AdBench

Adversarial Benchmarks for AI Agents

Coming Soon

First benchmarks dropping soon. We're testing frontier models in multi-agent games that actually require theory of mind, deception, and strategic reasoning.