The Openai model defeated Grak in the final of the artificial intelligence chess race

O3 AI O3 has succeeded in defeating the XAI Grok 4 model and becoming the championship of the all -encompassing artificial intelligence chess tournament. In the competition, Grok made gross and strange mistakes, and now the fire of competition between Sam Altman, CEO Openai, and Ilan Musk, the owner of XAI, has become more flame.

The competition, which was held on Google’s KAGGLE platform, was fundamentally different from historical competitions such as the Battle of Gary Casparov and the DeepBello PC. There was no news of chess computers here. Instead, the large -scale language models designed for everyday work were confronted with each other to measure their “reasoning” and “strategic learning”. In the three -day tournament, eight models of artificial intelligence from leading companies such as XAI, Google, Openai, Anthropic and Chinese Deepseek and Moonshot AI competed.

Grok failure of OpenAI model in chess race

Prior to the final, the Grok 4 seemed to be the strongest player on the field. Pedro Pinhata of the Chess website writes: “Before the half -half, it seemed that nothing could stop Grok … but this illusion was lost on the last day.” In the final games, Grok completely collapses against the O3’s O3 model by presenting a “flashless” game. “Grok made a lot of mistakes in these games, but Openai wasn’t,” said Hikaru Nakamura, a great chess professor.

Grok mistakes were so strange that in several games simply lost his minister; A mistake that is very rare at the top of the chess.

Ilan Musk, even before the final and perhaps predicting failure, tried to address the importance of the competition. He wrote in a post in X that the previous success of Grok in the tournament was a “side effect” and XAI “has made almost no attempt to succeed in chess.” This comment is considered an attempt to manage the defeat against his longtime rival, Sam Altman; The two people who both claim to have the smartest models of artificial intelligence in the world.

Overall, complex strategic games such as chess have long been used as a criterion for measuring the intelligence and ability of cars. The historical victory of Alfago (Google’s Dipmind product) over the world champion, Lee Se-Dol, which led to his retirement, showed that artificial intelligence can surpass human strategies. The new tournament also showed that large language models are learning deep reasoning skills beyond text production.