OpenAI Unveils o1: A Reasoning-Powered AI Model

According to OpenAI's official blog, OpenAI has released a reasoning-powered AI model called o1, internally codenamed "Strawberry". OpenAI o1 can reason through complex tasks and solve problems that are more challenging than previous scientific, coding, and mathematical models. In testing, OpenAI o1 performed similarly to PhD students on challenging benchmark tasks in physics, chemistry, and biology, excelling in mathematics and coding.

While GPT-4o only correctly solved 13% of the problems on the International Mathematical Olympiad (IMO) qualification exam, the reasoning model scored 83%. OpenAI o1's coding capabilities reached the 89th percentile in the Codeforces competition.

As an early model, OpenAI o1 lacks many useful features of ChatGPT such as browsing the web for information and uploading files and images. GPT-4o will be more powerful in the short term. However, this is a significant advancement for complex reasoning tasks, representing a new level of AI capability. In light of this, the counter has been reset to 1 and this series is named OpenAI o1.

Healthcare researchers can use o1 to annotate cell sequencing data, physicists can use o1 to generate complex mathematical formulas required for quantum optics, and developers across all fields can use o1 to build and execute multi-step workflows.

OpenAI also released OpenAI o1-mini, a faster, cheaper reasoning model that is particularly effective in coding. As a smaller model, o1-mini is 80% cheaper than o1-preview, making it a powerful and cost-effective model for applications that require reasoning but not extensive world knowledge. OpenAI plans to offer o1-mini access for free to all ChatGPT users.