AI Board Game Benchmark Report: Werewolf and Mafia

Mentiss compares AI models in Werewolf and Mafia, social deduction board games that require hidden-role reasoning, persuasion, deception detection, and strategic coordination.

Models and metrics

The benchmark evaluates models such as GPT, Claude, Gemini, Grok, DeepSeek, and custom BYOM models by win rate, voting accuracy, role-action accuracy, persuasion impact, and deception quality.