HKUST(GZ) & OLA Dimensions · robot AI model
UniWAM
China · 2026
Runs on PiPER by AgileX RoboticsSource and note: PiPERA arxiv.org/abs/2610.02054 · Retrieved 19 November 2026
Sources retrieved: 19 November 2026 (A, B, C, D)
- Model type
- World action modelSource and note: Model typeB raw.githubusercontent.com/UniWAM/UniWAM/main/README.md · Retrieved 19 November 2026Note. Repository title: “# UniWAM: Unified World-Action Model”
- What it controls
- Arms and handsSource and note: What it controlsA arxiv.org/abs/2610.02054 · Retrieved 19 November 2026Note. Real-world evaluation: “Experiments are conducted on the AgileX Piper platform, where each arm has 6-DoF joints and a parallel gripper. Our dual-arm system consists of two Piper”.
- Inputs
- Camera Language Robot stateSource and note: InputsA arxiv.org/abs/2610.02054 · Retrieved 19 November 2026Note. Paper: “Given the current observation ot, proprioceptive state st, and task instruction I, the model learns the joint distribution of language outputs, future observations, and actions”. Real-world cameras: “each equipped with a wrist-mounted camera, together with an additional third-person camera providing a global view of the workspace.”
- Access
- Open weights (download)Source and note: AccessB raw.githubusercontent.com/UniWAM/UniWAM/main/README.md · Retrieved 19 November 2026Note. UniWAM-base and UniWAM-robotwin-clean weights are published on Hugging Face and ModelScope (“pretained checkpoint on mixed robot,human and VQA datasets.”).
- Parameters
- 8BSource and note: ParametersB raw.githubusercontent.com/UniWAM/UniWAM/main/README.md · Retrieved 19 November 2026Note. README: “UniWAM is an 8B-parameter model that brings semantic understanding, visual prediction, and action generation into one MoT architecture.”
- Base model
- Qwen3-VL-2B-Instruct & Wan2.2-TI2V-5BSource and note: Base modelB raw.githubusercontent.com/UniWAM/UniWAM/main/README.md · Retrieved 19 November 2026Note. Pretrained backbones: “Qwen/Qwen3-VL-2B-Instruct” and “Wan-AI/Wan2.2-TI2V-5B”.
- Licence type
- Free, commercial useSource and note: Licence typeC huggingface.co/chenpyyy/UniWAM-base · Retrieved 19 November 2026Note. UniWAM-base model card: “license: apache-2.0”

Specifications
10 of 14 stated 4 not stated
2 fields hidden 2 not-stated fields with a note stay ·
How to read the figuresShow guideClose
- not statedNo source
- 0Reported as zero
- ASource
Select a source marker for source details and notes.
Release1/1 disclosed
Input and output3/4 disclosed
Size and speed1/2 disclosed
Where it runs and what it is built on1/2 disclosed
Openness3/3 disclosed
The developer’s own claims1/2 disclosed
Notes
Developers describe UniWAM: “UniWAM is an 8B-parameter model that brings semantic understanding, visual prediction, and action generation into one MoT architecture.”
Core contributors are affiliated with “The Hong Kong University of Science and Technology (Guangzhou)” and “OLA Dimensions”.
Sources
One line per source, not per figure: a source has one retrieval date.
- Aarxiv.org/abs/2610.02054Retrieved 19 November 2026
- Braw.githubusercontent.com/UniWAM/UniWAM/main/README.mdMain source. Retrieved 19 November 2026
- Chuggingface.co/chenpyyy/UniWAM-baseRetrieved 19 November 2026
- Darxiv.org/abs/2610.02054v1Retrieved 19 November 2026



