♠ AlphaHoldem

A 15 BB sit-and-go vs a self-play RL agent — last player standing wins. Runs entirely in your browser.

Loading model…

Hand 0 — your stack: 15 BB

Table AI advice

Hand log

Model: population self-play PPO (Trinal-Clip) over a 12-action abstraction, 1–20 BB stacks, 2–3 players. Inference via onnxruntime-web (WASM) with a pure-JS fallback; no data leaves your browser.