TurtleBench: Evaluating Top Language Models via Real-World Yes/No Puzzles Paper • 2410.05262 • Published Oct 7 • 9