# Farama-Foundation/Arcade-Learning-Environment

A simple framework that allows researchers and hobbyists to develop AI agents for Atari 2600 games

Repository: https://github.com/Farama-Foundation/Arcade-Learning-Environment
Canonical: https://ross.abutalabs.com/products/arcade-learning-environment
Homepage: https://ale.farama.org/
Language: C++
License: GPL-2.0
License Family: copyleft
Last push: 2026-08-19T23:51:14+00:00

## Health v2 (maintenance only)
Score: 94/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 98, release rhythm 86, longevity 100
- inputs: {"age_days": 5101, "days_push": 14, "days_rel": 17, "gap_med": 70, "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 2448, forks 477 (observed 2026-08-28T04:06:52.787475+00:00)

## What it is
The Arcade Learning Environment (ALE) is a framework built on the Stella Atari 2600 emulator that lets researchers and hobbyists develop and evaluate AI agents across more than 100 Atari games. It exposes C++, Python (ale-py), Gymnasium, and WebAssembly interfaces with automatic score and game-over extraction, decoupled emulation from rendering, and packaged ROMs.

## Use cases
- train reinforcement learning agents on Atari games
- benchmark RL algorithms against classic Atari environments
- get an Atari environment for Gymnasium
- evaluate AI agents on 2600 game scores and terminal states
- run Atari emulation headless for fast RL training
- use multi-agent Atari environments for competitive agent research
- run Atari environments in the browser via WebAssembly

## When to choose
- You need the standard, widely-cited benchmark environment for Atari RL research
- You want native Gymnasium or OpenAI Gym-style integration with ROMs included via pip
- You need a fast, headless emulator core decoupled from rendering, with C++ vectorized stepping across multiple ROMs
- You need multi-language access to the same environments (C++, Python, WASM)

## When to avoid
- You want to build or play games rather than train agents on an emulated 2600
- You need environments for modern consoles, 3D games, or non-Atari benchmarks
- You need free-threaded CPython, which is unsupported due to OpenCV wheel limitations

## Facets
- artifact type: framework
- maturity: stable
- function: reinforcement-learning, simulation, machine-learning
- domain: reinforcement-learning, artificial-intelligence, machine-learning, gaming-tools
- platform: windows, cross-platform, python, cpp, wasm
- tags: atari-2600, emulation, gymnasium, rl-environment, benchmark, stella-emulator, ai-agents, roms, farama-foundation, nanobind, linux, macos

## Member repositories
- Farama-Foundation/Arcade-Learning-Environment (main) score 94

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:52.787475+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 2026-08-30T02:30:27.173414+00:00, confidence not recorded.
  - readme: https://github.com/Farama-Foundation/Arcade-Learning-Environment (fetched 2026-08-28T04:06:52.787475+00:00, sha c3d7a7815558)
  - homepage: https://ale.farama.org/ (fetched 2026-08-29T10:11:39.831801+00:00, sha b3efe2f7ebcb)
  - site_page: https://ale.farama.org/getting-started (fetched 2026-08-29T10:11:39.834407+00:00, sha 3cd2213db8cb)
  - site_page: https://ale.farama.org/faq (fetched 2026-08-29T10:11:39.836322+00:00, sha 3d8b746a7641)
- Data as of 2026-08-30T08:39:29.467469+00:00.