Tongji University · robot AI model
Boundless-World-Model
China · 2026
Sources retrieved: 19 November 2026 (A, B, C, D)
- Model type
- World modelSource and note: Model typeB huggingface.co/BLM-Lab/Boundless-World-Model/raw/main/README.md · Retrieved 19 November 2026Note. Model card: “BWM: Physically consistent, action-conditioned video world model for robotic manipulation”
- Inputs
- Camera Robot stateSource and note: InputsA raw.githubusercontent.com/boundless-large-model/boundless-world-model/main/README.md · Retrieved 19 November 2026Note. GitHub README: “BWM autoregressively predicts future observation chunks from the initial observation, dynamic history, and action chunk.”
- Access
- Open weights (download)Source and note: AccessA raw.githubusercontent.com/boundless-large-model/boundless-world-model/main/README.md · Retrieved 19 November 2026Note. Weights are published on Hugging Face (BLM-Lab/Boundless-World-Model): “hf download BLM-Lab/Boundless-World-Model step-12000.safetensors --local-dir ckpt/BLM”
- Base model
- Wan2.2-TI2V-5BSource and note: Base modelB huggingface.co/BLM-Lab/Boundless-World-Model/raw/main/README.md · Retrieved 19 November 2026Note. Model card: “Wan2.2-TI2V-5B”

Specifications
5 of 14 stated 9 not stated
4 fields hidden 5 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
Released31 July 2026
Source and note: ReleasedD arxiv.org/abs/2607.29302 · Retrieved 19 November 2026Note. arXiv abstract page: “Submitted on 31 Jul 2026”
Input and output2/4 disclosed
Model typeWorld model
Source and note: Model typeB huggingface.co/BLM-Lab/Boundless-World-Model/raw/main/README.md · Retrieved 19 November 2026Note. Model card: “BWM: Physically consistent, action-conditioned video world model for robotic manipulation”
InputsCamera Robot state
Source and note: InputsA raw.githubusercontent.com/boundless-large-model/boundless-world-model/main/README.md · Retrieved 19 November 2026Note. GitHub README: “BWM autoregressively predicts future observation chunks from the initial observation, dynamic history, and action chunk.”
What it controlsNot stated
Source and note: What it controlsA raw.githubusercontent.com/boundless-large-model/boundless-world-model/main/README.md · Retrieved 19 November 2026Note. As a world simulator, BWM outputs predicted future observation video frames rather than robot control actions: “BWM autoregressively predicts future observation chunks from the initial observation, dynamic history, and action chunk.”
Degrees of freedom controlledNot stated
Size and speed0/2 disclosed
ParametersNot stated
Source and note: ParametersB huggingface.co/BLM-Lab/Boundless-World-Model/raw/main/README.md · Retrieved 19 November 2026Note. Built upon “Wan2.2-TI2V-5B” (5B-parameter base model) with an additional action encoder.
Control rateNot stated
Where it runs and what it is built on1/2 disclosed
On robot or cloudNot stated
Base modelWan2.2-TI2V-5B
Source and note: Base modelB huggingface.co/BLM-Lab/Boundless-World-Model/raw/main/README.md · Retrieved 19 November 2026Note. Model card: “Wan2.2-TI2V-5B”
Openness1/3 disclosed
AccessOpen weights (download)
Source and note: AccessA raw.githubusercontent.com/boundless-large-model/boundless-world-model/main/README.md · Retrieved 19 November 2026Note. Weights are published on Hugging Face (BLM-Lab/Boundless-World-Model): “hf download BLM-Lab/Boundless-World-Model step-12000.safetensors --local-dir ckpt/BLM”
Licence documentNot stated
Source and note: Licence documentB huggingface.co/BLM-Lab/Boundless-World-Model/raw/main/README.md · Retrieved 19 November 2026Note. The Hugging Face repository for the weights carries no licence. The GitHub code repository has a LICENSE file headed “Apache License”, which covers the code.
Licence typeNot stated
Source and note: Licence typeC arxiv.org/pdf/2607.29302 · Retrieved 19 November 2026Note. Paper describes BWM as “an open-source, low-cost, high-fidelity world simulator for robot manipulation”, but the Hugging Face weights repository carries no licence.
The developer’s own claims0/2 disclosed
Learns new tasksNot stated
Works across robotsNot stated
Source and note: Works across robotsYes only if the developer itself says the model runs on different robot models, for example robots from two manufacturers or two of one manufacturer’s models. Several units of the same robot do not count.
Notes
Developed by BWM Team (BLM-Lab) at Tongji University. “BWM: Physically consistent, action-conditioned video world model for robotic manipulation”
Inference code was released in May 2026 and the technical report was submitted to arXiv on 31 July 2026.
Sources
One line per source, not per figure: a source has one retrieval date.
- Araw.githubusercontent.com/boundless-large-model/boundless-world-model/main/README.mdMain source. Retrieved 19 November 2026
- Bhuggingface.co/BLM-Lab/Boundless-World-Model/raw/main/README.mdRetrieved 19 November 2026
- Carxiv.org/pdf/2607.29302Retrieved 19 November 2026
- Darxiv.org/abs/2607.29302Retrieved 19 November 2026

