OpenAI has launched a new reasoning model, internally code-named 'Strawberry,' now integrated into ChatGPT, which dramatically enhances the AI's ability to solve complex logic and math problems. According to OpenAI, this model represents a major advancement in AI reasoning, allowing ChatGPT to 'think' more deliberately before responding.
How Strawberry Improves Reasoning
The Strawberry model uses a technique called 'chain-of-thought' reasoning, which enables the AI to break down multi-step problems into smaller, logical parts. In benchmark tests, Strawberry scored 83% on the MATH dataset, compared to 58% for previous versions. OpenAI CEO Sam Altman stated, 'This is a fundamental step toward AI that can reason like humans.'
Integration into ChatGPT
ChatGPT Plus and Enterprise users now have access to Strawberry through a toggle in the interface. The model is particularly effective at solving math word problems, coding challenges, and scientific reasoning tasks. Early users report that the AI now explains its reasoning step-by-step, reducing errors in complex queries.
Impact on AI Development
Experts say Strawberry could accelerate AI applications in education, research, and software development. Dr. Emily Chen, an AI researcher at MIT, noted, 'This is a clear breakthrough. It shows that scaling reasoning capabilities is possible without massive increases in compute.' However, OpenAI warns that the model still makes mistakes and should not be relied upon for critical decisions without human oversight.
What's Next
OpenAI plans to refine Strawberry further and expand its availability. The company is also exploring how reasoning models can be applied to robotics and scientific discovery. With Strawberry, ChatGPT is no longer just a language model—it's becoming a reasoning engine.
