OpenAI has introduced a new series of OpenAI o1 AI models: OpenAI o1 and OpenAI o1-mini.
They can reason independently and solve more complex problems than previous models in the fields of science, coding, and mathematics. According to the company’s tests, o1 is almost 6 times smarter than GPT-4o.
We trained these models to spend more time thinking through problems before they react like a human would. Through training, they learn to refine their thought process, try different strategies, and admit their mistakes.
In our tests, the next iteration of the model performs similarly to graduate students on difficult benchmark problems in physics, chemistry, and biology. We also found that it excels at math and coding. On the International Mathematical Olympiad (IMO) qualifying exam, GPT-4o got only 13% of the problems right, while the reasoning model scored 83%.
— OpenAI
For now, the new models exist as preview versions. There is still a lot of work to be done on them.
However, ChatGPT Plus and Team users will have access to o1-preview and o1-mini starting today. Educational institutions will be given access next week. o1-mini will be available to everyone for free, but there is no release date yet. (OpenAI)
Source: www.iphones.ru
