# Farama-Foundation/Gymnasium

A standard API for single-agent reinforcement learning environments, with popular reference environments and related utilities (formerly Gym)

Repository: https://github.com/Farama-Foundation/Gymnasium
Canonical: https://ross.abutalabs.com/products/gymnasium
Homepage: https://gymnasium.farama.org
Language: Python
License: MIT
License Family: permissive
Topics: api, gym, reinforcement-learning
Last push: 2026-08-19T16:01:04+00:00

## Health v2 (maintenance only)
Score: 88/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 98, release rhythm 68, longevity 100
- inputs: {"age_days": 1456, "days_push": 14, "days_rel": 133, "gap_med": 88, "n_releases_24m": 8}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 12408, forks 1434 (observed 2026-08-28T04:10:52.618399+00:00)

## What it is
Gymnasium is a Python library providing a standard API for single-agent reinforcement learning environments, maintained by the Farama Foundation as the successor to OpenAI's Gym. It ships a diverse collection of reference environments (Classic Control, Box2D, Toy Text, MuJoCo, Atari) plus utilities like wrappers and vectorization.

## Use cases
- train and benchmark RL algorithms on standard environments
- create custom reinforcement learning environments with a standard API
- debug RL algorithm implementations on simple toy text environments
- vectorize environments for faster parallel RL training
- wrap environments to modify actions, observations, or rewards
- migrate code from OpenAI Gym to a maintained fork

## When to choose
- you need a well-maintained, standard RL environment API in Python
- you want ready-made reference environments like CartPole, Lunar Lander, or MuJoCo tasks
- you are developing or comparing single-agent RL algorithms
- you are starting a new RL project and need ecosystem compatibility

## When to avoid
- you need multi-agent RL environments (use PettingZoo instead)
- you need OpenAI Gym specifically for legacy compatibility
- you need GPU-accelerated environments (consider Brax or Isaac Gym)
- your project depends on the old gym package and cannot migrate

## Facets
- artifact type: library
- maturity: stable
- function: machine-learning, reinforcement-learning, simulation, sdk
- domain: reinforcement-learning, machine-learning, artificial-intelligence
- platform: python, cross-platform
- tags: reinforcement-learning, gym, environments, api-standard, mujoco, atari, farama-foundation, game-development

## Member repositories
- Farama-Foundation/Gymnasium (main) score 88

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:52.618399+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:14:06.004397+00:00, confidence not recorded.
  - readme: https://github.com/Farama-Foundation/Gymnasium (fetched 2026-08-28T04:10:52.618399+00:00, sha 37f4efd9d95d)
  - homepage: https://gymnasium.farama.org (fetched 2026-08-29T08:11:00.085687+00:00, sha 922b0ecf7172)
  - registry_pypi: https://pypi.org/pypi/gymnasium/json (fetched 2026-08-29T08:11:00.094812+00:00, sha e442a1a21263)
- Data as of 2026-08-30T08:39:29.467469+00:00.