OpenAI’s Next Frontier: Orion AI Model Set for December Release, Promising Unprecedented Power

According to sources familiar with the plan, OpenAI is gearing up to release its next frontier AI model, codenamed Orion, by December. Unlike previous OpenAI models, such as GPT-4o and o1, Orion will not initially be available to the general public through ChatGPT. Instead, OpenAI plans to offer early access to select partner companies to allow them to integrate the model into their products and develop new features, marking a departure from the company’s previous broad-release approach.

OpenAI To Launch New AI Model Orion By December: What You Need To Know | Times Now

Microsoft, a major partner of OpenAI in deploying AI models, is expected to host Orion on its Azure platform as early as November. This access would allow Microsoft engineers to begin working with Orion before the broader rollout. However, it is still unclear whether OpenAI will officially name Orion as “GPT-5” or choose a different title. Internally, OpenAI sees Orion as the successor to GPT-4, and the model is anticipated to be up to 100 times more potent than GPT-4. OpenAI has kept its branding plans private, and while Microsoft declined to comment, OpenAI’s spokesperson Niko Felix shared that “we do plan to release a lot of other great technology.” However, he denied that a model named Orion would be released this year.

Hints about Orion’s development have circulated for several months. OpenAI reportedly used its o1 model, “Strawberry,” to generate synthetic data during Orion’s training. A source familiar with the development process revealed that OpenAI celebrated completing Orion’s training in September, shortly before CEO Sam Altman posted a cryptic message on X (formerly Twitter), saying, “excited for the winter constellations to rise soon.” Many took this post as a subtle hint about Orion, the most visible winter constellation from November to February. Additionally, users of the o1-preview model on ChatGPT noted that Altman’s mention of winter constellations could be connected to the upcoming Orion release.

The Verge also reported that Orion’s gradual release marks a significant time for OpenAI, which recently closed a $6.6 billion funding round, elevating the company’s valuation to $157 billion. As part of its new funding structure, OpenAI is transitioning into a more for-profit entity, which has placed additional pressure on the company to continue producing increasingly advanced models to maintain its leadership in AI. This latest funding round has also coincided with substantial staff turnover at OpenAI, with CTO Mira Murati, chief research officer Bob McGrew, and VP of post-training Barret Zoph all announcing their departures in recent months.

Verge’s report on Orion’s development journey pointed out that there is still much speculation about the model’s capabilities. An OpenAI executive previously teased that Orion could be up to 100 times more powerful than GPT-4, positioning it as a significant advancement over current models. OpenAI aims to integrate its language models into a unified, competent model that could eventually achieve artificial general intelligence (AGI).