OpenAI Brings Again Fan-Favourite GPT-4o After a Large Person Revolt

After a disastrous 72 hours that noticed its most loyal customers in open revolt, OpenAI is making a significant U-turn.

In a collection of posts on X (previously Twitter) Sunday, CEO Sam Altman introduced that the corporate is bringing again its beloved older AI fashions, together with GPT-4o, and dramatically growing utilization limits for paying subscribers, a transparent peace providing to a livid buyer base.

The transfer comes simply days after the botched rollout of GPT-5, the corporate’s newest and strongest mannequin. The launch, which ought to have been a triumph, as an alternative sparked a firestorm.

the proportion of customers utilizing reasoning fashions every day is considerably growing; for instance, without spending a dime customers we went from <1% to 7%, and for plus customers from 7% to 24%.

i count on use of reasoning to enormously improve over time, so price restrict will increase are essential.

— Sam Altman (@sama) August 10, 2025

A Botched Launch and a Person Mutiny

On August 7, OpenAI launched GPT-5, presenting it as a “unified system” that might mechanically route consumer queries to the perfect mannequin for the job. In doing so, it eliminated the menu that allowed customers to decide on between older, trusted fashions like GPT-4o, which was launched in March 2023.

For patrons paying for subscriptions like ChatGPT Plus ($20/month), the sudden change felt like a betrayal. They’d constructed their skilled and inventive workflows round a toolkit of various fashions, every with its personal strengths: one for creativity, one other for pure logic, one other for deep analysis. Forcing everybody onto a single new mannequin broke these workflows and eliminated their potential to cross-reference solutions to verify for errors or hallucinations. The backlash was speedy and fierce, leading to a cascade of subscription cancellations and on-line petitions.

Altman’s Public Reversal

Because the outrage grew over the weekend, Sam Altman took to X to do harm management, acknowledging the corporate had misjudged the scenario and asserting a collection of concessions.

The most important information got here in a direct reply to a consumer asking the query on everybody’s thoughts: “And bringing again 4o?”

“It’s again! go to settings and decide ‘present legacy fashions’,” Altman responded, confirming the return of the fan-favorite mannequin. Whereas GPT-5 stays the default, customers can now choose again into the older variations.

it is again! go to settings and decide “present legacy fashions”

— Sam Altman (@sama) August 10, 2025

A consumer then jokingly requested if they might be “marked as bizarre” for utilizing the legacy fashions. “Not marked as bizarre in both case!” Altman replied.

To additional appease paying prospects, Altman introduced a large improve in utilization limits for GPT-5’s strongest options. “Right now we’re considerably growing price limits for reasoning for chatgpt plus customers, and all model-class limits will shortly be increased than they have been earlier than gpt-5,” he posted.

When a consumer requested for specifics on the brand new limits, Altman revealed the brand new cap: “Making an attempt 3000 per week now!” That’s an enormous improve for a $20 monthly subscription and a transparent incentive for indignant customers to stay round.

making an attempt 3000 per week now!

— Sam Altman (@sama) August 10, 2025

A Promise of Transparency

Lastly, Altman promised extra transparency. He introduced an upcoming UI change to point out customers which mannequin is actively responding to their queries and promised an in depth weblog put up this week explaining the corporate’s “considering on how we’re going to make capability tradeoffs.”

He additionally shared information displaying the immense reputation of the brand new reasoning fashions, revealing that each day utilization amongst Plus customers has jumped from 7% to 24%, subtly justifying the corporate’s preliminary concentrate on the brand new tech, even when the rollout was a catastrophe. “Reasoning” is the AI’s potential to “assume” step-by-step to resolve a posh downside.

Ultimately, OpenAI appears to have gotten the message loud and clear. This was a uncommon and highly effective show of consumer revolt, forcing one of the vital highly effective corporations in tech to pay attention and backtrack. The compromise—holding GPT-5 because the default whereas giving customers again their alternative—is a direct recognition of the facility of a loyal group that refuses to be ignored.