OpenAI’s Epic Reversal: GPT-4o is Back!

This Tuesday, OpenAI’s CEO Sam Altman shocked the AI community by announcing the triumphant return of GPT-4o to ChatGPT! This comes hot on the heels of a massive user revolt triggered by the abrupt removal of GPT-4o during last week's launch of GPT-5.

The eagerly anticipated AI model is now back in the model picker for all paid subscribers, including ChatGPT Plus accounts. This decision marks a rapid backtrack after thousands of users expressed their dissatisfaction, lamenting the loss of their favorite AI companion.

Why Did OpenAI Make the Change?

According to Altman, OpenAI didn't fully grasp how much users valued certain features of GPT-4o. In a bid to streamline the service, the company axed all previous models when introducing GPT-5 on August 7, leaving many users feeling blindsided.

The backlash was fierce, with a popular Reddit thread titled "GPT-5 is horrible" racking up over 2,000 comments in just a few days.

Enhanced User Control and Features!

In addition to restoring GPT-4o, OpenAI implemented a slew of adjustments to appease its user base. Rate limits for the new GPT-5's Thinking mode skyrocketed from 200 to a staggering 3,000 messages per week, with even more capacity available via the new "GPT-5 Thinking mini".

Moreover, new routing options like "Auto," "Fast," and "Thinking" grant users greater control over the variant of GPT-5 that tackles their inquiries.

Exclusive Features for Pro Users!

For premium users shelling out $200 a month, Altman confirmed that additional models such as o3, 4.1, and GPT-5 Thinking mini will be accessible through a "Show additional models" toggle in the ChatGPT web settings. However, GPT-4.5 will stay exclusive to Pro subscribers due to elevated GPU costs.

Addressing User Sentiments on Personality!

Beyond model availability, OpenAI took heed of widespread complaints about GPT-5's output style, which some users found "abrupt and sharp" compared to the more friendly tone of GPT-4o. Emotional users described their attachment to the old model as akin to losing a dear friend.

"We’re working on a personality update for GPT-5 that will be warmer yet not as off-putting as GPT-4o," Altman shared. He acknowledged the need for tailored personality options for users moving forward.

Issues Faced During GPT-5 Launch!

The launch of GPT-5 was riddled with complications; an automated routing system designed to choose the correct model malfunctioned, consistently defaulting to lesser capable versions. The company also faced backlash for including misleading performance graphs during the launch, which Altman labeled a "mega chart screwup."

What’s Next for ChatGPT Users?

While GPT-4o’s return is a win for users for now, OpenAI remains dedicated to refining GPT-5. The company hinted that rate limits might require further adjustments based on user engagement, and personalization updates are still in the works. For the moment, paid users can enjoy the freedom of selecting their preferred model, striking a balance between innovation and user choice.