# Visual token pruning and efficiency

Area: [Multimodal LLMs / vision-language models](https://mlsanity.com/mllm/index.md). HTML: https://mlsanity.com/mllm/visual-token-pruning-and-efficiency/

## Key facts

- Visual token pruning and efficiency: 635 papers on arXiv since 2022 (data through 14 Oct 2026).
- 158 of its 635 mapped papers (25%) cite “An Image is Worth 1/2 Tokens After Layer 2: Plug-and-Play Inference Acceleration for Large Vision-Language Models” (2024, arXiv 2403.06764).
- 123 of its 635 mapped papers (19%) cite “LLaVA-PruMerge: Adaptive Token Reduction for Efficient Large Multimodal Models” (2024, arXiv 2403.15388).
- 56 new papers 17 Sep–14 Oct.

## Most cited here

- [An Image is Worth 1/2 Tokens After Layer 2: Plug-and-Play Inference Acceleration for Large Vision-Language Models](https://mlsanity.com/paper/2403.06764/) (2024), cited by 158 papers here
- [LLaVA-PruMerge: Adaptive Token Reduction for Efficient Large Multimodal Models](https://mlsanity.com/paper/2403.15388/) (2024), cited by 123 papers here
- [SparseVLM: Visual Token Sparsification for Efficient Vision-Language Model Inference](https://mlsanity.com/paper/2410.04417/) (2024), cited by 119 papers here
- [PyramidDrop: Accelerating Your Large Vision-Language Models via Pyramid Visual Redundancy Reduction](https://mlsanity.com/paper/2410.17247/) (2024), cited by 93 papers here
- [VisionZip: Longer is Better but Not Necessary in Vision Language Models](https://mlsanity.com/paper/2412.04467/) (2024), cited by 88 papers here
- [Boosting Multimodal Large Language Models with Visual Tokens Withdrawal for Rapid Inference](https://mlsanity.com/paper/2405.05803/) (2024), cited by 63 papers here
- [DivPrune: Diversity-based Visual Token Pruning for Large Multimodal Models](https://mlsanity.com/paper/2503.02175/) (2025), cited by 59 papers here
- [Beyond Text-Visual Attention: Exploiting Visual Cues for Effective Token Pruning in VLMs](https://mlsanity.com/paper/2412.01818/) (2024), cited by 57 papers here
- [TokenPacker: Efficient Visual Projector for Multimodal LLM](https://mlsanity.com/paper/2407.02392/) (2024), cited by 49 papers here
- [Stop Looking for Important Tokens in Multimodal Language Models: Duplication Matters More](https://mlsanity.com/paper/2502.11494/) (2025), cited by 45 papers here

## New (17 Sep–14 Oct)

- [Visual Tokens on Demand: Visual Budgeting for Scalable and Efficient Multi-Image Reasoning in Multimodal LLMs](https://mlsanity.com/paper/2610.18858/) (2026-10-13)
- [SP2ec: Adaptive Self-Speculative Decoding for Vision-Language Models](https://mlsanity.com/paper/2610.18735/) (2026-10-13)
- [Bifocal: Training-Free Mixed Granularity Visual Token Pruning for Multimodal LLMs](https://mlsanity.com/paper/2610.18399/) (2026-10-13)
- [DVTR-KD: Distilling Vision-Language Models via Intermediate Visual Token Reconstruction](https://mlsanity.com/paper/2610.17853/) (2026-10-13)
- [TwinPrune: Density-Aware Two-Phase Token Pruning for Vision-Language Models](https://mlsanity.com/paper/2610.17530/) (2026-10-13)
- [When 1+1 Backfires: Rethinking Relevance-Diversity Balancing for Sample-Aware Visual Token Reduction](https://mlsanity.com/paper/2610.16615/) (2026-10-12)
- [Attention Sinks in VLMs Are Borrowed, Not Seen](https://mlsanity.com/paper/2610.15855/) (2026-10-11)
- [SheafPrune: Sheaf-Guided Graph Propagation for Visual Token Pruning](https://mlsanity.com/paper/2610.13263/) (2026-10-09)
- [From Glance to Focus: Mitigating Recognition Laziness in MLLMs for Fine-Grained Recognition](https://mlsanity.com/paper/2610.13154/) (2026-10-09)
- [GeoGR: Spatial-Semantic Token Compression and Group Routing for Efficient 3D VLMs](https://mlsanity.com/paper/2610.13035/) (2026-10-09)
- [Look Back, Think Ahead: Visual Memory on Demand for Efficient Multimodal Reasoning](https://mlsanity.com/paper/2610.12060/) (2026-10-08)
- [SAGE: Sink-Aware Guided Emphasis for Visual Grounding in Vision-Language Decoders](https://mlsanity.com/paper/2610.11469/) (2026-10-08)
- [V-CoLA: Vision Token Compression with Linear Attention](https://mlsanity.com/paper/2610.11251/) (2026-10-08)
- [Why VLMs Miss Small Objects, and When Zooming In Is Safe](https://mlsanity.com/paper/2610.09313/) (2026-10-07)
- [Adaptive Visual Token Reduction for Accelerated Image Understanding](https://mlsanity.com/paper/2610.09252/) (2026-10-07)
- [DIPrune: Task-Aware Token Pruning with Dual Importance for Efficient Multimodal Language Models](https://mlsanity.com/paper/2610.08341/) (2026-10-06)
- [VisionWeave: Weaving Elastic Visual Representations as a Native Capability of MLLMs](https://mlsanity.com/paper/2610.07987/) (2026-10-06)
- [Foveated Compression: Selective High-Resolution Preservation for Token-Efficient VLMs](https://mlsanity.com/paper/2610.07729/) (2026-10-06)
- [When and What to Prune? Stage-Aware Visual Token Pruning for Efficient VLA](https://mlsanity.com/paper/2610.05273/) (2026-10-04)
- [FlashGaze: Training-Free Multi-Scale Patch Pruning For Efficient Video Understanding](https://mlsanity.com/paper/2610.04225/) (2026-10-03)
- [From Patching to Pruning Visual Computation in Vision Language Models](https://mlsanity.com/paper/2610.03389/) (2026-10-02)
- [VETO: Video Efficient Token Optimization for Vision Language Models](https://mlsanity.com/paper/2610.01785/) (2026-10-01)
- [Architectural Sampling: Test-Time Scaling via Computational Diversity in Frozen Vision-Language Models](https://mlsanity.com/paper/2610.01687/) (2026-10-01)
- [MWOP: Modality-aware Width-wise Operation Pruning for Efficient MLLMs](https://mlsanity.com/paper/2610.01434/) (2026-10-01)
- [Fewer Tokens, More Self-Teaching: On-Policy Self-Distillation for Extreme Visual Token Reduction](https://mlsanity.com/paper/2609.32353/) (2026-10-01)
- [HAWK: Rethinking Multimodal Drafting for Speculative Decoding](https://mlsanity.com/paper/2610.00623/) (2026-09-30)
- [CoVisco: Codec-Native Vision Encoder with Native Token Compression for Unified Image-Video Understanding](https://mlsanity.com/paper/2609.39924/) (2026-09-30)
- [Feature-Aware Token Attack for Compression-Triggered Stealthy Failures in Large Vision-Language Models](https://mlsanity.com/paper/2609.39134/) (2026-09-30)
- [Representation Dynamics Reveal Semantic Saliency and Similarity for Visual Token Pruning in MLLMs](https://mlsanity.com/paper/2609.36916/) (2026-09-30)
- [MiCo: Mutual Information Coverage Optimization through Semantic Erasure Modeling for Efficient MLLM Inference](https://mlsanity.com/paper/2609.34330/) (2026-09-30)