The chatgpt model picker is back, and it’s complicated

When Openai launched GPT-5 Last week, the company said that the model would simplify the experience of Chatgpt. Openai hoped that the GPT-5 would act as a kind of “one-sized” AI model with a router that would automatically decide the best way to answer users’ questions. The company has declared that this unified approach would eliminate the need for users to navigate its model selector – a long and complicated menu of AI options that Openai CEO Sam Altman publicly said that he hates.

But it seems that GPT-5 is not the unified AI model, Openai hoped that this would be the case.

Altman said in a Publish On Tuesday, the company introduced new “Auto”, “Fast” and “Thinking” parameters for GPT-5 that all chatgpt users can select in the model selector. The automatic parameter seems to work as the GPT-5 model router that Optai initially announced; However, the company also offers users options to get it around, allowing them to directly access AI models in fast and slow response.

In addition to the new GPT-5 modes, Altman said that paid users can again access several inherited AI models-including GPT-4O, GPT-4.1 and O3-which were depreciated last week. GPT-4O is now in the default model selector, while other AI models can be added from chatgpt settings.

“We are working on an update of the personality of GPT-5 which should feel warmer than the current person but not as boring (for most users) as the GPT-4O”, wrote Altman in the post on X. “However, learning for us in recent days is that we really need to go to a world with more personalization by user of the customization of the model.”

The chatgpt model coach now seems as complicated as ever, which suggests that the GPT-5 model router has not satisfied universally satisfied users as the company hoped. The expectations for GPT-5 were Sky High, many hoping that Optai would push the limits of AI models as he did with the launch of GPT-4. However, the deployment of GPT-5 was more rough than expected.

The depreciation of GPT-4O and other models of AI in Chatgpt has triggered a Hinged among users who had attached themselves to the answers and personalities of the models of AI in a way that Optai had not planned. In the future, Altman claims that the company will give users a lot of notice if it ever depreciates GPT-4O.

The GPT-5 model router also seemed to be widely broken on the day of launch. This led some users to think that the AI model was not as efficient as the previous OpenAi models and forced Altman to solve the problem in a AMA session on Reddit. However, it seems that the GPT-5 router is still not satisfactory for all users.

“We are not going to always get everything on Try # 1, but I am very proud of the speed with which the team can iterate,” wrote the vice-president of Openai de Chatgpt, Nick Turley, in a Publish Tuesday.

Routing of prompts to the right AI model is a difficult task that requires align an AI model on a user’s preferences, as well as the specific question they ask. The router must then make a decision on which the IA model to send the prompt in a fraction of a second – in this way, if a prompt goes to a fast responding AI model, the answer can always be fast.

More broadly, some people have preferences for AI models that go beyond quick or slow responses. Some users may like the verbosity of an AI model, while others can appreciate the contrary responses 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 recently held the funeral For the anthropic AI model, Claude 3.5 Sonnet, when it was put out of line. In other cases, AI chatbots seem to contribute to mentally unstable people who descend Psychotic rabbit holes.

It seems that Optai has more work to do to align its AI models on the preferences of individual users.