OpenAI rolls back changes after massive user dissatisfaction
By ForGeeks
August 7-10, 2025 – Just 72 hours after the disastrous launch of GPT-5, OpenAI announced the return of old AI models, including the popular GPT-4o, under pressure from users. CEO Sam Altman personally stepped in to appease angry subscribers.
The company’s CEO Sam Altman has personally stepped in to calm angry subscribers.
What caused the riot?
On August 5, OpenAI introduced GPT-5 as a “universal system” that automatically selects the best model for each request. In doing so:
- The ability to manually select between different AI versions (GPT-4o, GPT-4 Turbo, etc.)
- Users lost control over model-specific workflows
- The new query redirection algorithm worked erratically, causing errors
is gone.
Professionals who depend on AI accuracy (researchers, developers, authors) complained en masse about:
- Inability to compare responses of different models for validation
- Loss of specialized features (e.g., GPT-4o was better at creative tasks)
- Deterioration in the quality of data analysis
Result: a wave of subscriptions from ChatGPT Plus ($20/month) and petitions demanding the old functionality be returned.
OpenAI response: concessions and apologies
- Return old models – there will be a “Show outdated models” option in the settings.
- Limit Increase for GPT-5:
- Limit of requests increased to 3000 per week (for Plus subscribers)
- Quotas will soon be increased for all categories of users
“We misjudged the situation. Thank you for your honest feedback,” Altman wrote.
The story OpenAI rolls back changes after massive user dissatisfaction was first published on itzine.ru.
