LLM Chess Tournament Concludes: OpenAI o3 Claims Championship, xAI Grok 4 Swept Without Winning a Single Game

The Kaggle AI International Chess Tournament has concluded, with the untrained o3 defeating Grok 4 with a score of 4-0, showcasing its strong reasoning ability. (Background: Musk threatened to sue Apple: App Store rankings exhibit monopolistic behavior, maliciously suppressing Grok) (Supplementary background: Grok 4 is now available for free use, and Musk's xAI is in direct confrontation with GPT-5) Recently, Google-owned Kaggle announced the results of the "Artificial Intelligence Chess Exhibition" on August 14, where OpenAI's general-purpose large language model o3 swept xAI's Grok 4 with a score of 4-0, winning the championship and becoming the first LLM to completely shut out its opponent without specialized training. A total of 8 AI participated in the tournament, which lasted three days and was decided by elimination matches. Highlights of the language model competition According to OpenTools.ai, o3 achieved a series of three consecutive 4-0 shutout victories during its progression, eliminating its lightweight version o4 mini in the semifinals. In contrast, Grok 4 often took the lead in the early rounds but lost multiple times at the end of the tournament (sacrificing the most powerful piece, the Queen). Chess grandmaster Hikaru Nakamura evaluated o3 as making "very few mistakes" and pointed out that Grok 4 frequently suffered from tactical self-destruction. Former world champion Magnus Carlsen described Grok's play style as: watching a child play chess. He estimated Grok's Elo at about 800, and o3 at about 1200, far below top human players or specialized chess AIs. Elo: A professional rating system (English: Elo rating system) created by Hungarian-American physicist Arpad Elo, is a method for measuring the level of various competitive activities and is recognized as the authoritative standard for assessing competitive levels today, widely used in chess, Go, football, basketball, and other sports. The highest Elo score in chess was set by Magnus Carlsen at 2882. General AI vs. Specialized AI Stockfish, a specialized system, relies on deep searching and domain scoring, maintaining an Elo of about 3644. General-purpose LLMs, on the other hand, learn from large-scale interdisciplinary corpora, with chess being just an extension of their reasoning ability. Although o3 can defeat Grok 4, it earlier this year still lost to Stockfish, indicating that general models still have gaps in stability and depth of computation in chess game reasoning. Related Reports Betting on OpenAI, Masayoshi Son has "turned the tables" again Ethereum developer installs "malicious AI plugin" and gets hacked, crypto wallet emptied in three days, ten years of cybersecurity experience proved useless a16z's latest insights: Is traditional e-commerce dead? AI-native platforms are redefining the concept of "shopping" The article "LLM Chess Tournament Concludes: OpenAI o3 Wins Championship, xAI Grok 4 Shut Out Without Winning a Single Game" was first published in BlockTempo, the most influential blockchain news media.