# CodeShrink: Adaptive Visual Compression for Efficient Multimodal Code Understanding

arXiv: https://arxiv.org/abs/2607.29637 · HTML: https://mlsanity.com/paper/2607.29637/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Visual token pruning and efficiency, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 32 of its 41 references that are in the map.
- First posted on arXiv 31 Jul 2026.

## Placed in

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

## Builds on (most cited first, 30 of 32)

- [LoRA: Low-Rank Adaptation of Large Language Models](https://mlsanity.com/paper/2106.09685/) (2021)
- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [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)
- [LongLLMLingua: Accelerating and Enhancing LLMs in Long Context Scenarios via Prompt Compression](https://mlsanity.com/paper/2310.06839/) (2023)
- [LLMLingua-2: Data Distillation for Efficient and Faithful Task-Agnostic Prompt Compression](https://mlsanity.com/paper/2403.12968/) (2024)
- [VisionZip: Longer is Better but Not Necessary in Vision Language Models](https://mlsanity.com/paper/2412.04467/) (2024)
- [DeepSeek-OCR: Contexts Optical Compression](https://mlsanity.com/paper/2510.18234/) (2025)
- [Qwen3.5-Omni Technical Report](https://mlsanity.com/paper/2604.15804/) (2026)
- [LongCoder: A Long-Range Pre-trained Language Model for Code Completion](https://mlsanity.com/paper/2306.14893/) (2023)
- [Interaction2Code: Benchmarking MLLM-based Interactive Webpage Code Generation from Interactive Prototyping](https://mlsanity.com/paper/2411.03292/) (2024)
- [DesignBench: A Comprehensive Benchmark for MLLM-based Front-end Code Generation](https://mlsanity.com/paper/2506.06251/) (2025)
- [MMCode: Benchmarking Multimodal Large Language Models for Code Generation with Visually Rich Programming Problems](https://mlsanity.com/paper/2404.09486/) (2024)
- [LongCodeZip: Compress Long Context for Code Language Models](https://mlsanity.com/paper/2510.00446/) (2025)
- [LongCodeBench: Evaluating Coding LLMs at 1M Context Windows](https://mlsanity.com/paper/2505.07897/) (2025)
- [Diet Code Is Healthy: Simplifying Programs for Pre-trained Models of Code](https://mlsanity.com/paper/2206.14390/) (2022)
- [MRWeb: An Exploration of Generating Multi-Page Resource-Aware Web Code from UI Designs](https://mlsanity.com/paper/2412.15310/) (2024)
- [SlideCoder: Layout-aware RAG-enhanced Hierarchical Slide Generation from Design](https://mlsanity.com/paper/2506.07964/) (2025)
- [Seeing is Coding: On the Effectiveness of Vision Language Models in Code Understanding](https://mlsanity.com/paper/2602.01785/) (2026)
- [Empowering Agile-Based Generative Software Development through Human-AI Teamwork](https://mlsanity.com/paper/2407.15568/) (2024)
- [The Hidden Cost of Readability: How Code Formatting Silently Consumes Your LLM Budget](https://mlsanity.com/paper/2508.13666/) (2025)
- [GPTCloneBench: A comprehensive benchmark of semantic clones and cross-language clones using GPT-3 model and SemanticCloneBench](https://mlsanity.com/paper/2308.13963/) (2023)
- [Natural Is The Best: Model-Agnostic Code Simplification for Pre-trained Large Language Models](https://mlsanity.com/paper/2405.11196/) (2024)
- [EfficientPosterGen: Semantic-aware Efficient Poster Generation via Token Compression and Accurate Violation Detection](https://mlsanity.com/paper/2603.00155/) (2026)
- [DesignRepair: Dual-Stream Design Guideline-Aware Frontend Repair with Large Language Models](https://mlsanity.com/paper/2411.01606/) (2024)
- [EfficientUICoder: A Bidirectional Token Compression Framework for Efficient MLLM-Based UI Code Generation](https://mlsanity.com/paper/2509.12159/) (2025)
- [ComUICoder: Component-based Reusable UI Code Generation for Complex Websites via Semantic Segmentation and Element-wise Feedback](https://mlsanity.com/paper/2602.19276/) (2026)
- [Envisioning Future Interactive Web Development: Editing Webpage with Natural Language](https://mlsanity.com/paper/2510.26516/) (2025)
- [Token Sugar: Making Source Code Sweeter for LLMs through Token-Efficient Shorthand](https://mlsanity.com/paper/2512.08266/) (2025)
- [UIBenchKit: A unified toolkit for design-to-code model evaluation](https://mlsanity.com/paper/2605.13141/) (2026)

## Abstract (arXiv)

Rendering source code as images offers a promising way to reduce the input costs of Multimodal Large Language Models (MLLMs). Adjusting image resolution can trade visual token cost against content fidelity. However, resolution scaling alone overlooks two sources of inefficiency: blank regions created by line breaks and indentation, and code regions irrelevant to the current instruction. Moreover, the best compression setting varies across inputs, tasks, and models, limiting fixed-ratio strategies. We propose CodeShrink, an adaptive visual compression framework with three components. Blank-Free Rendering replaces whitespace-dependent layouts with compact layouts and explicit structural markers, removing layout-induced tokens. Adaptive Compression Configuration uses a lightweight agent trained with reinforcement learning to predict a per-input setting that balances token efficiency and readability. Dominant Token Selection jointly analyzes the instruction and code image to prune task-irrelevant visual tokens during inference. We evaluate CodeShrink on code question answering, clone detection, and code completion. CodeShrink reduces visual token use by up to 71.2% while matching or exceeding uncompressed text-only inputs, and consistently outperforms text-based and visual compression baselines across all three tasks. These results show that combining layout compaction, adaptive configuration, and instruction-aware pruning can make multimodal code understanding more efficient. Our code is available at https://github.com/vinsontang1/CodeShrink.