OpenAI Upgrades Its Smartest AI Model With Improved Reasoning Skills

Open artificial intelligence Announcing its most powerful and improved version today AI Model so far – need more time to think things over – just one day later Google Announcing the first model of its type.

OpenAI’s new model is called o3, replacing the company’s o1 Launched in September. Like o1, the new model takes time to think about problems in order to provide better answers to questions that require step-by-step logical reasoning. (OpenAI chose to skip the “o2” moniker because it is already the name of a British mobile telecommunications company.)

“We think this is the beginning of the next phase of artificial intelligence,” OpenAI CEO Sam Altman said during a live broadcast on Friday. “You can use these models to complete increasingly complex tasks that require a lot of reasoning.”

OpenAI said the o3 model scored much higher than its predecessor on several metrics, including measures of complex coding-related skills and advanced math and science abilities. The ability to answer the following questions is three times that of o1 ARC-AGIa benchmark designed to test an artificial intelligence model’s ability to reason about extremely difficult mathematical and logical problems it encounters for the first time.

Google is conducting similar research. Yesterday, Google researcher Noam Shazeer Revealed in X’s post The company developed its own inference model called Gemini 2.0 Flash Thinking. Google CEO Sundar Pichai called it “our most thoughtful model yet” in his presentation own posts. Google’s new model achieved high scores on SWE-Bench, a test that measures a model’s agent capabilities.

However, OpenAI’s new o3 model is 20% better than o1. “o3 blew it out of the water,” said About newswas a postdoctoral researcher at Princeton University and helped develop SWE-Bench. “The growth is very surprising, no idea how they do it.”

The showdown between these two models shows that the competition between OpenAI and Google is more intense than ever. For OpenAI, proving that it can continue to improve is crucial as it seeks to attract more investment and build a profitable business. At the same time, Google is eager to show that it remains at the forefront of artificial intelligence research.

The new model also illustrates how AI companies are increasingly moving beyond simply expanding the scope of AI models to gain more intelligence from them.