In a recent AI competition hosted by OpenRouter, xAI’s Grok 4.1 Fast emerged as the top-performing large language model (LLM), winning 43% of the 30 matches played in a 2D battle royale format. The contest, held in early June 2026, featured eleven LLMs including Anthropic’s Claude, with Grok outperforming all rivals while being the most cost-efficient model by a factor of 27 on cost per win, according to openrouter.ai.
The competition involved deploying the eleven LLMs in a simulated environment where they competed in strategic matches. Grok 4.1 Fast demonstrated superior tactics and adaptability, while some models failed to secure any wins. The contest also highlighted behavioral differences, such as one model frequently attempting to form alliances by sharing its location. OpenRouter’s blog detailed the full data and provided video footage of the matches, emphasizing Grok’s dominance in both performance and cost-effectiveness.
This event underscores the growing capabilities and competitiveness of AI language models in dynamic, interactive scenarios beyond traditional benchmarks. Grok’s victory over Claude and other models signals a shift in the AI landscape, where efficiency and strategic interaction are becoming key differentiators. The cost advantage demonstrated by Grok 4.1 Fast could influence adoption decisions in applications requiring real-time, resource-sensitive AI deployment.
OpenRouter’s blog post published on June 4, 2026, includes comprehensive insights and data from the battle royale, offering a detailed look at each model’s performance. The video of a full match is available on YouTube, providing transparency and a resource for further analysis of AI model behaviors in competitive settings.