{"ID":2836291,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2511.21150","arxiv_id":"2511.21150","title":"LLaVA-UHD v3: Progressive Visual Compression for Efficient Native-Resolution Encoding in MLLMs","abstract":"Visual encoding followed by token condensing has become the standard architectural paradigm in multi-modal large language models (MLLMs). Many recent MLLMs increasingly favor global native- resolution visual encoding over slice-based methods. To investigate this trend, we systematically compare their behavior on vision-language understanding and attention patterns, revealing that global encoding enhances overall capability but at the expense of greater computational overhead. To address this issue, we present LLaVA-UHD v3, an MLLM centered upon our proposed Progressive Visual Compression (PVC) method, which can be seamlessly integrated into standard Vision Transformer (ViT) to enable efficient native-resolution encoding. The PVC approach consists of two key modules: (i) refined patch embedding, which supports flexible patch-size scaling for fine-grained visual model- ing, (ii) windowed token compression, hierarchically deployed across ViT layers to progressively aggregate local token representations. Jointly modulated by these two modules, a widely pretrained ViT can be reconfigured into an efficient architecture while largely preserving generality. Evaluated across extensive benchmarks, the transformed ViT, termed ViT-UHD, demonstrates competitive performance with MoonViT while reducing TTFT (time-to-first-token) by 2.4x, when developed within an identical MLLM architecture. Building upon ViT-UHD, LLaVA-UHD v3 also achieves competitive performance to Qwen2-VL, while further reducing TTFT by 1.9x. We will release all code and checkpoints to support future research on efficient MLLMs.","short_abstract":"Visual encoding followed by token condensing has become the standard architectural paradigm in multi-modal large language models (MLLMs). Many recent MLLMs increasingly favor global native- resolution visual encoding over slice-based methods. To investigate this trend, we systematically compare their behavior on vision...","url_abs":"https://arxiv.org/abs/2511.21150","url_pdf":"https://arxiv.org/pdf/2511.21150v1","authors":"[\"Shichu Sun\",\"Yichen Zhang\",\"Haolin Song\",\"Zonghao Guo\",\"Chi Chen\",\"Yidan Zhang\",\"Yuan Yao\",\"Zhiyuan Liu\",\"Maosong Sun\"]","published":"2025-11-26T08:11:10Z","proceeding":"cs.CV","tasks":"[\"cs.CV\",\"cs.AI\"]","methods":"[\"Vision Transformer\",\"Transformer\",\"Large Language Model\",\"Language Model\"]","has_code":false}