OpenAI Restores GPT-4o After User Criticism

• GPT-4o has been restored to ChatGPT following user criticism.
• OpenAI has introduced new settings for GPT-5.
• Users can choose between several working modes.

This is reported by Ars Technica, ZDNet, ZDNet.

On August 13, 2025, OpenAI's CEO Sam Altman announced the restoration of the GPT-4o model to ChatGPT after significant backlash from users who expressed dissatisfaction due to its removal during the launch of GPT-5.

The restoration of GPT-4o was a reaction to numerous complaints, as OpenAI simplified its proposal, removing all previous models during the launch of GPT-5 on August 7. This was one of the largest protests in ChatGPT's history, where users voiced their discontent on social media.

In addition to restoring GPT-4o, OpenAI has also introduced new settings, such as increasing the limit on responses for the GPT-5 Thinking mode from 200 to 3000 responses per week, as well as new routing options: "Auto", "Fast", and "Think". This allows users to better control which variant of GPT-5 processes their requests.

For professional users who pay $200 per month, Altman confirmed that additional models, such as o3, 4.1, and GPT-5 Thinking mini, will be accessible via the "Show Additional Models" toggle in ChatGPT settings.

Style Adjustment for GPT-5 Output

OpenAI also acknowledged widespread complaints about the output style of GPT-5, which many users described as "robotic" compared to the more conversational tone of GPT-4o. Altman noted that the company is working on updating the personality of GPT-5 to make it warmer.

Problems with the launch of GPT-5 were not only related to the removal of models. The automatic routing system, designed for selecting appropriate model variants, failed on launch day, consistently defaulting to a less powerful version.

While GPT-4o has been restored, OpenAI continues to develop GPT-5, emphasizing that limits may require further adjustments depending on user feedback.

Tags: Technology/AI