Openai launched its major model GPT-5 a few weeks ago. While everyone was expecting the new model that it would be much better than the previous, users are highlighting its flaws. But why did users criticize GPT-5? The main reason is that users realized that the new model is not able to be as GPT-4O in real-life conversation. To close the GPT-4o, Kadam shook all users who are using it for emotional support and connection. Yes, it is as problematic as it seems. But Openai believes in giving what its users want. Earlier, it brought back the GPT-4o and now, updating Openai GPT-5 to make it more favorable and heated.

Taking X (aka Twitter), OpenI recently posted that the company is working on the latest model to improve its personality. The post said, “We are making GPT -5 warmer and friendly on the basis of feedback which was formally felt earlier. Changes are subtle, but the chatgip should feel more acceptable now,” the post said.

But how will it change? You will look at small, real touch like “good questions” or “great start”, not flatter. Internal tests show no increase in sycophancy compared to previous GPT-5 personality, “said this. “The change may take up to a day to roll out, soon update more.”

GPT-5 vs. GPT 4O: Which is a hot personality?

The entire saga was closed when GPT-5 officially replaced old models, only to complain immediately to users that it could be clever, it was also less preferred. GPT-5 is billed as more reliable with “PHD-level” skills in coding, mathematics, writing and science. Even it comes with alternative personality settings-Cynic, Robot, Listener and NERD-but no one seemed to win over long-time fans of GPT-4o.

Addressing the situation, Openai CEO Sam Altman had earlier said that GPT-5 is now receiving complaints about it being favorable, but now it changes. In a post, Altman said, “Now you can choose between” Auto “,” Fast “and” Thinking “for GPT -5. Most users want autos, but additional control will be useful for some people.” Adding this, he said, “Rate limitations are now 3,000 messages/week with GPT -5 thinking, and then additional capacity on GPT -5 Thinking Mini after that limit. Reference limit for GPT -5 thinking is 196k tokens. We may have to update the rate limit over time over time.”

“Yes man” problem

A part of the GPT-4o’s appeal, the irony was that the very specialty is that Openai wanted to approve: his tendency to act like a smooth cheerleader. In April, the company admitted that the GPT-4O had become “excessive flattery or agree” and even “dissatisfied”.

But in a podcast recorded before the launch of GPT -5, Altman admitted that many users had begged to return to that behavior. “Some people had never supported anyone before,” he said on heavy conversation, given that GPT-4o’s “yes man” quality, while flawed, clearly matters something.

Considering a stormy rollout, Altman said that the level of emotional connection to the AI model feels that it is contrary to anything seen with previous technology. He said, “It seems differently and strong that people have posted on X, compared to the types of attachment for previous types of technology. “Suddenly demonstrating the old model that users depended in their workflows were a mistake.”

He said that while most users can distinguish reality from imagination, “cannot do a small percentage.” Balancing the user freedom with the responsibility of the company, he said, a “main principle” remains because Openai continues to roll out the new technology.