OpenRouter ran 11 LLMs in a 30-game battle royale — Grok 4.1 Fast won 43%

A Robot Is Sprinting Towards You: Do You Want It Running on Claude or Grok?

OpenRouter's Jacky Liang dropped 11 LLMs into a 2D battle royale for 30 matches. Grok 4.1 Fast won 13 games at $0.97 per win; Claude Sonnet 4.6 won 5 at $26.78 per win — a 27x gap. GPT 5.4 had the most kills (38) but only 2 wins, so killing more didn't mean winning more. GPT 5.4-mini, DeepSeek 4 Flash, and Kimi K2.6 spent $57 combined and won zero games. The models reasoned, called tools, and updated memory each turn — they weren't just generating control code. The post doesn't provide the full leaderboard or detailed behavioral differences across all models.

Why it matters: OpenRouter's official blog, author Jacky Liang ran 30 games himself with full data and replays. Grok 4.1 Fast's cost advantage is stark, Claude Sonnet 4.6 is expensive but consistent, GPT 5.4 is the kill leader but can't close — all three takeaways are concrete and verifiable....