# The Hidden Cost of Readability: How Code Formatting Silently Consumes Your LLM Budget arXiv: https://arxiv.org/abs/2508.13666 ยท HTML: https://mlsanity.com/paper/2508.13666/ ## Key facts - Cited by 12 papers in the map (data through 14 Oct 2026). - Placed in Efficient reasoning and agent inference, one of 12 directions in Efficient LLM inference and serving. - Builds on 18 of its 38 references that are in the map. - First posted on arXiv 19 Aug 2025. ## Placed in - Efficient LLM inference and serving / [Efficient reasoning and agent inference](https://mlsanity.com/inference/efficient-reasoning-and-agent-inference/) ## Builds on (most cited first, 18 of 18) - [QLoRA: Efficient Finetuning of Quantized LLMs](https://mlsanity.com/paper/2305.14314/) (2023) - [Neural Machine Translation of Rare Words with Subword Units](https://mlsanity.com/paper/1508.07909/) (2015) - [Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone](https://mlsanity.com/paper/2404.14219/) (2024) - [DeepSeek-Coder: When the Large Language Model Meets Programming -- The Rise of Code Intelligence](https://mlsanity.com/paper/2401.14196/) (2024) - [CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and Generation](https://mlsanity.com/paper/2109.00859/) (2021) - [StarCoder 2 and The Stack v2: The Next Generation](https://mlsanity.com/paper/2402.19173/) (2024) - [Think before you speak: Training Language Models With Pause Tokens](https://mlsanity.com/paper/2310.02226/) (2023) - [Magicoder: Empowering Code Generation with OSS-Instruct](https://mlsanity.com/paper/2312.02120/) (2023) - [Efficient Training of Language Models to Fill in the Middle](https://mlsanity.com/paper/2207.14255/) (2022) - [Diet Code Is Healthy: Simplifying Programs for Pre-trained Models of Code](https://mlsanity.com/paper/2206.14390/) (2022) - [Probing Model Signal-Awareness via Prediction-Preserving Input Minimization](https://mlsanity.com/paper/2011.14934/) (2020) - [Don't Complete It! Preventing Unhelpful Code Completion for Productive and Sustainable Neural Code Completion Systems](https://mlsanity.com/paper/2209.05948/) (2022) - [When Neural Code Completion Models Size up the Situation: Attaining Cheaper and Faster Completion through Dynamic Model Inference](https://mlsanity.com/paper/2401.09964/) (2024) - [AI Coders Are Among Us: Rethinking Programming Language Grammar Towards Efficient Code Generation](https://mlsanity.com/paper/2404.16333/) (2024) - [Comparing large language models and human programmers for generating programming code](https://mlsanity.com/paper/2403.00894/) (2024) - [Syntax-Guided Program Reduction for Understanding Neural Code Intelligence Models](https://mlsanity.com/paper/2205.14374/) (2022) - [STYLE-ANALYZER: fixing code style inconsistencies with interpretable unsupervised algorithms](https://mlsanity.com/paper/1904.00935/) (2019) - [Is Next Token Prediction Sufficient for GPT? Exploration on Code Logic Comprehension](https://mlsanity.com/paper/2404.08885/) (2024) ## Built on by (most cited first, 12 of 12) - [Reducing Cost of LLM Agents with Trajectory Reduction](https://mlsanity.com/paper/2509.23586/) (2025) - [Seeing is Coding: On the Effectiveness of Vision Language Models in Code Understanding](https://mlsanity.com/paper/2602.01785/) (2026) - [EfficientUICoder: A Bidirectional Token Compression Framework for Efficient MLLM-Based UI Code Generation](https://mlsanity.com/paper/2509.12159/) (2025) - [The Devil Is in the Interface: Evaluating How Tool Architecture Shapes Coding Agent Behavior](https://mlsanity.com/paper/2608.11386/) (2026) - [Characterizing Readability Issue Patterns and the Role of Prompt Design in LLM-Generated Code](https://mlsanity.com/paper/2605.13280/) (2026) - [The Best Programming Language for Tokenmaxxing: An Investigation of Coding Agent Behavior Across Programming Languages](https://mlsanity.com/paper/2607.22807/) (2026) - [AttnCompress: Dynamic Attention-Guided Trajectory Compression for Software Engineering Agents](https://mlsanity.com/paper/2609.08318/) (2026) - [FlowArk: Boosting Agentic Data-flow Analysis for Android Apps via Context-Aware Knowledge Reuse](https://mlsanity.com/paper/2607.11308/) (2026) - [CodeShrink: Adaptive Visual Compression for Efficient Multimodal Code Understanding](https://mlsanity.com/paper/2607.29637/) (2026) - [Talking to Itself While Coding: What Makes Comments Help Code Generation?](https://mlsanity.com/paper/2609.09242/) (2026) - [Analyzing and Mitigating Cost-Inefficient Behaviors in Coding Agents](https://mlsanity.com/paper/2609.30725/) (2026) - [Mind the Gaps: From Failure Attribution to Closed-Form Repair of Code Language Models](https://mlsanity.com/paper/2610.05277/) (2026) ## Abstract (arXiv) Source code is usually formatted with elements like indentation and newlines to improve readability for human developers. However, these visual aids do not seem to be beneficial for large language models (LLMs) in the same way since the code is processed as a linear sequence of tokens. Furthermore, these additional tokens can lead to increased computational costs and longer response times for LLMs. If such formatting elements are non-essential to LLMs, we can reduce such costs by removing them from the code. To figure out the role played by formatting elements, we conduct a comprehensive empirical study to evaluate the impact of code formatting on LLM performance and efficiency. Through large-scale experiments on Fill-in-the-Middle Code Completion tasks across four programming languages (Java, Python, C++, C#) and ten LLMs-including both commercial and open-source models-we systematically analyze token count and performance when formatting elements are removed. Key findings indicate that LLMs can maintain performance across formatted code and unformatted code, achieving an average input token reduction of 24.5% with negligible output token reductions. This makes code format removal a practical optimization strategy for improving LLM efficiency. Further exploration reveals that both prompting and fine-tuning LLMs can lead to significant reductions (up to 36.1%) in output code length without compromising correctness. To facilitate practical applications, we develop a bidirectional code transformation tool for format processing, which can be seamlessly integrated into existing LLM inference workflows, ensuring both human readability and LLM efficiency.