In a tournament to find the best artificial intelligence (AI) chess player, OpenAI’s o3 model has beaten Elon Musk’s Grok 4 in the final match. Unlike traditional chess machines built specifically for the game, this contest featured AI programs designed for everyday tasks, making the competition unusual and closely watched.
OpenAI’s o3 remained unbeaten throughout the tournament, defeating Grok 4 and reinforcing the rivalry between the two companies. Both Musk and Sam Altman, OpenAI’s co-founder, claim their AI models are the smartest globally. Google’s Gemini model secured third place by beating another OpenAI entry.
Despite these AIs being talented at many tasks, their chess skills are still developing. Grok 4 made several errors in the final games, including repeatedly losing its queen. Pedro Pinhata of Chess.com said Grok seemed unbeatable until the semifinals but faltered at the end, allowing o3 to secure convincing wins.
Chess grandmaster Hikaru Nakamura noted Grok’s blunders during his livestream, contrasting them with OpenAI’s flawless play. Musk admitted that Grok’s earlier success was a “side effect” since the team “spent almost no effort on chess.”
The competition took place on Google’s Kaggle platform, where eight large language models from various developers battled over three days. Chess remains a popular benchmark to test AI skills because of its complex rules and strategic depth.
Historically, chess has been used to measure computer progress, famously with IBM’s Deep Blue defeating Garry Kasparov in 1997—a landmark moment for AI development.
