OpenAI is planning to launch its next flagship artificial intellience (AI) model called Orion which has been teased as a hundred times more powerful than GPT-4, according to a report by The Verge which cited unnamed sources.
However, the AI giant will not release it through ChatGPT, but provide Orion access first to companies it closely works with, unlike GPT-4o and o1, according to the report which added that this is for the companies to use it to build their own products and features.
Orion may also be hosted on Microsoft Azure as early as November, according to the report.
Though OpenAI sees Orion as the successor to GPT-4, it remains unclear if it may be called GPT-5 externally.
OpenAI’s longer term goal is to combine its large lanuage models (LLMs) over time to create artificial general intelligence (AGI), which is an even more capable model.
Also Read: Onion prices to remain high in Diwali due to heavy rains, damaged crops, delayed harvest
CEO Sam Altman had also cryptically teased this earlier through a post on X saying he was “excited for the winter constellations to rise soon.”
A question to ChatGPT on what Altman’s post is hiding elicits a response stating he’s hinting at the word Orion, the winter constellation most visible in the night sky from November to February, according to the report.
OpenAI was also reported to have been using o1, codenamed Strawberry, to train Orion using synthetic data.
The release of Orion comes at a crucial time for the AI giant which has just closed a $6.6 billion funding round requiring the company to restructure itself as a for-profit entity.
Also Read: AI may be used to check traffic violations in new proposal, Nitin Gadkari says