It seems OpenAI is preparing to launch its next big AI model this December. The AI giant is expected to unveil a model codenamed “Orion”.

OpenAI, one of the world’s leading AI research firms, is gearing up to release its new generation AI model, Orion, in December. This significant launch coincides with the anniversary of ChatGPT’s debut two years ago.

Reports suggest that Orion will initially be available only to select OpenAI partner companies rather than the general public. Orion is expected to be hosted on Microsoft’s Azure platform, with preparations slated for completion in November.

A Successor to GPT-4?

While Orion is viewed as a successor to GPT-4, it is yet to be confirmed whether it will be publicly marketed as GPT-5. According to OpenAI’s Japan President, Tadao Nagasaki, Orion could be 100 times more powerful than GPT-4. Unlike the logic model codenamed “Strawberry” released in September, this new model takes another step toward artificial general intelligence (AGI). OpenAI's ultimate goal is to bring together various language models to create a more versatile and capable AI system over time.

Previous reports indicated that Orion used OpenAI’s o1 model to generate synthetic data during its training. A source within the company disclosed that the new model’s training process was completed in September. Sam Altman, CEO of OpenAI, hinted at Orion’s development on the social media platform X by expressing his excitement for the upcoming “rise of winter constellations.”