OpenAI's "secret" model becomes math champion
OpenAI has achieved a gold medal-level performance at the International Mathematical Olympiad (IMO), adding to its rapid progress in artificial intelligence research.
Alexander Wei, a researcher at the company working on large language models (LLM) and reasoning, announced on the X platform that an experimental research model has successfully completed the “long-awaited grand challenge” in the field of artificial intelligence.
A HIDDEN MODELAccording to Wei, an OpenAI model, which has not yet been released to the public, solved five out of six problems at the IMO, one of the world's most established and prestigious mathematics competitions, achieving a total score of 35 out of 42.
The IMO is known for its competitive nature, where countries send six students each to solve challenging algebra and precalculus problems. While the problems may appear simple on the surface, achieving a full score requires creative and in-depth solutions. Only 67 of the 630 participants in this year's competition earned a gold medal, representing a success rate of approximately 10 percent.
HUMAN-LIKE REASONINGWhile AI systems are often effective in areas such as analyzing large data sets or performing repetitive tasks, they may fall short in solving problems that require creative thinking or complex decision-making.
However, OpenAI says that with this latest success, the model they developed can solve complex mathematical problems with human-like reasoning.
“Thanks to this achievement, we have achieved a model that can construct detailed and robust logic chains at the level of human mathematicians,” said OpenAI CEO Sam Altman.
However, the company has stated that the model won't be released to the public anytime soon. This suggests that the highly anticipated GPT-5 will be more advanced than the previous version, but it won't yet be capable of solving math at the IMO level.
ntv
