About

About RL Research

A daily catalog of reinforcement-learning environments, AI benchmarks and model scores, with original research on who builds them and what they test.

RL Research tracks the companies and labs building environments for training AI agents, the benchmarks used to measure them, and how models score. Every environment, benchmark, model and category has its own page, and the whole catalog is rebuilt every morning.

The homepage shows the catalog as a slide under a microscope: each category is a colony, and each RL environment or AI benchmark is a cell in it. The research section adds original analysis of the RL environment economy.

These pages explain where the data comes from and how it is organized.

  1. How the data is built The public sources factored into the RL Research catalog, and how our pipeline cleans, merges, categorizes and rebuilds it every morning.
  2. Categories The categories RL Research uses to group RL environments and AI benchmarks, what each one covers, and how the pipeline builds and describes them.
  3. Authors Who writes the research, under public aliases.