AV-Phys Bench
Harbor: pendingDo Joint Audio-Video Generation Models Understand Physics?
Auto57%Multimodal
- Samples
- N/A
- Released
- 2026
- Scores
- 0
VA-Bench
VA-Bench VA-Bench evaluates embodied spatial intelligence through visual demonstrations, active perception, and metric robot control.
How to run
ltb · va-bench
$ npm i -g lettokenburn-cli$ ltb bench info va-bench
External harness. No Harbor adapter has been generated for this benchmark yet. `ltb bench info` prints the known links.
Runs happen on your machine with your own subscription; only the score summary is pushed. CLI docs
Scores
0 sourced entriesNo sourced scores recorded yet.
Every number links to where it was published. Low-confidence rows are muted and need per-cell verification.
Reproduced by the community
All runsLoading community runs…
About
VA-Bench
VA-Bench evaluates embodied spatial intelligence through visual demonstrations, active perception, and metric robot control. It contains 14 physics-simulated manipulation task families: 11 single-arm tasks and three dual-arm tasks.
Paper: VA-Bench Code and benchmark: github.com/zhangzhongbo2213/VABench Upstream simulator: RoboTwin Upstream assets: RoboTwin2.0
Contents
This repository distributes the versioned simulator assets needed by the 14 main…