OpenAI today announced an improved version of its most powerful version artificial intelligence The previous model – one that requires even more time to think about questions – just a day later Google announced its first model of its kind.
OpenAI’s new model, called o3, replaces o1, which the company developed introduced in September. Like o1, the new model spends time thinking about a problem to provide better answers to questions that require step-by-step logical reasoning. (OpenAI has chosen to omit the nickname “o2” as it is already the name of a mobile operator in the UK.)
“We view this as the beginning of the next phase of AI,” OpenAI CEO Sam Altman said in a livestream on Friday. “Where you can use these models to do increasingly complex tasks that require a lot of thought.”
According to OpenAI, the o3 model performs significantly better than its predecessor on several metrics, including measuring complex programming skills and advanced math and science skills. It is 3 times better than o1 in answering questions asked by ARC AGIa benchmark designed to test the ability of AI models to solve extremely difficult mathematical and logical problems they encounter for the first time.
Google is pursuing a similar line of research. Noam Shazeer, a Google researcher, yesterday revealed in a post on X that the company has developed its own reasoning model called Gemini 2.0 Flash Thinking. Google CEO Sundar Pichai called it “our most thoughtful model yet” in his post. own contribution.
The two duel models show that the competition between OpenAI and Google is fiercer than ever. It is critical for OpenAI to demonstrate that it can continue to make progress to attract more investment and build a profitable business. Google, meanwhile, is desperate to show that it remains at the forefront of AI research.
The new models also show that AI companies are increasingly moving beyond simply scaling AI models to unlock more intelligence from them.
According to OpenAI, there are two versions of the new model, o3 and o3-mini. The company is not yet making the models publicly available, but says it will invite outsiders to apply to conduct tests. OpenAI also revealed more details today about the techniques used to target o1. This includes asking the sample reason about the type of request being made and whether it may be violating its guidelines.
