deprecated in the future, “plenty of notice” will be provided.
In addition to GPT-5 and GPT-4o, paying users can toggle “Show additional models” in ChatGPT’s web settings to access o3, 4.1, and GPT-5 Thinking mini. The more powerful GPT-4.5 remains exclusive to the Pro subscribers tier due to high GPU costs.
(Source: Sam Altman, via X)
