OpenAI Destroys ChatGPT-4o Again.
https://enterprise.shutterstock.com/image-photo/openai-logo-displayed-on-smartphone-screen-2520388517 or https://enterprise.shutterstock.com/image-photo/chatgpt-logo-displayed-on-smartphone-screen-2520385879
Last August, OpenAI, the developers of ChatGPT, unceremoniously discontinued support for the popular GPT-4o model, but, after succumbing to complaints, reinstated it a week later. Now, the company is making a second attempt to persuade users to abandon GPT-4o. In a new post on its website, OpenAI announced it will once again deprecate this model.
This model is scheduled to disappear from the ChatGPT model list on February 13, along with other older models such as GPT-4.1, GPT-4.1 mini, and OpenAI o4-Mini. OpenAI is clearly concerned about this decision.
“While this announcement covers several older models,” OpenAI wrote, “GPT-4o deserves special attention.” The company says it took user dissatisfaction with 4o’s initial deprecation into account when developing its newest models, GPT-5.1 and GPT-5.2 , and designed these models to preserve the features most favored by fans of the older model. The company claims that GPT-4o is now used by “only 0.1% of users” daily.
Thus, the company wants to focus on “improving the models most people use today,” which presumably means deleting older models. “We understand that losing access to GPT-4o will be disappointing for some users, and we did not make this decision lightly,” the statement reads.
What is special about GPT-4o?
So why is OpenAI so cautious about its users, especially given that GPT-4o is several generations behind, and newer models exist that claim to do the same thing better? When GPT-4o was first deprecated, people were unhappy . Users called its successor, GPT-5, a ” complete disaster ” and accused OpenAI of ” the biggest hoax in the history of AI.”
Some criticized the model’s usefulness, claiming it gave wrong answers and broke code , but perhaps most strikingly, people pointed to its more concise tone.
GPT-4o was called ” sycophantic ” by critics, a fact the company acknowledged , stating its intention to address in future updates. But apparently, one person’s “yes-man” is another person’s “active listener.” When the company initially removed GPT-4o, users complained that its replacement was cold and less like a ” friend .” Even OpenAI acknowledged this, stating in a post today that users “preferred GPT-4o’s conversational style and warmth.”
In short, in the words of the Model 40’s supporters themselves, they ” mourned ” the model.
Is GPT-5.2 a good replacement for 4o?
However, given that many users seem to have already abandoned version 40, OpenAI’s decision seems understandable at first glance. Personally, I’m put off by AI because of how much soothing filler text is used to make most responses seem appealing (“you’re absolutely right,” and the like), apparently just to make me feel better. Shorter, more concise responses would be less off-putting to me.
To find a compromise, OpenAI redesigned the personalization feature in GPT-5.1 so users can simply choose how the chatbot interacts with them. There are options for more professional responses, more complex ones, more effective ones, and, for those who prefer an active listening style, friendlier ones.
Judging by OpenAI’s numbers, that was enough for most people, but some are still expressing dissatisfaction with the company’s new announcement.
There are still some staunch GPT-4o supporters.
In a Reddit thread dedicated to OpenAI’s new announcements, users questioned the accuracy of the 0.1% figure for GPT-4o, stating that requests ” are redirected to 5.2 regardless ” and that ” something somewhere in their calculations doesn’t add up.” Others pointed out that free users can’t use GPT-4o and that it isn’t enabled by default , which naturally inflates the numbers in its favor.
In light of this, calls to unsubscribe from ChatGPT are once again circulating among 4o’s most devoted fans. In a popular thread on the OpenAI subreddit, one user called 4o “OpenAI’s most advanced and beloved model,” praising its “personality, warmth, and stability,” stating that its fans have built long-term projects and “habits of emotional support” around it, and that its sudden loss, without even the option to return to the old regime, “feels jarring and deeply disappointing.”
“This isn’t about resisting innovation,” the statement reads. “It’s about respecting the connections users have established with specific models.”
Time will tell whether the fan protest will work again. However, since ChatGPT CEO Nick Turley has previously been skeptical of such connections , and since maintaining old models likely takes away developer resources that could be used to create new ones, I wouldn’t count on it.