OpenAI model wins gold at the mathematical Olympiad

OpenAI has announced a new achievement - an experimental AI model managed to score 35 out of 42 points at this year's International Mathematical Olympiad (IMO), which corresponds to the level of a gold medal.

Here's What We Know

This is the first time AI has achieved such a result at one of the most prestigious mathematical competitions in the world.

According to OpenAI researcher Alexander Wei on the X platform, the model solved 5 out of 6 complex problems usually solved by the best students from around the world. These tasks require not only knowledge of algebra and analysis, but also a high level of creativity and logic. For comparison, this year only 67 out of 630 participants won gold.

Usually, artificial intelligence is good at routine tasks but weaker at creative and complex tasks. However, this time, according to OpenAI, the model coped with the IMO problems, demonstrating human logical thinking.

Despite the success, OpenAI does not plan to open access to this model for the time being. According to the company's CEO Sam Altman, this level of reasoning will not be available for several months. This means that even in GPT-5, which may be released in the near future, such mathematical insight will probably not be implemented yet.

Source: @alexwei_