OpenAI to Unveil Next Gen Orion AI Model by Close of 2024, Reports Show
OpenAI is planning to unveil a new AI model dubbed Orion by December 2024, The Verge has reported. However, the AI startup’s rollup plan for the new model is different.
Unlike GPT-4o, OpenAI’s Orion AI model will not be rolled out through the ChatGPT platform. Instead, OpenAI will give the companies it works with access to the new model first and allow them to create their own products and features.
Powerful than GPT-4o
An executive at OpenAI has said the new AI model will be more powerful than GPT-4o. In a post shared on X social media platform, the executive said Orion could have up to 100 times more computational power than GPT-4o.
Although the Orion AI model uses similar resources as GPT-4o, it comes with improved efficiency and architecture. The Orion AI model is different from the o1 reasoning model that OpenAI unveiled in September 2024.
The o1, which is code-named Strawberry, is capable of reasoning through difficult problems logically and can solve problems that defeat GPT-4o. There are reports that OpenAI trained the new Orion AI model using synthetic data derived from the Strawberry. The AI startup completed the training in September 2024.
OpenAI CEO, Sam Altman hinted at the new model the same month when he shared a cryptic post on the X social media platform saying, “Excited for the winter constellations to rise soon; they are so great.”
Altman’s post was most likely a subtle nod to the upcoming AI model. Orion is one of the reasoning models that OpenAI has trained to respond to complex questions faster than humans. OpenAI’s long term goal is to consolidate its large language models to build a more capable model that would eventually be known as artificial general intelligence (AGI).
Microsoft’s Orion AI Model Plan
There are speculations that OpenAI’s new AI model may be hosted on Microsoft’s Azure cloud platform. Sources close to the company told The Verge that Microsoft engineers are already getting ready to host Orion on the Azure platform as early as November.
If this happens, Azure users will have an opportunity to explore the advanced AI model features before its rollout. Microsoft is OpenAI’s strategic partner for deploying AI models.
OpenAI insiders see Orion as the AI model that will succeed GPT-4. Microsoft already integrated the GPT-4o into Azure. In June 2024, Microsoft and OpenAI piloted a travel virtual agent that leverages GPT-4o to enable users to have video and voice interactions with customers.
It’s still unclear whether the AI startup will refer to the new model as GPT-5 externally. The schedule for releasing the new big AI model could change or slip, but neither Microsoft nor OpenAI has commented on the developments.
The Transition Challenge
The report about the release of OpenAI’s next generation model comes at a time when the AI startup is undergoing restructuring to become a for-profit company. The restructuring process saw CEO Altman receive a $150 million equity stake in the company.
OpenAI has also been experiencing significant leadership changes that add complexity to its evolving strategy. In September 2024, its Chief Technology Officer Mira Murati left the AI startup. At the time, President Greg Brockman was on temporary leave.
Bob McGrew, who served as the Chief Research Officer and Vice President for Post Training Barret Zoph, also announced their departure from OpenAI.
Focus on Growth
But even with these transitions, OpenAI has made significant progress in the month of October 2024. The AI startup has expanded its liquidity to $10 billion after it secured $6.6 billion in funding from investors and a $4B line of credit from leading banks that include Citi, JPMorgan Chase, Morgan Stanley, and Goldman Sachs.
OpenAI’s valuation also rose to an impressive $157 billion, effectively positioning the company among the most valuable private entities globally. With the latest funding, OpenAI has put its global expansion plan in motion.
Earlier this month, the company announced plans to open new offices in five cities across five cities across the US, Europe, and Asia. OpenAI plans to make Singapore its global operations hub and is set to open its offices there by the end of the year.