Efficient multimodal large language models: a survey
Abstract In the past years, multimodal large language models (MLLMs) have demonstrated remarkable performance in tasks such as visual question answering and visual understanding and reasoning. However, the extensive model size and high training and inference costs have hindered the widespread application of MLLMs in academia and industry. Thus, studying efficient and lightweight MLLMs has enormous potential, especially in edge computing scenarios. In this survey, we provide a comprehensive and systematic review of the current state of efficient MLLMs. Specifically, this survey summarizes the timeline of representative efficient MLLMs, the current state of research in structures and strategies, and the applications. Finally, the limitations of current efficient MLLM research and promising future directions are discussed.
Authors
- Yexin Liu
- Chengjie Wang (ORCID: https://orcid.org/0000-0003-4216-8090)
- Yabiao Wang (ORCID: https://orcid.org/0000-0002-6592-8411)
- Jinxiang Lai
- Bo Zhao (ORCID: https://orcid.org/0000-0002-7834-6672)
- Zhenye Gan (ORCID: https://orcid.org/0000-0002-2431-1159)
- Xin Tan (ORCID: https://orcid.org/0000-0001-9346-1196)
- Lizhuang Ma (ORCID: https://orcid.org/0000-0003-1653-4341)
- Jian Li (ORCID: https://orcid.org/0000-0002-0242-6481)
- Yizhang Jin
- Tianjun Gu (ORCID: https://orcid.org/0009-0001-4757-2685)
Institutions
- Shanghai Jiao Tong University (CN)
- Hong Kong University of Science and Technology (HK)
- Tencent (China) (CN)
- Beijing Academy of Artificial Intelligence (CN)
- East China Normal University (CN)
Publication Details
- Journal
- Visual Intelligence
- Published
- 2025-12-01
- DOI
- https://doi.org/10.1007/s44267-025-00099-6
- Citations
- 7
- Primary Topic
- Multimodal Machine Learning Applications
- Type
- article
- Field-Weighted Citation Impact
- 8.93
Funders
- National Natural Science Foundation of China