gymnasium

A standard API for multi-agent reinforcement learning environments, with popular reference environments and related utilities

Simple and easily configurable grid world environments for reinforcement learning

PyBullet Gymnasium environments for single and multi-agent reinforcement learning of quadcopter control

Reinforcement Learning environments based on the 1993 game Doom

Clean PyTorch implementations of imitation and reward learning algorithms

A standard format for offline reinforcement learning datasets, with popular reference datasets and related utilities

Modular Reinforcement Learning (RL) library (implemented in PyTorch, JAX, and NVIDIA Warp) with support for Gymnasium/Gym, NVIDIA Isaac Lab, MuJoCo Playground and other environments

Reinforcement Learning environments for Traffic Signal Control with SUMO. Compatible with Gymnasium, PettingZoo, and popular RL libraries.

A collection of robotics simulation environments for reinforcement learning

Unified Reinforcement Learning Framework

Multi-objective Gymnasium environments for reinforcement learning

A fork of gym-retro with additional games, emulators and supported platforms

An open framework to simulate and deploy perception-based PX4/ArduPilot drone swarms with ROS2, YOLO, LiDAR, NVIDIA Jetson

Multi-Objective Reinforcement Learning algorithms implementations.

Base Mujoco Gymnasium environment for easily controlling any robot arm with operational space control. Built with dm-control PyMJCF for easy configuration.

Model Predictive Path Integral Control (MPPI) with PyTorch

PettingZoo and Gymnasium bindings for popular reinforcement learning environments outside of Farama

Gymnasium trading environment for FX and crypto: backtrader execution bridge, protected SL/TP orders, solvency curricula and audit-grade trade diagnostics

A framework for creating rich, 3D, Minecraft-like single and multi-agent environments for AI research. (Accepted at ICML 2025).

⚡ ⚡ Deep RL (PPO) agent that manages a Uniswap V3 concentrated liquidity position, deciding when to hold, collect fees, or rebalance. Trained and validated against real on-chain data