Is OpenAI Orion the next big thing in AI technology?

According to reports, the Microsoft-backed AI company OpenAI has been preparing to launch its next major model, codenamed Orion, by the end of this year. Unlike previous models such as GPT-4o and the recently launched o1-series, which focus on reasoning capabilities, Orion will initially be exclusive to OpenAI’s partner companies, giving them a head start in building innovative products and features. ChatGPT users, therefore, may not see Orion integrated into the chatbot until later.

The Verge has reported that Microsoft, OpenAI’s primary partner for model deployment, is gearing up to host Orion on its Azure cloud platform, with hosting preparations likely to be complete as early as next month. The move underscores Microsoft’s strategic interest in being a frontrunner in next-gen AI infrastructure, particularly with a model that could surpass anything OpenAI has previously released.

How does Orion compare to past AI models?

OpenAI insiders revealed that Orion is projected to be up to 100 times more powerful than the current GPT-4 model, with capabilities that separate it from the reasoning-focused o1 model. In an ambitious training process, OpenAI reportedly leveraged synthetic data generated by the o1 model, which has significantly accelerated Orion’s development.

The company finished the model’s training just last month, and CEO Sam Altman subtly hinted at the name in a recent X post, referring to the “winter constellations”—a probable nod to Orion, one of the most prominent star constellations.

OpenAI reportedly views Orion as the true successor to GPT-4 and possibly the company’s next generational leap in AI modeling. However, it remains uncertain if OpenAI will officially label it GPT-5 or keep the Orion moniker. There’s also a chance that the release could be postponed into next year, allowing more time for finalization and refinement.