Google is introducing the Kaggle Gaming Arena, a platform designed to evaluate the reasoning capabilities of leading artificial intelligence models through live, competitive matches. The inaugural event was a chess tournament featuring six prominent AI systems, including ChatGPT, Gemini, Claude, Grok, Deepseek, and Kimi.
The chess competition had an esports-similar format, streaming daily matches on YouTube and featuring commentary from prominent names from the sport like Grand Masters Magnus Carlsen, Hikaru Nakamura and David Howell, and creator Levy “GothamChess” Rozman.
In a social media post, Google says the initiative moves beyond traditional benchmarks by testing how AI models think, adapt, and recover under pressure, with all reasoning steps visible to viewers. The company hopes the approach will reveal differences in strategic ability that standard testing methods may overlook.
New games will be seen soon at Kaggle Game Arena, as the initiative aims to be “a new, open benchmark platform where top AI models compete in complex, strategic games in streamed match-ups,” as it reads on Kaggle’s profile on X. Kaggle as a whole is already positioned as a platform for data science and machine learning professionals to compete while developing the most effective models.
For starters, chess has long been a measure of machine intelligence, dating back to IBM’s Deep Blue defeating Garry Kasparov in 1997. The traditional game is blending into modern esports, as streamers use gaming-first platforms like Twitch to cast their online matches and tournaments like the Esports World Cup (EWC) included the game in its schedule through platform Chess.com. The result, according to Esports Charts, was the highest livestream viewership for chess in 2025.
