University of Science and Technology of China · robot AI model
VLA-JEPA
China · 2026
Sources retrieved: 13 November 2026 (A, B, C, D)
- Model type
- Vision-language-action (VLA)Source and note: Model typeA arxiv.org/abs/2602.10098 · Retrieved 13 November 2026Note. Paper title: “VLA-JEPA: Enhancing Vision-Language-Action Model with Latent World Model”.
- What it controls
- Arms and handsSource and note: What it controlsB arxiv.org/html/2602.10098 · Retrieved 13 November 2026Note. Paper, real-world experiment: “we designed table-top manipulation tasks using a Franka Research 3 arm equipped with a Robotiq 2F-85 gripper”.
- Access
- Open weights (download)Source and note: AccessC github.com/ginwind/VLA-JEPA · Retrieved 13 November 2026Note. GitHub README: “Download the model checkpoints from Hugging Face: https://huggingface.co/ginwind/VLA-JEPA”
- Base model
- Qwen3-VL-2BSource and note: Base modelB arxiv.org/html/2602.10098 · Retrieved 13 November 2026Note. Paper appendix: “We use Qwen3-VL-2B as the VLM backbone”.
- Licence type
- Free, commercial useSource and note: Licence typeD huggingface.co/ginwind/VLA-JEPA · Retrieved 13 November 2026Note. Model card header: “license: apache-2.0”
Specifications
7 of 14 stated 7 not stated
5 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
Released10 February 2026
Source and note: ReleasedA arxiv.org/abs/2602.10098 · Retrieved 13 November 2026Note. arXiv listing: “Submitted on 10 Feb 2026”.
Input and output2/4 disclosed
Model typeVision-language-action (VLA)
Source and note: Model typeA arxiv.org/abs/2602.10098 · Retrieved 13 November 2026Note. Paper title: “VLA-JEPA: Enhancing Vision-Language-Action Model with Latent World Model”.
InputsNot stated
What it controlsArms and hands
Source and note: What it controlsB arxiv.org/html/2602.10098 · Retrieved 13 November 2026Note. Paper, real-world experiment: “we designed table-top manipulation tasks using a Franka Research 3 arm equipped with a Robotiq 2F-85 gripper”.
Degrees of freedom controlledNot stated
Size and speed0/2 disclosed
ParametersNot stated
Source and note: ParametersB arxiv.org/html/2602.10098 · Retrieved 13 November 2026Note. Paper appendix: “We use Qwen3-VL-2B as the VLM backbone”.
Control rateNot stated
Where it runs and what it is built on1/2 disclosed
On robot or cloudNot stated
Base modelQwen3-VL-2B
Source and note: Base modelB arxiv.org/html/2602.10098 · Retrieved 13 November 2026Note. Paper appendix: “We use Qwen3-VL-2B as the VLM backbone”.
Openness3/3 disclosed
AccessOpen weights (download)
Source and note: AccessC github.com/ginwind/VLA-JEPA · Retrieved 13 November 2026Note. GitHub README: “Download the model checkpoints from Hugging Face: https://huggingface.co/ginwind/VLA-JEPA”
Licence documentApache-2.0
Source and note: Licence documentD huggingface.co/ginwind/VLA-JEPA · Retrieved 13 November 2026Note. Model card header: “license: apache-2.0”
Licence typeFree, commercial use
Source and note: Licence typeD huggingface.co/ginwind/VLA-JEPA · Retrieved 13 November 2026Note. Model card header: “license: apache-2.0”
The developer’s own claims0/2 disclosed
Learns new tasksNot stated
Works across robotsNot stated
Source and note: Works across robotsA arxiv.org/abs/2602.10098 · Retrieved 13 November 2026Note. Paper, real-world experiment: “we designed table-top manipulation tasks using a Franka Research 3 arm equipped with a Robotiq 2F-85 gripper”.Yes 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
The paper lists “University of Science and Technology of China” as its first affiliation and “Zhongguancun Academy, Beijing, China” as its second.
The paper describes the method as: “We introduce VLA-JEPA, a JEPA-style pretraining framework”. The authors publish checkpoints for pre-training, LIBERO, SimplerEnv and a real-world setup.
Sources
One line per source, not per figure: a source has one retrieval date.
- Aarxiv.org/abs/2602.10098Main source. Retrieved 13 November 2026
- Barxiv.org/html/2602.10098Retrieved 13 November 2026
- Cgithub.com/ginwind/VLA-JEPARetrieved 13 November 2026
- Dhuggingface.co/ginwind/VLA-JEPARetrieved 13 November 2026
