Project AURIONAI Use-Cases Research & Innovation

LIBERO

Simulated robot manipulation task suites; average success rate

world models and roboticssaturated

Score in %, OpenVLA-OFT paper

  1. 1Stanford UniversityOpenVLA-OFT97.1 %vendor-reported
  2. 2Stanford UniversityOpenVLA76.5 %vendor-reported
3scores
3models
0independent
0maintainer-run
3vendor-reported

Definition

Version
not in the notes
Category
world models and robotics
Maintainer
not in the notes
Status
saturated
Home page
not in the notes

Sources

  1. Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Success arxiv.org https://arxiv.org/abs/2502.19645
  2. AGIBOT releases GO-2 foundation model for embodied AI - The Robot Report therobotreport.com https://www.therobotreport.com/agibot-releases-go-2-foundation-model-embodied-ai/

Scores at a Glance

Score in %, OpenVLA-OFT paper

Higher is better for this unit.

1Stanford UniversityOpenVLA-OFTStanford UniversityAverage
97.1 %
vendor-reported2Stanford UniversityOpenVLAStanford University, UC Berkeley, Toyota Research Institute (TRI)Average; baseline reported in OpenVLA-OFT paper
76.5 %
vendor-reported
0255075100
Where this comes from
Each bar is one model, with its most independent recorded score. Hatched bars are vendor-reported; the word beside each bar says who ran it. Settings and dates differ between rows; the Scores table below has every result.

Scores

3 of 3 scores shown libero-scores.csv
ModelScoreWho ran itSettingView or populationDateSource
AgiBot (Zhiyuan)GO-298.5 %vendor-reportedAverageAgiBot GO-2 releaseApr 2026therobotreport.com
Stanford UniversityOpenVLA-OFT97.1 %vendor-reportedAverageOpenVLA-OFT paperFeb 2025arxiv.org
Stanford UniversityOpenVLA76.5 %vendor-reportedAverage; baseline reported in OpenVLA-OFT paperOpenVLA-OFT paperFeb 2025arxiv.org
Where this comes from
Scores from different settings, harnesses or dates are not directly comparable; the row says which. Sorted by score, highest first, as a reading aid, not a ranking.

To cite this page: Project AURION, AI Use-Cases Research & Innovation, "LIBERO", https://aurion.semai.id/benchmarks/libero/, records of October 2026.