Published on: August 10, 2025 12:22 pm IST
Groke 4 started in the competition, but stumbled in the final match against Chat O3.
The Chatgpt O3 model of Openai defeated Elon Musk’s XAI model Grock 4 in the final of a Kaggle-Hosted tournament, determined to find the strongest chess-game large language model (LLM). The program, organized in three days, raised general-purpose LLM against each other instead of special chess engines.
Tournament format and participant
Eight models participated, including Openai, XAI, Google, Ethropic and Chinese Developers Deepsek and Moonshot AI entries. The competition used standard chess rules, but tested multi -purpose LLM, the system that are not particularly adapted to play chess. The BBC coverage of the incident noted that Google’s Mithun finished third by defeating another openiI entry.
Mobile finder: iPhone 17 Air is expected to debut next month
Last and major moments
Groke 4 started in the competition but stumbled in the final match against O3. Commentators and supervisors highlighted several strategic errors by Groke 4, which also included the Queen’s disadvantages, who rotated the match in favor of O3. Chess.com writer Pedro Pinhata said: “By the semi -finals, it seemed that anything would not be able to stop Groke 4,” but said Groke’s drama “fell under pressure”. Grandmaster Hikaru Nakamura, who made a live comment, said: “Groke made a lot of mistakes in these games, but Openai did not.”
Reactions and extensive references
Elon Musk reduced the necklace, stating that the earlier strong results of Groke were a “side effects” and Xai “had almost no attempt on chess.” The result combines a public dimension to the rivalry between the XAI and Openai of the Musk, both established by those who once worked together in Openai.
Chess has long been used to measure AI progression. The previous milestone includes special systems such as Deepmind’s Alfago, which defeated the top human players in the game of Go of Go. This Kagal tournament differs by test of normal LLM on strategic, sequential functions rather than using a dedicated chess engine.
What does it mean
The result in the consequences shows how LLM handles structured, adverse functions like chess. While the performance of the O3 shows that some LLM can maintain strategic game under the terms of the tournament, the fall of Groke 4 suggests that the results may still be inconsistent. The organizers and commentators are likely to continue using chess and similar tasks, as the area develops to check for logic, planning and strength in large language models.
