# Seeing is Free, Speaking is Not: Uncovering the True Energy Bottleneck in Edge VLM Inference

arXiv: https://arxiv.org/abs/2607.09520 · HTML: https://mlsanity.com/paper/2607.09520/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Visual token pruning and efficiency, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 21 of its 37 references that are in the map.
- First posted on arXiv 10 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, 21 of 21)

- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [Microsoft COCO: Common Objects in Context](https://mlsanity.com/paper/1405.0312/) (2014)
- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution](https://mlsanity.com/paper/2409.12191/) (2024)
- [RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control](https://mlsanity.com/paper/2307.15818/) (2023)
- [OpenVLA: An Open-Source Vision-Language-Action Model](https://mlsanity.com/paper/2406.09246/) (2024)
- [π0: A Vision-Language-Action Flow Model for General Robot Control](https://mlsanity.com/paper/2410.24164/) (2024)
- [PaLM-E: An Embodied Multimodal Language Model](https://mlsanity.com/paper/2303.03378/) (2023)
- [InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models](https://mlsanity.com/paper/2504.10479/) (2025)
- [Gemma 3 Technical Report](https://mlsanity.com/paper/2503.19786/) (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)
- [SparseVLM: Visual Token Sparsification for Efficient Vision-Language Model Inference](https://mlsanity.com/paper/2410.04417/) (2024)
- [PyramidDrop: Accelerating Your Large Vision-Language Models via Pyramid Visual Redundancy Reduction](https://mlsanity.com/paper/2410.17247/) (2024)
- [A Survey on Vision-Language-Action Models for Embodied AI](https://mlsanity.com/paper/2405.14093/) (2024)
- [SmolVLM: Redefining small and efficient multimodal models](https://mlsanity.com/paper/2504.05299/) (2025)
- [Edge-First Language Model Inference: Models, Metrics, and Tradeoffs](https://mlsanity.com/paper/2505.16508/) (2025)
- [PRISM: Privacy-Aware Routing for Adaptive Cloud-Edge LLM Inference via Semantic Sketch Collaboration](https://mlsanity.com/paper/2511.22788/) (2025)
- [Where Do the Joules Go? Diagnosing Inference Energy Consumption](https://mlsanity.com/paper/2601.22076/) (2026)
- [Understanding the Performance and Power of LLM Inferencing on Edge Accelerators](https://mlsanity.com/paper/2506.09554/) (2025)
- [HyperVL: An Efficient and Dynamic Multimodal Large Language Model for Edge Devices](https://mlsanity.com/paper/2512.14052/) (2025)
- [Dissecting Transformers: A CLEAR Perspective towards Green AI](https://mlsanity.com/paper/2510.02810/) (2025)

## Built on by (most cited first, 2 of 2)

- [Trains but Doesn't Learn: A Post-Training Delivery Benchmark for LLM Agents as Forward-Deployed Engineers](https://mlsanity.com/paper/2609.25237/) (2026)
- [Seeing at the Edge: Energy Characterization of Vision-Language Model Inference on Edge Platforms](https://mlsanity.com/paper/2610.20136/) (2026)

## Abstract (arXiv)

Vision-Language Models (VLMs) are the perceptual backbone of embodied AI, but their energy footprint on edge hardware remains poorly understood. Existing efficiency efforts focus predominantly on reducing visual tokens, implicitly treating visual processing as the dominant energy cost. We overturn this implicit assumption through the first systematic energy profiling of on-device VLM inference, spanning five models across three architecture families, four input resolutions, and two hardware platforms (NVIDIA RTX 3070 and Jetson Orin NX). Our analysis yields three findings. First, average inference power is a model-intrinsic constant, invariant to input resolution, image complexity, and prompt type, with less than 5% variation across all conditions. This means that all energy variation across inputs must arise from variation in inference time, not from variation in power draw. Second, each output token costs 11 to 39x more wall-clock time than each input token due to the compute-bound and memory-bound asymmetry between prefill and decode, making output token count the dominant driver of both latency and energy. Third, image complexity, measured by the number of objects in an image, induces up to 4.1x energy differences at identical resolution. This variation arises not from increased visual processing cost, but from differences in output length. These findings expose a fundamental limitation of visual token pruning: even removing all visual tokens saves at most 10% of total energy for fixed-token models. Across models spanning 1 billion to 8 billion parameters, controlling output length saves up to 97% of total energy, with the energy dominance of decoding growing stronger at larger model scale. In short, the true energy bottleneck in edge VLM inference is not what the model sees, but how much it says. Code is available at https://github.com/Junfei-Z/seeing-is-free.