OpenAI Upgrades Its Smartest AI Model With Improved Reasoning Skills

OpenAI Today announced an improved version of its most capable Artificial intelligence The model to date—one that takes more time to think through questions—is a day later Google Announced its first model of this kind.

OpenAI’s new model, called o3, replaces o1, which is the company Launched in September. Like o1, the new model spends time mulling over a problem to better answer questions that require step-by-step logical reasoning. (OpenAI chose to skip the “o2” moniker because it’s already the name of a UK mobile carrier.)

“We see this as the beginning of the next phase of AI,” OpenAI CEO Sam Altman said in a livestream Friday. “Where you can use these models to do increasingly complex tasks that require a lot of logic.”

The O3 model scores higher on several measures than its predecessor, OpenAI says, including measuring complex coding-related skills and advanced math and science skills. The answer to the question posed by it is three times better than o1 ARC-AGIA benchmark designed to test an AI model’s ability to reason on extremely difficult mathematical and logical problems that they are encountering for the first time.

Google is following a similar line of research. Noam Shazir, a Google researcher, yesterday As revealed in a post by X The company has developed its own logic model, called Gemini 2.0 Flash Thinking Google CEO Sundar Pichai called it “our most thoughtful model yet”. own post. Google’s new model earned a high score on SWE-Bench, a test that measures a model’s agentic capabilities.

However, OpenAI’s new o3 model is 20 percent better than o1. “The o3 blew it out of the water,” said About the pressis a post-doctoral researcher at Princeton University who helped develop SWE-Bench. “Pretty amazing growth, not sure how they did it.”

The two dual models show that the competition between OpenAI and Google is fiercer than ever. It is crucial for OpenAI that it continues to progress as it seeks to attract more investment and build a profitable business. Google is already desperate to show it’s at the forefront of AI research.

The new models also show how AI companies are increasingly looking to extract more intelligence from AI models beyond scaling.