OpenAI has clarified that it does not plan to release an AI model known as Orion in 2024, addressing recent speculation around its product development roadmap.
A spokesperson for OpenAI stated via email to TechCrunch, “We don’t have plans to release a model code-named Orion this year. We do plan to release a lot of other great technology.” This statement follows a report by The Verge, which indicated that Orion, expected to be OpenAI’s next advanced model, would be available by December, with select partners previewing it through ChatGPT. According to The Verge’s sources, Microsoft—an OpenAI collaborator and investor—was expected to access Orion as early as November.
OpenAI previously informed TechCrunch that The Verge’s report was inaccurate but did not provide additional details. Orion, as described in the report, is expected to expand upon OpenAI’s current GPT-4 model and is partially trained on synthetic data from o1, OpenAI's “reasoning” model. OpenAI has indicated plans to continue evolving both its GPT models and reasoning models like o1, which serve distinct functions.
The company’s statement allows for some interpretation. While Orion may not launch this year, OpenAI could potentially introduce a new model that does not match Orion’s anticipated capabilities by December.
