OpenAI is preparing to launch its next breakthrough in AI, the Orion AI model, by December. Unlike previous releases like GPT-4o and o1, OpenAI won’t introduce Orion AI through ChatGPT initially. Instead, the company will offer early access to select partner companies, allowing them to integrate it into their own products and features.
A Strategic Shift for OpenAI
While Orion AI model is considered the successor to GPT-4, it’s unclear if OpenAI will officially name it GPT-5. Engineers at Microsoft, OpenAI’s key partner, are already preparing to host the Orion AI model on Azure by November. However, like many tech releases, the timeline might shift.
The Orion AI model is expected to be up to 100 times more powerful than GPT-4. It represents a leap forward from the o1 reasoning model, introduced in September. Over time, OpenAI aims to combine its large language models to create an even more advanced system that could potentially achieve artificial general intelligence (AGI).
Training Orion AI
In September, OpenAI completed the training of the Orion AI model, marking a significant milestone. According to sources, OpenAI used o1, codenamed “Strawberry,” to supply synthetic data for training the AI model. This coincides with OpenAI CEO Sam Altman’s cryptic hint on X about being excited for “winter constellations,” referring to the Orion constellation most visible in winter months.
Critical Timing for OpenAI’s Growth
The launch of the Orion AI is significant as OpenAI shifts to a for-profit structure, following its historic $6.6 billion funding round. Key executives, including CTO Mira Murati, chief research officer Bob McGrew, and VP of post-training Barret Zoph, have announced their departures, but OpenAI is moving forward with its mission.
