♠ AlphaHoldem

A Spin & Go vs a self-play RL agent — 1,500 chips, blinds rise every 7 hands, last player standing wins. Runs entirely in your browser.

Loading model…

Hand 0 · Blinds 50/100 — 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.