Detailed view of chess plate and pieces (photo by Dean Moohopoulos/Getty Images) The opening day of the AI Chess exhibition tournament, which organized the Kaggle Game Arena project, witnessed four large language models (LLM) ensuring the dominant victory of 4-0 to advance to the semifinals. Gemini 2.5 Pro, O4-Mini, Grok 4 and O3 defeated their relevant opponents Claude 4 Opus, Deepseek R1, Gemini 2.5 Flash and Kimi K2, which represents the AI models in general purpose in a strategic game.The aim of the new Kaggle Game Arena, a new Google initiative, has an assessment of how LLMS works in a competitive environment. The tournament contains eight leading LLM contestants in a one -off knockout console, while the games broadcast live on multiple platforms.
Musk fights with its own AI on x | Grok fired as a “awakened propaganda machine” right -wing voices
Go beyond the border with our YouTube channel. Subscribe!Google has worked with Deepmind to organize this unique tournament where LLM uses to visualize positions and perform a universal controller called “Harness”. Each AI has four attempts to take a legal step and fails, leading to a loss of play.The match between Kimi K2 and O3 ended quickly, and none of the games lasted over eight moves. Kimi K2 could not consistently make legal movements, although it showed the ability to observe the theory of opening for initial movements.The O4-Mini against Deepseek R1 showed a pattern of strong opening movements, followed by the declining quality of the game. Despite the inconsistency, two control friends managed to reach the O4-mini during the match.“This is a side effect of BTW. @Xai spent almost any effort on chess,” Elon Musk said, responding to the impressive Grok 4 performance in the tournament.The Gemini 2.5 Pro against the Claude 4 OPUS was more control teammates than the illegal forfeiture. The first game showed that how AIS keeps good movements, until nine, when Claude 4 opus made a critical mistake with 10 … G5.Grok 4 presented the strongest performance of the day and demonstrated a special skill in identifying and capitalizing for undefined pieces in his match against Flash Gemini 2.5 Flash.The tournament revealed three primary challenges for LLMS in chess: visualization of the entire album, understanding of interactions of pieces and legal steps. These limitations differ between different AI models.The competition continues on Wednesday 6 August, starting with 13:00 ET / 19:00 trips / 22:30. Viewers can watch the event alive on GM Hikaru Nakamura’s Twitch and YouTube channels, as well as on the tournament dedicated events.
