---
title: lingbot-vla-v2-6b
canonical_url: "https://www.modelscope.cn/models/Robbyant/lingbot-vla-v2-6b"
md_url: "https://www.modelscope.cn/models/Robbyant/lingbot-vla-v2-6b.md"
repository: Robbyant/lingbot-vla-v2-6b
chinese_name: lingbot-vla-v2-6b
last_updated: 2026-07-07
pipeline_tag: robotics
tasks:
  - robotics
base_model_relation: finetune
parameters: 6.4B
tensor_type:
  - F32
library_name:
  - safetensors
  - pytorch
frameworks:
  - Pytorch
downloads: 1275
stars: 12
---

# lingbot-vla-v2-6b

> lingbot-vla-v2-6b - Robbyant 在 ModelScope 开源的模型。LingBot-VLA-V2: From Foundation to Application

- **Repository**: Robbyant/lingbot-vla-v2-6b
- **Tasks**: robotics
- **Parameters**: 6.4B
- **Downloads**: 1275
- **Stars**: 12
- **Last updated**: 2026-07-07

Source: https://www.modelscope.cn/models/Robbyant/lingbot-vla-v2-6b

---

# LingBot-VLA-V2: From Foundation to Application

<p align="center">
  <img src="assets/lingbot_vla2_framework.png" width="100%">
</p>

---

**LingBot-VLA-V2** is a practical Vision-Language-Action foundation model designed to move from large-scale pre-training toward reliable real-world robot applications.

Compared with LingBot-VLA, V2 improves three core capabilities:

- **Generalization across tasks and embodiments**: a redesigned data pipeline curates around **60,000 hours** of pre-training data, including **50,000 hours** of robot trajectories across **20 robot configurations** and **10,000 hours** of egocentric human videos.
- **Expanded action space**: the unified representation supports arms, end-effectors, grippers, dexterous hands, waist, head, and mobile-base signals instead of only standard dual-arm manipulation.
- **Predictive dynamics modeling**: future prediction is used as a proxy task, with DINO-Video providing semantic temporal priors and LingBot-Depth providing geometric cues.


### MoE Action Expert

To improve cross-embodiment scaling, LingBot-VLA-V2 uses sparse MoE layers inside the action expert. Fine-grained expert segmentation and shared expert isolation allow universal priors and specialized embodiment/task patterns to coexist under the same active compute budget.

<p align="center">
  <img src="assets/lingbot_vla2_loss_mse_comparison.png" width="90%">
</p>

### Dual-Query Distillation

LingBot-VLA-V2 appends current and future perceptual queries to the visual/text tokens. These queries are distilled from LingBot-Depth and DINO-Video, encouraging causal inference to capture both current scene geometry and future scene evolution.

<p align="center">
  <img src="assets/lingbot_vla2_vis_distillation.png" width="100%">
</p>

---

## Model Sources

- Paper: From Foundation to Application: Improving VLA Models in Practice
- Repository: https://github.com/robbyant/lingbot-vla-v2 
- Project Page: https://technology.robbyant.com/lingbot-vla-v2


---

## License Agreement
This project is licensed under the [Apache-2.0 License](LICENSE).

## Acknowledgement
This codebase is builded on the [VeOmni](https://arxiv.org/abs/2508.02317) project. Thanks for their excellent work!