Openai eliminates O3 AI model for the next generation of “uniform” ncvrs.com

Openai has actually resigned the O3 edition, which was considered to be the company’s next big AI model, in order to call CEO Sam Altman a “simplified” product range.

THE Enter on X on WednesdayAltman said that in the coming months, Openai will release GPT-5, which will integrate “many (Openai) technologies”, including the O3, AI-powered Chatgpt and API ChatBot platform. As a result of the schedule, Openai no longer plans to release the O3 as an independent model.

The company originally said in December that it plans to launch the O3 sometime earlier this year.

“We want to do a better job to share the planned schedule and a much better job that simplifies our product range,” Altman wrote in the post. “We want AI to“ just work ”for you; We realized how complicated the model and our product range are. We hate the model selector (in Chatgpt) as you and we want to return to Magic Unified Intelligence. ”

Altman also announced that Openai is planning unlimited chat access to the GPT-5 with “standard intelligence setting”, with “abuse thresholds” after the model is generally available. (Altman refused to detail in more detail that this setting and these are the thresholds of the abuse.) It is capable of running the GPT-5 “even higher levels of intelligence”.

“These models incorporate sound, canvas, search, deep research and much more,” Altman said, referring to many features launched in ChatGpt over the past few months. “(A) The most important goal is to combine () models () models by using all the tools, know when to think for a long time, and are generally useful for very wide tasks.”

Before the GPT-5 starts, Openai plans to release the GPT-4.5 “Orion” code, “Orion”, compared to Altman X. -S thought model. “Unlike O3 and Openai’s other so -called reasoning models, non -chain models are generally less reliable in mathematics and physics.

According to the sounds, Openai fully encompasses the trend of the argument model, which he undoubtedly launched with his first argument model, the O1, at the end of last year. The reasoning models are actually checking themselves, which helps to avoid errors that usually take the models. This fact-inspection process comes with some delays-the argument models last for a little longer, usually for a minute to get to the solutions. But they are generally more reliable and capable.

The Chinese AI Lab Deepseek recently drew the world’s attention with the R1 model, which adjusted the O1 to many reference values. Unlike O1, R1 is a “open” model under a permissive license, which is downloadable and usable as developers see it as appropriate.

In recent social media comments, Altman admitted that Deepseek had reduced Openai’s technology leadership in AI and said that Openai would make “some editions” to compete better.