Show HN: Redactle LLM Leaderboard (redactle.net)

A new leaderboard titled "Redactle LLM Leaderboard" has been unveiled, showcasing the performance of various language models in solving the Redactle puzzle game. Notably, models from Google's Gemini series have dominated the leaderboard, with the Gemini 3.8 Flash configurations achieving perfect solve rates across their evaluated attempts. This performance highlights advancements in model training and efficiency, as both speed and cost metrics were recorded, with Gemini models running at minimal costs ranging from $0.005 to $0.009 per run, and solving times averaging between 5 to 10 seconds. The significance of this leaderboard lies in its comprehensive evaluation of language models under real-world constraints, demonstrating their ability to efficiently perform complex tasks using minimal guesses. Not only does this present a competitive landscape for AI/ML researchers to assess model effectiveness, but it also raises implications for practical applications, such as improved chatbots or smarter interactive agents. As the AI/ML community continues to push the boundaries of model capabilities, leaderboards like this provide essential benchmarks for progress and an opportunity for further innovation in AI technologies.