Photo: Jamesonwu1972 / Shutterstock.com
OpenAI is set to introduce Orion, its latest high-powered AI model, by December this year, with early access anticipated in November for key collaborators. The rollout is poised to begin exclusively with select companies, granting them first access to integrate Orion into their products and services.
However, unlike the previous launches of models like GPT-4o and o1, Orion won’t immediately be available to the public via ChatGPT, reports The Verge.
Microsoft, OpenAI’s primary partner for model deployment, is reportedly preparing to host Orion on its Azure cloud platform as early as next month. Although OpenAI views Orion as a successor to GPT-4, it remains unclear if the company will officially launch the GPT-5 model upon release.
With OpenAI plans historically susceptible to change, the company could still adjust this timeline.
Hints from OpenAI executives suggest that Orion could be up to 100 times more powerful than GPT-4, an exponential leap that signals the company’s ongoing quest toward developing artificial general intelligence (AGI).
Orion also marks a departure from o1, OpenaAI’s previous model, which focused on reasoning abilities and debuted in September. According to those familiar with OpenAI’s roadmap, the long-term vision is to eventually unify its language models into one highly capable system.
OpenAI reportedly utilised synthetic data from o1 (internally dubbed ‘Strawberry’) to train Orion. The Verge also reports that the completion of Orion’s training was celebrated in September at an OpenAI gathering.
Further stoking curiosity about Orion, OpenAI CEO Sam Altman recently posted on X about his excitement for the “winter constellations to rise soon.” Orion is a prominent winter constellation visible from November through February.
The company has transitioned to a for-profit model to meet the ambitious performance expectations.
In the News: Change Healthcare hack exposed data of 100 million Americans
