The OpenAI o3 artificial intelligence model defeated Grok 4 from xAI, a company linked to billionaire Elon Musk, in a chess tournament, the BBC reported on Friday. The competition involved computers designed for everyday use, not chess-playing computers.
Google's Gemini model took third place in the competition, beating another OpenAI model.
Chess.com journalist Pedro Pinhata said that until the semifinals, everything pointed to Grok 4 winning. In his opinion, he was by far the “strongest chess player” until the final game. Pinhata stated that Grok's “incomprehensible” and “clumsy” play allowed o3 to achieve a series of “convincing victories.”
American chess grandmaster Hikaru Nakamura said during the live broadcast of the final that “Grok made a lot of mistakes, but OpenAI didn't.”
A chess tournament featuring eight AI models created by Anthropic, Google, OpenAI, xAI, and the Chinese developers of DeepSeek and Moonshot AI took place on Google's Kaggle platform. The tournament lasted three days. Kaggle is a platform that allows data scientists to evaluate AI systems by competing against each other.
The BBC reminded that chess and Go are complex rule-based strategic games that are often used to assess the abilities of an artificial intelligence model, including reasoning and coding.
From London Marta Zabłocka (PAP)
mzb/ brick/
