---
license: apache-2.0
base_model:
- Qwen/Qwen2.5-VL-3B-Instruct
pipeline_tag: image-text-to-text
library_name: transformers
---
**EN** | [中文](README_CN.md) 

# SenseNova-SI: Scaling Spatial Intelligence with Multimodal Foundation Models

<a href="https://github.com/OpenSenseNova/SenseNova-SI" target="_blank">
    <img alt="Code" src="https://img.shields.io/badge/SenseNova_SI-Code-100000?style=flat-square&logo=github&logoColor=white" height="20" />
</a>
  <a href="https://arxiv.org/abs/2511.13719" target="_blank">
      <img alt="arXiv" src="https://img.shields.io/badge/arXiv-SenseNova_SI-red?logo=arxiv" height="20" />
  </a>
<a href="https://github.com/EvolvingLMMs-Lab/EASI" target="_blank">
    <img alt="Code" src="https://img.shields.io/badge/EASI-Code-100000?style=flat-square&logo=github&logoColor=white" height="20" />
</a>
<a href="https://easi.lmms-lab.com/leaderboard" target="_blank">
    <img alt="Leaderboard" src="https://img.shields.io/badge/%F0%9F%A4%97%20_EASI-Leaderboard-ffc107?color=ffc107&logoColor=white" height="20" />
</a>

## Overview
Despite remarkable progress, multimodal foundation models still exhibit surprising deficiencies in spatial intelligence.
In this work, we explore scaling up multimodal foundation models to cultivate spatial intelligence within the **SenseNova-SI family**,
built upon established multimodal foundations including visual understanding models (i.e., Qwen3-VL and InternVL3) and unified understanding and generation models (i.e., Bagel).
We take a principled approach to constructing high-performing and robust spatial intelligence by systematically curating SenseNova-SI-8M:
eight million diverse data samples under a rigorous taxonomy of spatial capabilities.
SenseNova-SI demonstrates unprecedented performance across a broad range of spatial intelligence benchmarks: 68.7% on VSI-Bench, 43.3% on MMSI, 85.6% on MindCube,
54.6% on ViewSpatial, and 50.1% on SITE, while maintaining strong general multimodal understanding (e.g., 84.9% on MMBench-En).
More importantly, we analyze the impact of data scaling, discuss early signs of emergent generalization capabilities enabled by diverse data training,
analyze the risk of overfitting and language shortcuts, present a preliminary study on spatial chain-of-thought reasoning, and validate the potential downstream application. SenseNova-SI is an ongoing project, and this report will be updated continuously.
All newly trained multimodal foundation models are publicly released to facilitate further research in this direction.
*In the future, SenseNova-SI will be integrated with larger-scale in-house models.*

## Release Information
Currently, we build SenseNova-SI upon popular open-source foundation models to maximize compatibility with existing research pipelines.
In this release, we present 
[**SenseNova-SI-1.2-InternVL3-8B**](https://huggingface.co/sensenova/SenseNova-SI-1.2-InternVL3-8B), [**SenseNova-SI-1.1-Qwen2.5-VL-3B**](https://huggingface.co/sensenova/SenseNova-SI-1.1-Qwen2.5-VL-3B), [**SenseNova-SI-1.1-Qwen2.5-VL-7B**](https://huggingface.co/sensenova/SenseNova-SI-1.1-Qwen2.5-VL-7B), and [**SenseNova-SI-1.1-Qwen3-VL-8B**](https://huggingface.co/sensenova/SenseNova-SI-1.1-Qwen3-VL-8B), of which **SenseNova-SI-1.2-InternVL3-8B** achieve state-of-the-art performance among open-source models of comparable size across eight recent spatial intelligence benchmarks: 
**VSI**, **MMSI**, **MindCube**, **ViewSpatial**, **SITE**, **BLINK**, **3DSRBench**, **EmbSpatial-Bench**.

<table>
  <thead>
    <tr>
      <th>Model</th>
      <th>VSI</th>
      <th>MMSI</th>
      <th>MindCube-Tiny</th>
      <th>ViewSpatial</th>
      <th>SITE</th>
    </tr>
  </thead>
  <tbody>
    <tr>
      <td colspan="6" align="center"><em>Open-source Models (~2B)</em></td>
    </tr>
    <tr>
      <td>InternVL3-2B</td><td>32.9</td><td>26.5</td><td>37.5</td><td>32.5</td><td>30.0</td>
    </tr>
    <tr>
      <td>Qwen2.5-VL-3B-Instruct</td><td>27.0</td><td>28.6</td><td>37.6</td><td>31.9</td><td>33.1</td>
    </tr>
    <tr>
      <td>Qwen3-VL-2B-Instruct</td><td>50.3</td><td>28.9</td><td>34.5</td><td>36.9</td><td>35.6</td>
    </tr>
    <tr>
      <td>MindCube-3B-RawQA-SFT</td><td>17.2</td><td>1.7</td><td>51.7</td><td>24.1</td><td>6.3</td>
    </tr>
    <tr>
      <td>SpatialLadder-3B</td><td>44.8</td><td>27.4</td><td>43.4</td><td>39.8</td><td>27.9</td>
    </tr>
    <tr>
      <td>SpatialMLLM-4B</td><td>46.3</td><td>26.1</td><td>33.4</td><td>34.6</td><td>18.0</td>
    </tr>
    <tr>
      <td>VST-3B-SFT</td><td><strong>57.9</strong></td><td>30.2</td><td>35.9</td><td><strong>52.8</strong></td><td>35.8</td>
    </tr>
    <tr>
      <td>Cambrian-S-3B</td><td>57.3</td><td>25.2</td><td>32.5</td><td>39.0</td><td>28.3</td>
    </tr>
    <tr>
      <td><strong>SenseNova-SI-1.1-Qwen2.5-VL-3B</strong></td>
      <td>54.9</td>
      <td><strong>30.8</strong></td>
      <td><strong>52.6</strong></td>
      <td>43.5</td>
      <td><strong>37.8</strong></td>
    </tr>
    <tr>
      <td colspan="6" align="center"><em>Proprietary Models</em></td>
    </tr>
    <tr>
      <td>Gemini-2.5-pro-2025-06</td><td>53.5</td><td>38.0</td><td>57.6</td><td>46.0</td><td>57.0</td>
    </tr>
    <tr>
      <td>Grok-4-2025-07-09</td><td>47.9</td><td>37.8</td><td>63.5</td><td>43.2</td><td>47.0</td>
    </tr>
    <tr>
      <td>GPT-5-2025-08-07</td><td>55.0</td><td>41.8</td><td>56.3</td><td>45.5</td><td>61.8</td>
    </tr>
  </tbody>
</table>

## Evaluation

To reproduce the benchmark results above, please refer to [EASI](https://easi.lmms-lab.com/leaderboard/) to evaluate SenseNova-SI on mainstream spatial intelligence benchmarks.


## Citation

```bib
@InProceedings{sensenova-si,
  title = {Scaling Spatial Intelligence with Multimodal Foundation Models},
  author = {Cai, Zhongang and Wang, Ruisi and Gu, Chenyang and Pu, Fanyi and Xu, Junxiang and Wang, Yubo and Yin, Wanqi and Yang, Zhitao and Wei, Chen and Sun, Qingping and Zhou, Tongxi and Li, Jiaqi and Pang, Hui En and Qian, Oscar and Wei, Yukun and Lin, Zhiqian and Shi, Xuanke and Deng, Kewang and Han, Xiaoyang and Chen, Zukai and Fan, Xiangyu and Deng, Hanming and Lu, Lewei and Pan, Liang and Li, Bo and Liu, Ziwei and Wang, Quan and Lin, Dahua and Yang, Lei},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2026}
}
```