A Gymnasium environment for Tetris on The Nintendo Entertainment System (NES) based on the nes-py emulator. It currently supports CPython 3.13 and 3.14 in CI.

The preferred installation of gym-tetris is from pip:

pip install gym-tetris

Python 3.13 or newer is required. The supported CI targets are CPython 3.13 and 3.14.

You must import gym_tetris before trying to make an environment. This is because Gymnasium environments are registered at runtime. By default, gym_tetris environments use the full NES action space of 256 discrete actions. To constrain this, gym_tetris.actions provides an action list called MOVEMENT (12 discrete actions) for the nes_py.wrappers.JoypadSpace wrapper. There is also SIMPLE_MOVEMENT with a reduced action space (6 actions). For exact details, see gym_tetris/actions.py.

import gymnasium as gym
import gym_tetris
from gym_tetris.actions import MOVEMENT
from nes_py.wrappers import JoypadSpace

env = gym.make('TetrisA-v0', render_mode='rgb_array')
env = JoypadSpace(env, MOVEMENT)

observation, info = env.reset(seed=123)
terminated = False
truncated = False

for step in range(5000):
    if terminated or truncated:
        observation, info = env.reset(seed=123)
        terminated = False
        truncated = False
    observation, reward, terminated, truncated, info = env.step(
        env.action_space.sample(),
    )
    frame = env.render()

env.close()

NOTE: gym_tetris.make is just an alias to gymnasium.make for convenience.

NOTE: remove calls to render in training code for a nontrivial speedup.

gym_tetris features a command line interface for playing environments using either the keyboard, or uniform random movement.

gym_tetris -h
gym_tetris --env TetrisA-v0 --mode human --actionspace simple
gym_tetris --env TetrisA-v0 --mode random --steps 100 --render --seed 123
gym_tetris --env TetrisA-v0 --mode random --steps 100 --no-render --actionspace simple --seed 123
gym_tetris --env TetrisB-v0 --mode random --steps 100 --render --actionspace standard

Human mode requires rendering, so --mode human --no-render is rejected. Use --seed/-S to seed only the first environment reset in CLI playback.

There are two game modes defined in NES Tetris, namely, A-type and B-type. A-type is the standard endurance Tetris game and B-type is an arcade style mode where the agent must clear a certain number of lines to win. There are three potential reward streams: (1) the change in score, (2) the change in number of lines cleared, and (3) a penalty for an increase in board height. The table below defines the available environments in terms of the game mode (i.e., A-type or B-type) and the rewards applied.

The info dictionary returned by the step method contains the following keys:

PyPI releases are published by the Publish to PyPI GitHub Actions workflow through PyPI trusted publishing, not by local twine credentials. Configure the PyPI project publisher with owner Kautenja, repository gym-tetris, workflow filename publish.yml, and environment pypi.

Releases should follow the current GitHub Actions flow:

1. Create and push a tag that matches pyproject.toml's version, with or without a leading v.
2. Let CI build the tagged distribution artifacts.
3. Publish a GitHub Release from that tag to trigger the trusted-publishing workflow.

Build distributions locally with python -m build only for verification, not for authenticated upload.

Please cite gym-tetris if you use it in your research.

@misc{gym-tetris,
  author = {Christian Kauten},
  howpublished = {GitHub},
  title = {{Tetris (NES)} for {Gymnasium}},
  URL = {https://github.com/Kautenja/gym-tetris},
  year = {2019},
}

The following references contributed to the construction of this project.

1. Tetris (NES): RAM Map. Data Crystal ROM Hacking.
2. Tetris: Memory Addresses. NES Hacker.
3. Applying Artificial Intelligence to Nintendo Tetris. MeatFighter.