OpenAI has introduced o1, a new and improved AI model designed to tackle more complex tasks through enhancing reasoning capabilities.
O1 is the rumoured “Project Strawberry” that OpenAI has been quietly working on for some time and was expected to launch in the Fall.
The first versions of O1, o1-preview and o1-mini, will be released in preview to ChatGPT Plus and Team subscribers, with users limited to 30 and 50 messages, respectively.
Subscribe today for free
“We trained these models to spend more time thinking through problems before they respond, much like a person would,” an OpenAI announcement read “Through training, they learn to refine their thinking process, try different strategies, and recognize their mistakes.”
AI models often produce poor outputs, known as hallucinations, as they’re not trained to think how to produce a response to a query, instead focusing on predicting the next likely word or sentence.
OpenAI’s o1 is designed to think more carefully about how it responds, which in turn, creates more accurate outputs.
OpenAI said that the o1 models can reason through complex tasks and handle queries related to science and maths.
The Microsoft-backed startup suggested o1 can be used by healthcare researchers to annotate cell sequencing data or help physicists generate complicated mathematical formulas needed for quantum optics.
The models are also adept at accurately generating and debugging complex code.
Beyond its initial launch, o1-preview and o1-mini will be made available to ChatGPT Enterprise and Edu users next week.
They’ll also come to OpenAI’s API, enabling developers to start prototyping with both models with a rate limit of 20 RPM.
OpenAI said it also plans to bring o1-mini to ChatGPT Free users but offered no time frame.
The previews of o1 and o1-mini are just the start — with OpenAI planning to add browsing, file and image uploading, and other features to make the models “more useful to everyone.”
RELATED STORIES
OpenAI vs xAI: The battle for AI supremacy
OpenAI and News Corp partnership marks 'significant evolution' for AI and news
