ChatGPT’s model picker is back, and it’s complicated

When OpenAi launched GPT-5 last week, the company said that the model would simplify the Chatgpt experience. OpenAI hoped that GPT-5 would act as a kind of “one size fits all” AI model with a router that would automatically decide how they can best answer users’ questions. The company said that this uniform approach would eliminate the need for users to navigate in the model picker – a long, complicated menu of AI options that OpenAi CEO Sam Altman has publicly said he hate.

But it seems that GPT-5 is not the uniform AI model that Openai hoped it would be.

Said Altman in one Post on X Tuesday that the company introduced new “car”, “fast” and “thinking” settings for GPT-5 that all chatgpt users can choose from the model selector. The automatic setting seems to work as the model router of GPT-5 that Openai initially announced; However, the company also gives users options to bypass it, giving them direct access to fast and slowly reacting AI models.

In addition to the new GPT-5 modes, Altman said that paid users have access to various Legacy AI models-including GPT-4O, GPT-4.1 and O3 that were abolished last week. GPT-4O is now standard in the model selector, while other AI models can be added from the Chatgpt settings.

“We are working on an update of the personality of GPT-5 that should feel warmer than the current personality, but not as annoying (for most users) as GPT-4O,” Altman wrote in the post on X. “However, we learn one for us from the past days, however, that we really have to go to a world with more per-heer modernity.”

Chatgpt’s model selector now seems to be just as complicated as always, which suggests that the model router of GPT-5 is not universally satisfied with users like the company hoped. The expectations for GPT-5 were heavenly, with many hope that OpenAI would push the boundaries of AI models as it had with the launch of GPT-4. However, the roll-out of GPT-5 is rougher than expected.

The depreciation of GPT-4O and other AI models in Chatgpt led to one Backlash among users who was attached to the answers and personalities of the AI models in ways that OpenAi did not expect. In the future, Altman says that the company will give users sufficient advance if the GPT-4O will ever write off.

The model router of GPT-5 also seemed to be largely broken on the launch day. That made sure that some users had the feeling that the AI model was not as performance as earlier OpenAi models, and Altman forced the problem to tackle Reddit in an AMA session. However, it seems that the GPT-5 router may still not be satisfactory for all users.

“We are not always going to get everything on Try #1, but I am very proud of how quickly the team can repeat,” wrote OpenAi’s VP from Chatgpt, Nick Turley, in one Post on X Tuesday.

Routing instructions to the right AI model is a difficult task that requires an AI model to be tailored to the preferences of a user, as well as the specific question they ask. The router must then make a decision about which AI model to send the prompt on just a fraction of a second -in this way, if a promptly responding AI model goes, the reaction can still be fast.

More generally, some people show preferences for AI models that go beyond fast or slow reactions. Some users may like the Verbosity of the one AI model, while others appreciate the contrary answers of another.

Human attachment to certain AI models is a relatively new concept that is not well understood. For example, hundreds of people in San Francisco held a funeral For the AI model of Anthropic, Claude 3.5 Sonnet, when it was taken offline. In other cases, AI chatbots seem to contribute to mentally unstable people who go down Psychotic rabbit holes.

It seems that OpenAi has more work to do its AI models to individual user preferences.