OpenAI Launches o1 Pro, a New Reasoning Model That Tops Key Benchmarks
Read the full article for context, quotes, and updates from the team.
OpenAI has released o1 Pro, a new reasoning-focused AI model that the company says delivers major gains over earlier versions such as o1-preview on demanding tasks in math, coding, and science. The model is designed to improve step-by-step problem solving, allowing it to tackle complex questions with greater accuracy and consistency.
According to OpenAI, o1 Pro has achieved strong results on several challenging benchmarks. It scored 93% on AIME 2024, a difficult math competition benchmark, and 79% on GPQA Diamond, a test used to measure performance on graduate-level science questions. These results suggest the model is better equipped to handle tasks that require careful reasoning rather than simple pattern recognition.
The launch highlights OpenAI’s continued push into advanced reasoning systems, an area that has become increasingly important as AI models are expected to perform more reliably on technical and analytical work. o1 Pro is now available to ChatGPT Pro subscribers for $200 per month, positioning it as a premium offering for users who need higher-end capabilities.
With its improved benchmark performance and focus on structured reasoning, o1 Pro represents a notable step forward in OpenAI’s model lineup and in the broader race to build more capable AI systems.
