Grok 4, developed by xAI, has emerged as a leading artificial intelligence model in competitive AI benchmarks and chess competitions. It recently defeated Google's Gemini 2.5 Pro with a 3-2 score in the Kaggle AI Chess semi-finals, advancing to the grand finale. Grok 4 also outperformed GPT-5 on the ARC-AGI benchmark, showcasing strong capabilities in AI reasoning tasks. xAI has released several updates to the Grok app, including improved audio for Imagine videos, enhanced sharing features, and better handling of long PDF files. Despite Grok 4's competitive edge in some benchmarks, GPT-5 leads on others, such as the WeirdML and SCALE benchmarks, and scores higher on certain IQ tests when run with additional tools. Grok 4 recently climbed to second place on a global AI leaderboard with a score of 60.5%, surpassing GPT-5 (high) and Claude 4.1 Opus, with only Gemini 2.5 Pro ahead. The AI community continues to monitor the evolving performance of these models as xAI expands Grok's capabilities and presence in AI competitions.