OpenAI has recently launched a new family of AI models under the “o1” series, which introduces significant advancements in reasoning and problem-solving capabilities. This model family, including *o1-preview* and *o1-mini*, is designed to excel at handling complex tasks, particularly in fields like science, math, coding, and advanced problem-solving.
The *o1-preview* model stands out for its ability to “think” more deeply before providing answers, making it particularly useful for tasks requiring higher-level reasoning, such as generating mathematical formulas or analyzing complex scientific data. This model has demonstrated impressive performance in tests, including solving 83% of problems in the International Mathematics Olympiad, far exceeding the capabilities of its predecessor, GPT-4.
In parallel, the *o1-mini* model offers similar reasoning abilities but at a lower cost and faster performance, making it ideal for coding and workflow management. It is 80% cheaper than *o1-preview* while still delivering strong performance in areas like math and coding.
Though still in the preview stage and lacking some features like browsing and file uploads, these models are expected to evolve quickly. They are currently available to ChatGPT Plus and Team users, with plans to expand access to other user tiers(openAI)