BEAR: Benchmarking and Enhancing Multimodal Language Models with Atomic Embodied Capabilities

Project Page | arXiv | Twitter | Huggingface

🚀 Todos

• Release BEAR Benchmark with evaluation scripts
• Release the official code of BEAR-Agent

🥳 BibTex

We would appreciate it if you find this work useful and consider citing it🎉!

@article{qi2025bear,
  title={BEAR: Benchmarking and Enhancing Multimodal Language Models for Atomic Embodied Capabilities},
  author={Qi, Yu and Zhao, Haibo and Guo, Ziyu and Ma, Siyuan and Chen, Ziyan and Han, Yaokun and Zhang, Renrui and Lin, Zitiantao and Xin, Shiji and Huang, Yijian and others},
  journal={arXiv preprint arXiv:2510.08759},
  year={2025}
}

🤔 Contact

Please contact Yu ([email protected]) or Haibo ([email protected]) if you would like discussions!