OpenAI, the AI company which has been backed by Microsoft, is reportedly planning to launch its latest AI model, which is said to be codenamed 'Orion,' by December 2023, according to The Verge. This anticipated release is expected to mark a significant advancement, as the Orion model is said to be up to 100 times more powerful than GPT-4, OpenAI’s current flagship model.
Orion to debut with partner companies, not ChatGPT
Unlike OpenAI's previous models, Orion will initially be exclusive to partner companies rather than the ChatGPT platform, offering these partners an opportunity to create unique products and features. Microsoft, OpenAI's primary collaborator, is already gearing up to host the model on its Azure cloud platform, which could start as early as November, potentially giving Azure users early access.
Advanced training with synthetic data
According to The Verge, Orion’s training process was completed last month. The model leveraged synthetic data generated by OpenAI's newly introduced o1 reasoning model, which focuses on high-level problem-solving and reasoning capabilities. This training approach positions Orion as an evolution of GPT-4, with the potential for even broader capabilities.
Teasing Orion’s release: OpenAI’s CEO hints at the future
CEO Sam Altman hinted at the model’s name on social platform X (formerly Twitter) with a cryptic post on September 14, saying, “Excited for the winter constellations to rise soon; they are so great.” The post likely referenced Orion, a winter constellation and a subtle nod to the forthcoming AI model.
Will Orion be the Next GPT-5?
OpenAI insiders view Orion as a true successor to GPT-4, and there is speculation it might eventually be named GPT-5. However, OpenAI remains cautious and might even push the release date into 2024.
ALSO READ: Zomato hikes platform fee ahead of Diwali, making online food orders pricier
ALSO READ: Indian YouTube Creators can earn by tagging Flipkart and Myntra products in videos: Know-how
