The next model of OpenAI, 100 times more powerful than GPT-4

According to The Verge, OpenAI plans to release Orion, its next frontier model, by December.

The Verge has claimed that, unlike previous OpenAI models, the company plans to release Orion to closely-partnered companies so they can creatively build their own products instead of making it widely available through ChatGPT.

According to engineers at Microsoft, OpenAI’s main partner, the company is preparing to host Orion on Microsoft Azure in early November. Regarding the naming, although Orion is seen in OpenAI as a successor to GPT-4, it is not clear whether the company will call it GPT-5 or not.

However, as always, the release plan is subject to change. The Verge says Microsoft declined to comment for this story, and OpenAI declined to comment on the claim.

After company CEO Sam Altman called the story “fake news,” OpenAI spokesman Nico Felix told The Verge that the company has no plans to release a model codenamed Orion this year; But they intend to offer amazing technologies.

According to one of the managers of OpenAI, the Orion model is a hundred times more powerful than GPT-4; It was previously reported that OpenAI uses o1, codenamed “Strawberry”, to train Orion.

Previously, Sam Altman, in a post on the X social network, wrote, “I love staying home, especially when I’m in the Midwest; The night sky is very beautiful. “I’m excited to see the Winter Circle, they’re fantastic.”

Sam Altman’s post coincides with news that OpenAI is targeting September for its event.

Interestingly, if you ask ChatGPT o1-preview what Sam Altman’s post is hiding, he will tell you that he is referring to the word Orion, a constellation that appears in the night sky in winter.

If the Orion unveiling happens, it will be a critical event for OpenAI; Because this company has recently struggled with the relocation of its employees and has attracted 6.6 billion dollars in capital.