OpenAI’s o3 model has claimed victory over Elon Musk’s Grok 4 in the final of a unique artificial intelligence chess tournament, marking another chapter in the growing rivalry between the two AI giants.
The event, hosted on Google-owned data science platform Kaggle, brought together eight large language models (LLMs) from leading AI developers including OpenAI, xAI, Google, Anthropic, and Chinese firms DeepSeek and Moonshot AI. Unlike traditional chess engines, these were general-purpose AI systems designed for everyday tasks, making their performance in the strategy-heavy contest a test of reasoning and adaptability rather than raw computational power.
OpenAI’s o3 emerged unbeaten throughout the three-day competition, sealing its win in the final with a series of decisive games against Grok 4. Google’s Gemini took third place after defeating another OpenAI model in the playoff match.
While Musk had downplayed Grok’s chess abilities ahead of the final — describing its earlier wins as a “side effect” of its design and claiming xAI had “spent almost no effort on chess” — the defeat was notable given Grok’s strong run through the early stages.
“Up until the semi-finals, it seemed like nothing would be able to stop Grok 4,” said Pedro Pinhata, a Chess.com writer covering the tournament. “But the illusion fell through on the last day. Grok’s play became unrecognisable, with repeated blunders, including losing its queen multiple times.”
Chess grandmaster Hikaru Nakamura, who livestreamed commentary on the final, noted: “Grok made so many mistakes in these games, but OpenAI did not.”
Chess has long been used to measure AI progress. In the late 1990s, IBM’s Deep Blue defeated world champion Garry Kasparov in a landmark moment for computing. More recently, Google’s DeepMind developed AlphaGo, which famously beat Go champion Lee Se-dol in 2016, leading to the player’s retirement three years later.
For AI developers, chess provides a valuable benchmark to test a model’s ability to navigate complex, rule-based environments and strategise toward victory. This competition offered a modern twist: assessing whether conversational AI systems, without specific chess training, could adapt and excel at the game.
While OpenAI’s o3 now holds bragging rights, the tournament also revealed that general-purpose AI still has room to improve before rivaling dedicated chess engines. Yet for Sam Altman’s company, the win over Musk’s xAI adds momentum in the broader battle over which firm leads the race to create the world’s most capable AI.
