OpenAI releases new o1 models to help ChatGPT answer more complex questions

OpenAI

For many months, there have been rumours swirling around OpenAI releasing a new large language model with codename Strawberry. The company has released the new models under o1 series, which are designed to help ChatGPT answer faster than a human. According to a blog post, the new models can reason through complex tasks and solve harder problems than previous models in science, coding, and math.

OpenAI said that it has trained these models — o1 and o1-mini — to spend more time thinking through problems before they respond, much like a person would. “Through training, they learn to refine their thinking process, try different strategies, and recognise their mistakes,” the company said.

What are the new models capable of?

The company said that the o1 models performs similarly to PhD students on challenging benchmark tasks in physics, chemistry, and biology. “We also found that it excels in math and coding. In a qualifying exam for the International Mathematics Olympiad (IMO), GPT-4o correctly solved only 13% of problems, while the reasoning model scored 83%,” the company revealed. Furthermore, the models’ coding abilities were evaluated in contests and reached the 89th percentile in Codeforces competitions.

The new models, however, don’t have many features for which users turn to ChatGPT for. For example, it can’t browse the web for information and answer questions. Nor can it upload images and files. “For many common cases GPT-4o will be more capable in the near term,” OpenAI noted.

However, for complex reasoning tasks, the new models will come in more handy. “These enhanced reasoning capabilities may be particularly useful if you’re tackling complex problems in science, coding, math, and similar fields,” said OpenAI

OpenAI also said that it is now “resetting the counter back to 1” since these are brand new models and hence it named the series OpenAI o1.

Availability of new models

ChatGPT Plus and Team users will be able to access o1 models in ChatGPT starting today. Both o1-preview and o1-mini can be selected manually in the model picker, and at launch, weekly rate limits will be 30 messages for o1-preview and 50 for o1-mini, said OpenAI. ChatGPT Enterprise and Edu users will get access to both models beginning next week.