Current multimodal large lanauge models possess strong perceptual and reasoning capabilities, however high computational and memory requirements make them difficult to deploy directly on on-device environments. While small-parameter models are progressively endowed with strong general capabilities, standard Vision Transformer (ViT) encoders remain a critical bottleneck, suffering from excessive latency and memory consumption when processing high-resolution inputs.To address these challenges, we introduce HyperVL, an efficient multimodal large language model tailored for on-device inference. HyperVL adopts an image-tiling strategy to cap peak memory usage and incorporates two novel techniques: (1) a Visual Resolution Compressor (VRC) that adaptively predicts optimal encoding resolutions to eliminate redundant computation, and (2) Dual Consistency Learning (DCL), which aligns multi-scale ViT encoders within a unified framework, enabling dynamic switching between visual branches under a shared LLM. Extensive experiments demonstrate that HyperVL achieves state-of-the-art performance among models of comparable size across multiple benchmarks. Furthermore, it significantly significantly reduces latency and power consumption on real mobile devices, demonstrating its practicality for on-device multimodal inference.
@article{arxiv.2512.14052,
title = {HyperVL: An Efficient and Dynamic Multimodal Large Language Model for Edge Devices},
author = {HyperAI Team and Yuchen Liu and Kaiyang Han and Zhiqiang Xia and Yuhang Dong and Chen Song and Kangyu Tang and Jiaming Xu and Xiushi Feng and WenXuan Yu and Li Peng and Mingyang Wang and Kai Wang and Changpeng Yang and Yang Li and Haoyu Lu and Hao Wang and Bingna Xu and Guangyao Liu and Long Huang and Kaibin Guo and Jinyang Wu and Dan Wu and Hongzhen Wang and Peng Zhou and Shuai Nie and Shande Wang and Runyu Shi and Ying Huang},
journal= {arXiv preprint arXiv:2512.14052},
year = {2025}
}