Research
156d agoNew Benchmark Outperforms Top Poker AI
The GTO Wizard Benchmark introduces a rigorous, public framework for evaluating heads‑up poker agents, surpassing the previous champion Slumbot by an average of 19.4 big blinds per 100 hands.
The Signal
By integrating the AIVAT variance‑reduction method, the benchmark achieves statistical significance with tenfold fewer simulations, accelerating global research into game‑theoretic AI and agent robustness.