Benchmark Almanac

Robocurve

We track 7 benchmarks that Robocurve made or helped make.

0 in use 4 with public code 4 with a leaderboard here

Benchmarks

Name Kind Released Best score Status
Roboharm Agents and VLAs controlling bimanual YAM arms get five harmful-looking instructions in fixed scenes, such as stabbing a doll. Reviewers label refusals and completions. Embodied reasoning Sep 2026 20% System cards only
DataCenterBench Tier 1 Frontier models watch through cameras and drive two real robot arms on tabletop tasks rebuilt from data-center equipment manuals. LLM judges score each try from photos. Real robots Oct 2026 40% Not in cards
Perceptual Attacks on Embodied AI (Robocurve) Frontier models drive two real robot arms on five unsafe tasks with unaltered, blurred or doctored camera footage. An AI judge labels attempts and refusals. Jailbreaks and prompt injection Oct 2026 0% Not in cards
Robocurve simple tasks (LLM control of YAM arms) Frontier LLMs control bimanual I2RT YAM arms through Inspect Robots tools on block-into-bowl, puzzle, push-ball and stack-bowls. Staged scores, 20 trials. Real robots Sep 2026 52.5% Not in cards
StationeryBench Robots use both arms to handle common desk supplies. A human grades progress from zero to 100 on five tasks. Real robots Sep 2026 46.0 Not in cards
Robocurve physical tool-use benchmark (unidentified) Unnamed Robocurve physical tool-use robot task. Real robots Not in cards
Robot Capabilities Index (Robocurve) An index fitted like Epoch ECI over five RoboDojo-RC Tier 1 tasks run by frontier LLMs on real YAM arms. Anchored at GPT-5.6 Sol minus 3 and GPT-6 Astra plus 3. Real robots Oct 2026 Not in cards