OpenAI has officially stated that it will not release a new AI model codenamed Orion this year. This model was expected to follow the current GPT-4o but is now off the 2024 roadmap.
“We have no plans to release a model codenamed Orion this year. We have a lot of other great technologies we are planning to release,” an OpenAI representative told TechCrunch. Earlier reports suggested that Orion could be OpenAI’s next flagship model, initially planned for release by December. However, instead of launching with ChatGPT, preliminary access would go to trusted partners, with Microsoft—OpenAI’s primary investor—expected to receive access in November.
Orion and Future Plans for GPT Models
Orion is expected to represent a significant improvement over GPT-4o. According to unconfirmed reports, it was trained using synthetic data generated by o1—a neural network developed by OpenAI with reasoning capabilities. OpenAI has expressed intentions to continue working on both the GPT family of models and reasoning networks like o1 in parallel, as they serve different purposes in the company’s ecosystem.
What This Means for OpenAI’s Strategy
Despite the clear announcement, OpenAI’s wording leaves room for speculation, notes NIXSolutions. It is possible that the next flagship model may not be called Orion, or that another, less powerful model could still launch before the end of the year. We’ll keep you updated on any developments or new releases from OpenAI as they become available.