- OpenAI debuts o1-preview models, advancing AI in solving science, math, and coding problems.
- O1-preview AI excels in reasoning, surpassing PhD students in physics, chemistry, and more.
- OpenAI introduces o1-mini, an affordable model for developers with advanced coding needs.
OpenAI presented the long-awaited o1-preview AI models with the support of its CEO Sam Altman. Such systems are a step up in the development of artificial intelligence particularly in solving challenging issues in science, coding, and mathematics. The o1-preview models are now integrated with ChatGPT and API and are open for early adopters to test its functionalities.
Ongoing Updates Planned
OpenAI plans ongoing updates that will continually improve the models. On Thursday, CEO Sam Altman shared his optimism in a post on X, expressing enthusiasm for what these systems may achieve. He considered this release critical for advancing AI concerning more advanced multi-step tasks.
here is o1, a series of our most capable and aligned models yet:https://t.co/yzZGNN8HvD
— Sam Altman (@sama) September 12, 2024
o1 is still flawed, still limited, and it still seems more impressive on first use than it does after you spend more time with it. pic.twitter.com/Qs1HoSDOz1
A standout feature of the o1-preview models is enhanced reasoning. These models take extra time to process information before responding, permitting greater precision in problem-solving.
During initial testing, the models matched PhD students physics, chemistry, and biology performance.
Their results were even more impressive for math and coding competitions. For one, on a qualifying exam for the International Mathematics Olympiad, the o1 model scored 83%, vastly better than GPT-4o’s 13%.
Limitations and Focus
Despite progress, the o1-preview has limitations. This needs some practical GPT-4 capabilities like web browsing and file uploading. However, OpenAI stresses that the o1-preview focuses more on complex problem-solving, appealing to industries that require sophisticated AI.
OpenAI Reaches 1 Million Users, Plans Premium SubscriptionFurthermore, OpenAI also launched the o1-mini model which is comparatively cheaper than the standard o1 model. Specifically designed for developers who need to code complex applications without the need for a general knowledge of the world, the o1-mini is 80% cheaper than the o1-preview.
Expanded User Access
ChatGPT Plus and Team users could now select o1-preview and o1-mini from the model menu. The o1-preview allows up to 30 messages, while the o1-mini permits 50. API users at the highest usage tier also gained access.
OpenAI plans to increase the availability of o1-mini to the users of ChatGPT Free version. The company is aiming at including the browsing and uploading functions in the subsequent versions of the software. Despite acknowledging its flaws, Sam Altman stated that it has great potential, saying that this is the ‘beginning of a new AI paradigm’.
