Meta-World

An open benchmark of simulated robotic manipulation tasks for developing and evaluating multi-task and meta reinforcement learning algorithms.

Made by Farama Foundation.

Licenses

Code: MIT (opens in a new tab)
Read by Robotheca from LICENSE (opens in a new tab) (MIT) at commit c0dafaf (opens in a new tab), on 16 Oct 2026.
Commercial use: Yes

Provides continuous-control robot manipulation environments for reinforcement learning research.
Offers MT1, MT10 and MT50 multi-task benchmarks, where MT10 and MT50 append one-hot task IDs to observations.
Offers ML1, ML10 and ML45 meta-learning benchmarks with separate train and test environments.
Follows the Gymnasium API for creating and interacting with environments.
Supports synchronous and asynchronous vectorized environments for running many tasks at once.
Allows building custom benchmarks by combining any set of its environments.

Written by Robotheca from the project's README (opens in a new tab) at commit c0dafaf.

Checked on GitHub on 16 Oct 2026.

No robot on Robotheca is linked to it yet. Links come from a robot's robot.yaml, from its dependency files read as text, or from Robotheca's editors.

Robotheca lists software as a reference: nothing is hosted or run here, and every fact links to where it was read.