Orion is supposedly OpenAI’s biggest upgrade in the LLM landscape and is set to launch on the tech giant's second anniversary.
OpenAI is working on its next big revelation in the AI industry. The tech giant may release its flagship Orion AI model by the time it reaches second-anniversary celebrations. The AI model reportedly more advanced than OpenAI’s GPT-4 will be made available for companies it works closely with before rolling out for everyone else.
Orion was recently teased indirectly by Sam Altman, OpenAI’s CEO, in a post on X. The post caption mentioned, “Excited for the winter constellations to rise soon.” When users asked ChatGPT what Sam Altman was hiding in his post, it prompted toward Orion, the most visible constellation in the sky from November to February.
One of the OpenAI executives confirmed the latest AI model under development. The executive said that Orion could be potentially 100 times more powerful than GPT-4. He also hinted that Orion is a separate technology from the o1 reasoning model that OpenAI released in September. However, using o1, the development team created a codename Strawberry to provide synthetic data to train Orion. The research team also threw an open-hour celebration after finishing training the new model, according to a source familiar with the event.
The company aims to combine all LLMs and their training data to create a singular Artificial General Intelligence product equipped for tasks of all types and scales. Open AI plans to release Orion by December as a separate entity rather than launching it as the successor of GPT-4. Earlier, some speculations revolved around online that Orion would be names GPT-5 and will be placed as the next flagship model of the company.
The release of the next model is crucial for OpenAI as it recently closed a $6.6 billion funding round, making it one of the company's historic fundraising initiatives. The new funding also requires OpenAI to transition from a non-profit organization to a for-profit organization. Significant staff turnover is also occurring within the company as its CTO, Mira Murati, just announced her departure. Bob McGrew also resigned from his role as OpenAI’s Chief Research Officer, along with Barret Zoph, VP of post-training.
