OpenAI has promised advance warning of disabling older ChatGPT models

After criticism over the replacement of the GPT-4o model with GPT-5, OpenAI has changed its approach to managing the ChatGPT lineup. The company will no longer remove older versions without first notifying users.

In an interview with The Verge, ChatGPT CEO Nick Turley admitted that stopping access to GPT-4o “at least for a transition period” was a mistake. What was unexpected, he said, was the extent to which users were locked into a particular version:

“It’s not just that people have a hard time getting used to new things. It turns out they can have very strong feelings about the ‘personality’ of a model.”

GPT-5 want to make “warmer” and less annoying

Turley said the OpenAI team is working to bring to GPT-5 the “warmth” of communication that GPT-4o users enjoyed. CEO Sam Altman wrote in X Tuesday evening that the update should also be “less annoying (for most users) than GPT-4o.”

Turley said OpenAI is working to bring the “warmth” of communication that GPT-4o users enjoyed to GPT-5.

Why GPT-4o was removed

Turley said the decision to drop the old model was due to a desire to simplify choices for ChatGPT’s 700 million weekly users, most of whom only use the default model.

“It certainly wasn’t cost-related,” he emphasized. – “The main goal we’ve been striving for for a long time is simplicity. For the average user, and there are a lot of them, the idea of sorting out which model to choose for a particular answer turns out to be too complicated.”