OpenAI is set to introduce its latest AI model, code-named Orion, by December, sources have revealed.

This introduction coincides with the second anniversary of ChatGPT.

Orion will initially be accessible to companies working closely with OpenAI. This is different from previous models, which were available through ChatGPT.

Microsoft’s engineering team plans to host Orion on Azure, potentially starting as soon as November.

This new model is expected to represent a substantial advancement over GPT-4o and o1. However, it is unclear if Orion will be named GPT-5.

The model’s development has been supported by synthetic data from OpenAI’s o1 model, which is also referred to as Strawberry.