Gym Retro (https://github.com/openai/retro) is a relatively new (released on May 25, 2018) Python library released by OpenAI (https://blog.openai.com/gym-retro/) as a research platform for developing reinforcement learning algorithms for game playing. Although the Atari suite of 60+ games was available in OpenAI Gym, the total number of games available was limited. Gym Retro supports the use of games developed for several console/retro gaming platforms, such as Nintendo's NES, SNES, Game Boy consoles, Sega Genesis, and Sega Master System to name a few. This is made possible with the use of emulators using the Libretro API:
Gym ...
