Show HN: Redactle LLM Leaderboard

Category: ai-ml

Tags: llm, benchmark, leaderboard, games, evaluation

Score: 4.0/10 (Innovation: 5, Technical: 3, Documentation: 3, Utility: 4)

A leaderboard that evaluates different LLMs on their ability to solve Redactle puzzles, a game where players guess words to reveal a redacted Wikipedia article. The project provides a dashboard comparing model performance and cost, revealing surprising results about which models excel.

Target audience: AI researchers, LLM enthusiasts, and developers evaluating model performance

Repository: https://redactle.net/llm-leaderboard

View on Hacker News