Exclusive: RoboRank, the 'LeetCode for roboticists,' will open evaluation environments and run a public scoreboard
RoboRank, a LeetCode-like benchmarking project for robotics at roborank.dev, will separate and open-source its evaluation environments, target VLAs and world models, and harden sandboxed code execution so researchers can contribute tasks and models in the open.
By Ryan Merket · Published · Updated
RUNTIMEWIRE INVESTIGATION — Exclusive
Original reporting by RuntimeWire.

RoboRank, a LeetCode-like benchmarking project for robotics at roborank.dev, will separate and open-source its evaluation environments, target VLAs and world models, and harden sandboxed code execution so researchers can contribute tasks and models in the open.