A unique chess tournament brought together AI models from leading global tech companies, pitting consumer-ready systems against each other. The three-day event, hosted on Google-owned Kaggle, featured eight models from Anthropic, Google, OpenAI, xAI, and Chinese firms DeepSeek and Moonshot AI.
In the decisive final, OpenAI’s O3 model defeated xAI’s Grok 4 to claim the title, remaining undefeated throughout the tournament. Grok 4 had been a favorite up to the semifinals but faltered in the final, repeatedly making major blunders, including losing its queen.
Google’s Gemini model secured third place after beating another OpenAI model in the playoff. Chess grandmaster Hikaru Nakamura remarked that Grok made many mistakes, while OpenAI avoided them.
While the AI models displayed impressive chess skills, experts noted they are still learning the finer aspects of the game. The contest aimed to assess the strategic thinking and problem-solving abilities of widely accessible AI systems, rather than specialist chess engines.
Chess has long been a benchmark for measuring computer intelligence, but this tournament offered a fresh glimpse into how modern AI models handle strategy under competition. OpenAI’s flawless run highlighted its model’s tactical edge, setting a high bar for future AI-versus-AI contests.