OpenAI has made changes that seem to walk back its original promise after a week of introducing GPT-5.
The company had said GPT-5 would simplify the user experience by automatically deciding how to respond, using a built-in system that picked the best method without needing input from users.
This approach was meant to make the old model picker unnecessary.
Did you know?
Subscribe – We publish new crypto explainer videos every week!
What is Chainlink? LINK Explained Simply (ANIMATED)
However, on August 12, CEO Sam Altman posted on X that the model picker has become available again, with three options: Auto, Fast, and Thinking.
The Auto setting works like the original plan, where GPT-5 handles the decision-making. Meanwhile, Fast is for quicker answers, while Thinking is for more careful responses.
Additionally, OpenAI has reintroduced some older models for subscribers. GPT-4o has been added to the picker by default. GPT-4.1 and o3 can also be switched on through the settings.
Altman also mentioned that updates to GPT-5’s tone and personality are in progress. According to his post, the team is trying to make it feel “warmer” than the current version but not as overly expressive as GPT-4o.
He added that the recent rollout has shown them that users want more control over how the model communicates with them.
Recently, xAI, Elon Musk’s artificial intelligence (AI) company, plans to take Apple to court. What happened? Read the full story.
