OpenAI plans to release new GPT-4.5 model in coming weeks, says Altman

By Rachel Metz  OpenAI is nearing the release of a new artificial intelligence model called GPT-4.5 after hitting stumbling blocks in developing the AI system last year.  

In a series of posts on social network X on Wednesday, OpenAI Chief Executive Officer Sam Altman said the model, which the company has referred to as Orion internally, would come in a matter of “weeks.” Bloomberg News reported in November that Orion had not been hitting OpenAI’s desired performance.

 

The release of GPT-4.5 will mark the end of an era of sorts for OpenAI. Altman said the model will be the last that the company introduces that doesn’t use additional computing power to mimic the ways humans reason. OpenAI has embraced the reasoning approach with some of its newer models, including o1 and o3.

 

Altman said in the post that OpenAI will move to combine its GPT models, which powered the original ChatGPT chatbot, with its newer o-series of models to build AI systems that can automatically determine how long to ruminate over a query before spitting out an answer. The goal, he said, is to simplify the experience so users aren’t forced to choose from an increasingly complicated list of options. 

 

“We want AI to ‘just work’ for you; we realize how complicated our model and product offerings have gotten,” Altman wrote. “We hate the model picker as much as you do.”

 

Altman also said OpenAI is moving forward with a GPT-5 model for later this year that will pull together much of the company’s AI technology, including o3. The company had released a smaller version of that system called o3-mini, but hadn’t yet launched the full model.

 

An OpenAI spokesperson declined to comment beyond Altman’s statements.