LiberAI Launches Liber-0, a Groundbreaking World Action Model for Robot Manipulation
- Published
- Source
- TheRobotReport.com
LiberAI has unveiled Liber-0, a world action model pretrained on over 2.2 million hours of human experience, designed to facilitate zero-shot robot manipulation. In a test involving 60 tasks proposed by GPT-6, Liber-0 achieved an overall success rate of 36.0%, demonstrating its capability to perform tasks without specific robot demonstrations. The model integrates large-scale human video data, glove-recorded hand movements, and embodiment alignment to effectively transfer human manipulation skills to robots.
The significance of Liber-0 lies in its potential to revolutionize robot manipulation, which has traditionally depended on task-specific demonstrations. By leveraging abundant human activity data, Liber-0 aims to reduce the time and costs associated with collecting robot-specific training data. The model's performance on various tasks, including pick-and-place and complex manipulation, highlights its versatility and adaptability in real-world applications.
LiberAI's evaluation of Liber-0 included 600 trials across 60 tasks, with notable success rates of 63.5% for pick-and-place tasks and 34.0% for non-pick-and-place tasks. Future assessments will focus on further refining the model's capabilities and exploring its application in diverse robotic platforms, although no specific timeline for these developments was disclosed at the time of publication.
Original report
LiberAI Unveils Liber-0, a World Action Model That Grounds Human Experience in Robot Action
TheRobotReport.com · Sponsored ContentThis briefing is an independently written summary based on publicly available reporting and is provided for industry information and news discovery. The original report and source publication are credited and linked where applicable. RobotToday does not claim ownership of third-party source material.
Rights concerns? Contact [email protected] with the relevant URL and details. We will review the matter and take appropriate action where warranted.
More related news
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
- 11
-
12
RobotToday Weekly October 26 – 30, 2026 RobotToday · 1 week ago
Related Suppliers
Bifrost AI
United States
Bifrost AI builds simulation infrastructure for physical AI: the Manifold robot-policy evaluation platform and the Stardust synthetic 3D sensor data platform.
Mabo Engineering & Automation
Belgium
Belgian AGV and AMR builder in Lier making automated forklifts, VNA shuttles and lift robots with its own AGV Master fleet software
Intelligence Factory
United States
New York startup founded 2026 (YC P26) training robot manipulation foundation models on human demonstration data recorded with custom sensing gloves.
FiveAges 中科第五纪
China
Embodied-AI startup founded in 2024 by CASIA-trained Liu Nianfeng; its few-shot FAM manipulation models run on its own wheeled dual-arm robot and Unitree G1.
commissure Inc.
Japan
University of Tokyo spin-off founded 2023 in Tokyo making tactile sensing, haptic devices and touch-data tools for robot imitation learning; NEDO NEP pick 2026.
Goddard Technologies
United States
US product development firm providing design, engineering and prototyping for robotic systems, automation and medical devices.
Cyphra Autonomy
United States
US start-up renting PAL autonomous haulers that move rebar, fencing and debris on jobsites, at about USD 1,800 per unit per month.
Meinong Robot Co., Limited 美农机器人
Hong Kong
Hong Kong Science Park startup founded 2023 developing AI fruit-picking robots for grapes, tree fruit and melons, plus transport robots and a 4WD-4WS chassis.
Instinct Robotics
United States
Carnegie Mellon spinout in Pittsburgh building physical-AI foundation models that let industrial robots grind, polish and weld high-mix parts.
Matrix Robotics 矩阵超智能
China
Chinese developer of general-purpose humanoid robots, with flagship MATRIX-3 and MATRIX-1 driven by embodied AI.
General Intuition
United States
Frontier research lab building world and action foundation models for spatial-temporal reasoning in virtual and physical environments, including robots.
ShengShu Technology 生数科技
China
Beijing-founded AI company building general world models; its Motubrain product is a world-action model for physical robot control and embodied intelligence.
Robotics needs a service framework.
RSF defines a common language for robot service capability, lifecycle operations, certification pathways, and service-provider networks.
Join the RobotToday community on LinkedIn
Daily robotics news, in-depth analysis, conference highlights, and discussions with professionals worldwide.
