What it measures
Language-conditioned dexterous tabletop manipulation by a humanoid with two dexterous hands, measured as average task success rate over trials.
SimulationActiveHigh relevance
GR-1 Tabletop Tasks
GR-1 Tabletop Tasks, a simulated benchmark introduced in GR00T N1: An Open Foundation Model for Generalist Humanoid Robots
GR-1 Tabletop Tasks is a simulated benchmark of 24 tabletop manipulation tasks for the Fourier GR-1 humanoid with dexterous hands. It was introduced in the GR00T N1 paper and focuses on dexterous hand control: 18 rearrangement tasks and 6 tasks that place objects into articulated containers such as cabinets, drawers and microwaves. The GR00T repository lists it among its simulation benchmarks.
- First published
- 2025
- Organizers
- NVIDIA
- Tasks
- 24
- Simulator
- —
- License
- —
How it works
The paper's protocol for simulation benchmarks reports the average success rate over 100 trials, following RoboCasa: the maximum score over the last five checkpoints, saved every 500 training steps. GR00T N1 is the model evaluated on these tasks. The GR00T repository notes that the benchmark's results were refreshed in its N1.7 release.
Why it matters
This is a humanoid-specific dexterous benchmark: the GR-1 with Fourier dexterous hands is a configuration that most arm-based benchmarks do not cover. It is listed in an actively updated NVIDIA repository.
Scoring
Metrics
| Metric | Key | Unit | Direction |
|---|---|---|---|
| Average task success rate (over 100 trials) | success_rate | Unitless | Higher is better ↑ |
Hardware
Robots involved
Fourier GR-1 humanoid with Fourier dexterous hands, per the GR00T N1 paper. The roster entry fourier-gr-1 is marked 'verify', so the model variant should be confirmed. The simulator is not named in the sections read.
Recorded outcomes
Results
No result files yet
This record describes the evaluation, but no verified structured results have been added.
Evidence
Sources
- [1]GR00T N1: An Open Foundation Model for Generalist Humanoid Robots ↗arXiv · accessed Oct 8, 2026
- [2]GR00T N1 (arXiv HTML, v2) ↗arXiv · accessed Oct 8, 2026
- [3]Isaac-GR00T (GitHub repository) ↗NVIDIA · accessed Oct 8, 2026
Related evaluations