# openai/universe

Universe: a software platform for measuring and training an AI's general intelligence across the world's supply of games, websites and other applications.

Repository: https://github.com/openai/universe
Canonical: https://ross.abutalabs.com/products/universe
Homepage: https://universe.openai.com
Language: Python
License: MIT
License Family: permissive
Archived: true
Last push: 2018-04-05T18:58:15+00:00

## Health v2 (maintenance only)
Score: 10/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 3559, "days_push": 3072, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: archived
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 7494, forks 954 (observed 2026-08-28T04:10:00.298332+00:00)

## What it is
Universe is an OpenAI library that wraps games, websites, and other applications into OpenAI Gym environments so AI agents can be trained and evaluated on them via keyboard/mouse events and screen pixels. It packages programs into Docker containers and exposes over 1,000 environments, many with reward signals for reinforcement learning.

## Use cases
- train reinforcement learning agents on games and browser tasks
- turn any desktop application into a Gym environment
- benchmark AI general intelligence across diverse environments
- run Atari and Flash game environments over VNC
- evaluate agents on World of Bits browser tasks

## When to choose
- you need the original Universe VNC-based environments for reproducing historical research
- you want to wrap arbitrary GUI applications as Gym environments

## When to avoid
- starting new reinforcement learning projects - use openai/retro instead
- you need maintained software or modern Python support
- you need Windows support or Python beyond 2.7/3.5

## Facets
- artifact type: library
- maturity: abandoned
- function: reinforcement-learning, simulation, machine-learning, testing
- domain: reinforcement-learning, artificial-intelligence, gaming-tools, machine-learning
- platform: python
- tags: openai-gym, vnc, reinforcement-learning-environments, deprecated, docker-environments, agent-training, linux, macos, docker

## Member repositories
- openai/universe (main) score 10

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:00.298332+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 2026-08-29T17:37:34.116633+00:00, confidence not recorded.
  - readme: https://github.com/openai/universe (fetched 2026-08-28T04:10:00.298332+00:00, sha 8d360247f46d)
  - registry_pypi: https://pypi.org/pypi/universe/json (fetched 2026-08-29T08:32:50.983482+00:00, sha 44d55548c967)
- Data as of 2026-08-30T08:39:29.467469+00:00.