OpenAI Unveils New AI Model ‘o1’ Capable of ‘Reasoning’, Integrated into ChatGPT

▶ “Step-by-step Problem Solving”

OpenAI, the developer of ChatGPT, has launched a new version of ChatGPT equipped with reasoning abilities. The new model, ‘OpenAI o1’ (also known as ‘o1’), is integrated into this new ChatGPT.

According to OpenAI, the new chatbot is based on ‘o1’, which can ‘reason’ through tasks related to mathematics, coding, and coding-related work. This model enables the chatbot to think, analyze in English, and find the right angles to present the best solution.

Jakub Pachocki, OpenAI’s Chief Scientist, explained that “previous models like ChatGPT would start responding immediately after receiving a question, but this model may take some time. It thinks, analyzes in English, and finds the right angles to present the best solution.”

Although it may take longer to provide an answer than previous models, it solves complex problems through a step-by-step thought process. This new model could mitigate issues such as struggling with simple math problems, being prone to bugs, and generating incorrect or fabricated answers.

OpenAI stated that this model achieved an 83% accuracy rate in the International Mathematical Olympiad (IMO) preliminary exam, compared to the 13% accuracy rate of previous models. Additionally, this model could assist physicists in formulating complex mathematical equations and support medical researchers in their experiments.

This new model is part of OpenAI’s effort to develop Artificial General Intelligence (AGI), AI at a human level. Along with the basic ‘o1’ model, OpenAI also unveiled a smaller version, ‘o1-mini’. The ‘o1’ model provides text-based answers and does not generate images or videos.

ChatGPT Plus and Chat Teams subscribers, as well as businesses, will be able to access this new technology starting today.