LiberAI
About
About LiberAI
LiberAI is building foundation systems for general-purpose home robots around scalable human demonstration data. Its handheld UMI equipment records first-person imagery, gripper state and movement while a person completes household manipulation tasks. Demonstrations can be collected without occupying the final robot and later converted into an action representation suitable for policy training.
The company links this collection hardware with world models and embodied policies intended to transfer across robot platforms. Training, simulation and physical deployment form a closed loop: human sessions provide initial behaviour, the model is tested on a robot, and failures generate targeted collection or model updates. Founded in December 2025, LiberAI remains at the foundation-system and early hardware-scaling stage.
Articles
Funding
- 2026-06-11 - Pre-A - LiberAI raises hundreds of millions of RMB Pre-A for embodied intelligence models
LiberAI completed a Pre-A round worth several hundred million RMB led by Shunwei Capital for embodied-intelligence models and UMI data systems. - source: view.inews.qq.com
- 2026-05-15 - Angel+ - LiberAI raises approximately RMB 500M Angel+ round
LiberAI raised an Angel+ financing of approximately RMB 500 million. - source: finance.eastmoney.com
Investors
- Shunwei Capital - Lead - VC - 2 funding events - repeat backer - first seen 2026-05-15 - latest 2026-06-11
- ZhenFund - Participant - VC - 2 funding events - repeat backer - first seen 2026-05-15 - latest 2026-06-11
- Cathay Capital - Participant - VC - 1 funding event - first seen 2026-06-11
- HSG - Participant - Unknown - 1 funding event - first seen 2026-05-15
- Meituan DragonBall - Participant - Corporate / CVC - 1 funding event - strategic/corporate - first seen 2026-05-15
- Mufar Science and Technology Innovation - Participant - Unknown - 1 funding event - first seen 2026-06-11
- Sequoia China - Participant - Angel / Individual - 1 funding event - first seen 2026-06-11
- Yuanhe Origin - Participant - Angel / Individual - 1 funding event - first seen 2026-06-11
Team
- Liu Songming - Founder
Research
- LiberAI previews Liber-0 world model, topping RoboDojo benchmark
2026-09-17 - Research Publication - On RoboDojo's 42-task assessment Liber-0 Preview scored an average of 30.74 with a 25.52% mean success rate, against 11.41 and 6.91% for the pi-0.5 model. Liber-0 Preview completed long-horizon dexterous-hand manipulation on real robots, including a brunch task spanning more than 30 dexterous subtasks; the full Liber-0 release is to follow.
Source
Timeline
Events
- 2026-09-17 - Research Publication - LiberAI previews Liber-0 world model, topping RoboDojo benchmark
On RoboDojo's 42-task assessment Liber-0 Preview scored an average of 30.74 with a 25.52% mean success rate, against 11.41 and 6.91% for the pi-0.5 model. Liber-0 Preview completed long-horizon dexterous-hand manipulation on real robots, including a brunch task spanning more than 30 dexterous subtasks; the full Liber-0 release is to follow.
Source - 2026-09-17 - Funding - LiberAI completes Pre-A++ round as cumulative funding passes RMB1B
Investors in the round include Zhongguancun Science City, Dachen Caizhi, Gaoxin Capital and Huaying Capital. Backers across the five-month run include Sequoia China, ZhenFund, Meituan Longzhu and Shunwei Capital. LiberAI has pursued high-degree-of-freedom dexterous hands from the outset rather than single-degree-of-freedom grippers, an approach it frames as the route to human-level physical intelligence.
Source - 2026-06-11 - Funding - LiberAI raises hundreds of millions of RMB Pre-A for embodied intelligence models
The Pre-A round was led by Shunwei Capital. Other named investors included Cathay Innovation, Origin Ventures and Tsinghua Science Park Innovation. The round is linked to human UMI data, world models, UMI hardware, data collection and model training.
Source - 2026-05-15 - Funding - LiberAI raises approximately RMB 500M Angel+ round
Investors included HSG, ZhenFund, Meituan DragonBall and Shunwei. The amount is approximate and retained in native RMB rather than converted to an exact USD figure.
Source
Context
Comparable profiles
- Intelligence Factory
stack adjacency - Physical AI Training Data
- Lupa Robotics
stack adjacency - Physical AI Training Data
- LabyrinthAI
stack adjacency - Physical AI Training Data
- Ropedia
stack adjacency - Physical AI Training Data
- FORMOVE GmbH
stack adjacency - Physical AI Training Data
- Vision Lab
stack adjacency - Physical AI Training Data
- Sureform
stack adjacency - Physical AI Training Data
- Kognic
stack adjacency - Physical AI Training Data
- Worldmodeldata
stack adjacency - Physical AI Training Data
- MiraiBit
stack adjacency - Physical AI Training Data
- Algomatic Dynamics
stack adjacency - Physical AI Training Data
- JOYX
stack adjacency - Physical AI Training Data
Context tags
- Physical AI training data - Stack Layer - LiberAI works on UMI hardware, data collection and embodied model training systems.