Farama-Foundation/Minigrid

Health v2 · maintenance only

Flags: no_license

Adoption not part of the score

2499 stars · 647 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Minigrid is a Python library of lightweight, easily configurable discrete grid-world environments for reinforcement learning research, following the Gymnasium API. It includes the original Minigrid environments plus BabyAI and WFC-generated environment suites with customizable wrappers and observation spaces.

Use cases

• train RL agents on gridworld navigation tasks
• benchmark reinforcement learning algorithms on simple environments
• create custom gridworld environments for RL research
• test hierarchical instruction-following agents with BabyAI tasks
• generate procedural mazes and dungeons with WFC environments
• prototype RL experiments without heavy simulation overhead

When to choose

• you need fast, lightweight RL environments with the Gymnasium API
• you want easily customizable gridworld tasks for research
• you need hierarchical or instruction-based missions like BabyAI
• you want a well-established benchmark cited in many publications

When to avoid

• you need photorealistic or 3D simulation environments
• you require official Windows support
• you need continuous control or robotics-style environments

Facets

library · maturity active

simulation machine-learning reinforcement-learning reinforcement-learning machine-learning simulation python gridworld gymnasium environments babyai rl-benchmarks research linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="Farama-Foundation/Minigrid")