oro-ai/orobench-trajectories
Viewer • Updated • 5.51k • 125 • 1
ORO Bench
ORO’s human-in-the-loop automated research competition.
Note 90 shopping tasks, 10 sampled agent versions and 1,800 recorded episodes from 20 successful runs. Public research archive with historical grading checks and outcomes. Website: https://oroagents.com