OpenAI said it does not intend to release a new artificial intelligence model, codenamed Orion, this year. It is expected to be a continuation of the current GPT-4o.
Image source: Mariia Shalabaieva / unsplash.com
«We have no plans to release a model codenamed Orion this year. We plan to release many other great technologies,” an OpenAI representative told TechCrunch. Earlier, media reported that Orion, which is expected to be OpenAI’s new flagship, will be released by December. At the same time, it will debut not with the ChatGPT chatbot, but with the company’s trusted partners – they will get preliminary access to it first. Microsoft, as OpenAI’s main investor, expects to gain access to it as early as November.
Orion is a step up from OpenAI’s current flagship GPT-4o. The model, according to anecdotal evidence, was trained on synthetic data from o1, a neural network created by OpenAI that can reason. In the foreseeable future, the company intends to continue developing new models of the GPT family along with reasoning neural networks like o1 – they will exist in parallel, because they are intended for fundamentally different working scenarios.
However, the statement made by OpenAI leaves it room for maneuver. The company’s next flagship may not actually be the Orion. Or by December OpenAI will still release a new model, but it will be less powerful than Orion.