Blackjack-v1
A simulated blackjack benchmark for testing decisions under uncertainty with a small observation space.
Published by Farama Foundation
Setup and evaluation
The agent chooses whether to hit or stand using its hand total, the dealer's visible card, and whether it has a usable ace. Use it to practice evaluating a policy over many short episodes or compare Monte Carlo estimates across hand states.
Install with pip install gymnasium and create gymnasium.make('Blackjack-v1'). Use the same payout settings for each policy in a comparison.
- Framework
- Gymnasium
- Access
- Local Gymnasium simulation. No account required.