Video-Reason/VBVR-Wan2.2
13260
1---2base_model:3- Wan-AI/Wan2.2-I2V-A14B-Diffusers4library_name: diffusers5license: apache-2.06pipeline_tag: image-to-video7---8 9# VBVR: A Very Big Video Reasoning Suite10 11<a href="https://video-reason.com" target="_blank">12 <img alt="Project Page" src="https://img.shields.io/badge/Project%20-%20Homepage-4285F4" height="20" />13</a>14<a href="https://github.com/Video-Reason/VBVR-EvalKit" target="_blank">15 <img alt="Code" src="https://img.shields.io/badge/Evaluation_code-VBVR_Bench-100000?style=flat-square&logo=github&logoColor=white" height="20" />16</a>17<a href="https://github.com/Video-Reason/VBVR-Wan2.2" target="_blank">18 <img alt="Code" src="https://img.shields.io/badge/Training_code-VBVR_Wan2.2-100000?style=flat-square&logo=github&logoColor=white" height="20" />19</a>20<a href="https://github.com/Video-Reason/VBVR-DataFactory" target="_blank">21 <img alt="Code" src="https://img.shields.io/badge/Data_code-VBVR_DataFactory-100000?style=flat-square&logo=github&logoColor=white" height="20" />22</a>23<a href="https://huggingface.co/papers/2602.20159" target="_blank">24 <img alt="arXiv" src="https://img.shields.io/badge/arXiv-VBVR-red?logo=arxiv" height="20" />25</a>26<a href="https://huggingface.co/datasets/Video-Reason/VBVR-Dataset" target="_blank">27 <img alt="Dataset" src="https://img.shields.io/badge/%F0%9F%A4%97%20_VBVR_Dataset-Data-ffc107?color=ffc107&logoColor=white" height="20" />28</a>29<a href="https://huggingface.co/datasets/Video-Reason/VBVR-Bench-Data" target="_blank">30 <img alt="Bench Data" src="https://img.shields.io/badge/%F0%9F%A4%97%20_VBVR_Bench-Data-ffc107?color=ffc107&logoColor=white" height="20" />31</a>32<a href="https://huggingface.co/spaces/Video-Reason/VBVR-Bench-Leaderboard" target="_blank">33 <img alt="Leaderboard" src="https://img.shields.io/badge/%F0%9F%A4%97%20_VBVR_Bench-Leaderboard-ffc107?color=ffc107&logoColor=white" height="20" />34</a>35 36🔥Please check out our newly released [**VBVR-Wan2.1**](https://huggingface.co/Video-Reason/VBVR-Wan2.1) (Diffusers format),37[**VBVR-Wan2.1-diffsynth**](https://huggingface.co/Video-Reason/VBVR-Wan2.1-diffsynth) (DiffSynth LoRA format), and38[**VBVR-LTX2.3-diffsynth**](https://huggingface.co/Video-Reason/VBVR-LTX2.3-diffsynth) (DiffSynth LoRA format; Diffusers does not yet support LTX-Video-2.3, so only the DiffSynth LoRA format is released for this model).39 40## Overview41Video reasoning grounds intelligence in spatiotemporally consistent visual environments that go beyond what text can naturally capture, 42enabling intuitive reasoning over motion, interaction, and causality. Rapid progress in video models has focused primarily on visual quality. 43Systematically studying video reasoning and its scaling behavior suffers from a lack of video reasoning (training) data. 44 45To address this gap, we introduce the Very Big Video Reasoning (VBVR) Dataset, an unprecedentedly large-scale resource spanning 200 curated reasoning tasks 46and over one million video clips—approximately three orders of magnitude larger than existing datasets. We further present VBVR-Bench, 47a verifiable evaluation framework that moves beyond model-based judging by incorporating rule-based, human-aligned scorers, 48enabling reproducible and interpretable diagnosis of video reasoning capabilities. 49 50Leveraging the VBVR suite, we conduct one of the first large-scale scaling studies of video reasoning and observe early signs of emergent generalization 51to unseen reasoning tasks. **Together, VBVR lays a foundation for the next stage of research in generalizable video reasoning.**52 53The model was presented in the paper [A Very Big Video Reasoning Suite](https://huggingface.co/papers/2602.20159).54 55## Models Zoo56 57| Model | Base Architecture | Other Remarks |58|-------|-------------------|---------------|59| [VBVR-Wan2.1](https://huggingface.co/Video-Reason/VBVR-Wan2.1) | Wan2.1-I2V-14B-720P | Diffusers format |60| [**VBVR-Wan2.2**](https://huggingface.co/Video-Reason/VBVR-Wan2.2) | Wan2.2-I2V-A14B | Diffusers format |61| [VBVR-Wan2.1-diffsynth](https://huggingface.co/Video-Reason/VBVR-Wan2.1-diffsynth) | Wan2.1-I2V-14B-720P | DiffSynth LoRA format |62| [VBVR-Wan2.2-diffsynth](https://huggingface.co/Video-Reason/VBVR-Wan2.2-diffsynth) | Wan2.2-I2V-A14B | DiffSynth LoRA format |63| [VBVR-LTX2.3-diffsynth](https://huggingface.co/Video-Reason/VBVR-LTX2.3-diffsynth) | LTX-Video-2.3 | DiffSynth LoRA format |64 65## Release Information66VBVR-Wan2.2 is trained from Wan2.2-I2V-A14B without architectural modifications, as the goal of VBVR-Wan2.2 is to *investigate data scaling behavior* and provide a *strong baseline model* for the video reasoning research community. Leveraging the VBVR-Dataset, which constitutes one of the largest video reasoning datasets to date, VBVR-Wan2.2 achieved highest score on VBVR-Bench.67 68In this release, we present 69[**VBVR-Wan2.2**](https://huggingface.co/Video-Reason/VBVR-Wan2.2), 70[**VBVR-Dataset**](https://huggingface.co/datasets/Video-Reason/VBVR-Dataset),71[**VBVR-Bench-Data**](https://huggingface.co/datasets/Video-Reason/VBVR-Bench-Data) and 72[**VBVR-Bench-Leaderboard**](https://huggingface.co/spaces/Video-Reason/VBVR-Bench-Leaderboard).73 74<table>75 <tr>76 <th>Model</th>77 <th>Overall</th>78 <th>ID</th>79 <th>ID-Abst.</th>80 <th>ID-Know.</th>81 <th>ID-Perc.</th>82 <th>ID-Spat.</th>83 <th>ID-Trans.</th>84 <th>OOD</th>85 <th>OOD-Abst.</th>86 <th>OOD-Know.</th>87 <th>OOD-Perc.</th>88 <th>OOD-Spat.</th>89 <th>OOD-Trans.</th>90 </tr>91 <tbody>92 <tr>93 <td><strong>Human</strong></td>94 <td>0.974</td><td>0.960</td><td>0.919</td><td>0.956</td><td>1.00</td><td>0.95</td><td>1.00</td>95 <td>0.988</td><td>1.00</td><td>1.00</td><td>0.990</td><td>1.00</td><td>0.970</td>96 </tr>97 <tr style="background:#F2F0EF;font-weight:700;text-align:center;">98 <td colspan="14"><em>Open-source Models</em></td>99 </tr>100 <tr>101 <td>CogVideoX1.5-5B-I2V</td>102 <td>0.273</td><td>0.283</td><td>0.241</td><td>0.328</td><td>0.257</td><td>0.328</td><td>0.305</td>103 <td>0.262</td><td><u>0.281</u></td><td>0.235</td><td>0.250</td><td><strong>0.254</strong></td><td>0.282</td>104 </tr>105 <tr>106 <td>HunyuanVideo-I2V</td>107 <td>0.273</td><td>0.280</td><td>0.207</td><td>0.357</td><td>0.293</td><td>0.280</td><td><u>0.316</u></td>108 <td>0.265</td><td>0.175</td><td><strong>0.369</strong></td><td>0.290</td><td><u>0.253</u></td><td>0.250</td>109 </tr>110 <tr>111 <td><strong>Wan2.2-I2V-A14B</strong></td>112 <td><strong>0.371</strong></td><td><strong>0.412</strong></td><td><strong>0.430</strong></td>113 <td><strong>0.382</strong></td><td><strong>0.415</strong></td><td><strong>0.404</strong></td>114 <td><strong>0.419</strong></td><td><strong>0.329</strong></td>115 <td><strong>0.405</strong></td><td>0.308</td><td><strong>0.343</strong></td>116 <td>0.236</td><td><u>0.307</u></td>117 </tr>118 <tr>119 <td><u>LTX-2</u></td>120 <td><u>0.313</u></td><td><u>0.329</u></td><td><u>0.316</u></td>121 <td><u>0.362</u></td><td><u>0.326</u></td><td><u>0.340</u></td>122 <td>0.306</td><td><u>0.297</u></td>123 <td>0.244</td><td><u>0.337</u></td><td><u>0.317</u></td>124 <td>0.231</td><td><strong>0.311</strong></td>125 </tr>126 <tr style="background:#F2F0EF;font-weight:700;text-align:center;">127 <td colspan="14"><em>Proprietary Models</em></td>128 </tr>129 <tr>130 <td><u>Seedance 2.0</u></td>131 <td><u>0.544</u></td><td><strong>0.570</strong></td><td>0.593</td><td><u>0.498</u></td><td><strong>0.618</strong></td><td><u>0.514</u></td><td><strong>0.602</strong></td>132 <td><u>0.517</u></td><td><strong>0.643</strong></td><td>0.398</td><td><u>0.492</u></td><td>0.427</td><td><strong>0.556</strong></td>133 </tr>134 <tr>135 <td>Runway Gen-4 Turbo</td>136 <td>0.403</td><td>0.392</td><td>0.396</td><td>0.409</td><td>0.429</td><td>0.341</td><td>0.363</td>137 <td>0.414</td><td>0.515</td><td><u>0.429</u></td><td>0.419</td><td>0.327</td><td>0.373</td>138 </tr>139 <tr>140 <td><strong>Sora 2</strong></td>141 <td><strong>0.546</strong></td><td><u>0.569</u></td><td><u>0.602</u></td>142 <td>0.477</td><td><u>0.581</u></td><td><strong>0.572</strong></td>143 <td><u>0.597</u></td><td><strong>0.523</strong></td>144 <td><u>0.546</u></td><td><strong>0.472</strong></td><td><strong>0.525</strong></td>145 <td><strong>0.462</strong></td><td><u>0.546</u></td>146 </tr>147 <tr>148 <td>Kling 2.6</td>149 <td>0.369</td><td>0.408</td><td>0.465</td><td>0.323</td><td>0.375</td><td>0.347</td><td>0.519</td>150 <td>0.330</td><td>0.528</td><td>0.135</td><td>0.272</td><td>0.356</td><td>0.359</td>151 </tr>152 <tr>153 <td>Veo 3.1</td>154 <td>0.480</td><td>0.531</td><td><strong>0.611</strong></td>155 <td><strong>0.503</strong></td><td>0.520</td><td>0.444</td>156 <td>0.510</td><td>0.429</td>157 <td><u>0.577</u></td><td>0.277</td><td>0.420</td>158 <td><u>0.441</u></td><td>0.404</td>159 </tr>160 <tr style="background:#F2F0EF;font-weight:700;text-align:center;">161 <td colspan="14"><em>Data Scaling Strong Baseline</em></td>162 </tr>163 <tr>164 <td><strong>VBVR-LTX2.3</strong></td>165 <td>0.516</td><td>0.580</td><td>0.608</td><td>0.631</td><td>0.529</td><td>0.454</td><td>0.680</td>166 <td>0.453</td><td>0.608</td><td>0.577</td><td><u>0.409</u></td><td>0.414</td><td><u>0.388</u></td>167 </tr>168 <tr>169 <td><strong>VBVR-Wan2.1</strong></td>170 <td><u>0.592</u></td><td><u>0.724</u></td><td><u>0.705</u></td><td><u>0.710</u></td><td><u>0.727</u></td><td><u>0.719</u></td><td><u>0.784</u></td>171 <td><u>0.461</u></td><td><u>0.674</u></td><td><strong>0.592</strong></td><td>0.387</td><td><u>0.461</u></td><td>0.387</td>172 </tr>173 <tr>174 <td><strong>VBVR-Wan2.2</strong></td>175 <td><strong>0.685</strong></td><td><strong>0.760</strong></td><td><strong>0.724</strong></td>176 <td><strong>0.750</strong></td><td><strong>0.782</strong></td><td><strong>0.745</strong></td>177 <td><strong>0.833</strong></td><td><strong>0.610</strong></td>178 <td><strong>0.768</strong></td><td><u>0.572</u></td><td><strong>0.547</strong></td>179 <td><strong>0.618</strong></td><td><strong>0.615</strong></td>180 </tr>181 </tbody>182</table>183 184## QuickStart185 186### Installation187 188We recommend using [uv](https://docs.astral.sh/uv/) to manage the environment.189 190> uv installation guide: <https://docs.astral.sh/uv/getting-started/installation/#installing-uv>191 192```bash 193pip install torch>=2.4.0 torchvision>=0.19.0 transformers Pillow huggingface_hub[cli]194uv pip install git+https://github.com/huggingface/diffusers195```196 197#### Example Code198 199```bash200huggingface-cli download Video-Reason/VBVR-Wan2.2 --local-dir ./VBVR-Wan2.2201 202python example.py \203 --model_path ./VBVR-Wan2.2204```205 206## Citation207 208```bibtex209@article{vbvr2026,210 title = {A Very Big Video Reasoning Suite},211 author = {Wang, Maijunxian and Wang, Ruisi and Lin, Juyi and Ji, Ran and212 Wiedemer, Thadd{\"a}us and Gao, Qingying and Luo, Dezhi and213 Qian, Yaoyao and Huang, Lianyu and Hong, Zelong and Ge, Jiahui and214 Ma, Qianli and He, Hang and Zhou, Yifan and Guo, Lingzi and215 Mei, Lantao and Li, Jiachen and Xing, Hanwen and Zhao, Tianqi and216 Yu, Fengyuan and Xiao, Weihang and Jiao, Yizheng and217 Hou, Jianheng and Zhang, Danyang and Xu, Pengcheng and218 Zhong, Boyang and Zhao, Zehong and Fang, Gaoyun and Kitaoka, John and219 Xu, Yile and Xu, Hua bureau and Blacutt, Kenton and Nguyen, Tin and220 Song, Siyuan and Sun, Haoran and Wen, Shaoyue and He, Linyang and221 Wang, Runming and Wang, Yanzhi and Yang, Mengyue and Ma, Ziqiao and222 Milli{\`e}re, Rapha{\"e}l and Shi, Freda and Vasconcelos, Nuno and223 Khashabi, Daniel and Yuille, Alan and Du, Yilun and Liu, Ziming and224 Lin, Dahua and Liu, Ziwei and Kumar, Vikash and Li, Yijiang and225 Yang, Lei and Cai, Zhongang and Deng, Hokin},226 journal = {arXiv preprint arXiv:2602.20159},227 year = {2026},228 url = {https://arxiv.org/abs/2602.20159}229}230```231