OpenAI to Celebrate ChatGPT's 2nd Anniversary with Next Flagship Model 'Orion' Reports suggest that OpenAI has been using o1, codenamed Strawberry, to train its model Orion with synthetic data
You're reading Entrepreneur India, an international franchise of Entrepreneur Media.
On September 14 2024, OpenAI CEO Sam Atlman took to microblogging platfrom X to share a cryptic post.
"Excited for the winter constellations to rise soon; they are so great," the post also read.
Many speculated that the post was related to Orion, the organisation's new Artificial Intelligence (AI) model.
On October 25, The Verge confirmed the speculations stating that Orion was expected to be a hundred times more powerful than GPT-4.
Notably, unlike GPT-4o and o1, Orion will not be released widely in the public. Rather, OpenAI will deploy it first to the companies its working closely with, said a source familiar with the plan. Microsoft is speculated to host Orion on Azure as early as November.
Reports suggest that OpenAI has been using o1, codenamed Strawberry, to train its model Orion with synthetic data.
The long term goal is to combine its Large Language Model (LLMs) to create a more capable model that could eventually be termed artificial general intelligence (AGI).
The news of Orion follows OpenAI's histroic funding of USD 6.6 billion which requires the company to restructure itself as a for-profit entity. On the leadership front, the it has seen departure of its significant talents including CTO Mira Murati, chief research officer Bob McGrew and VP of post training Barret Zoph.