Keyword Matters: Unveiling the Energy Sensitivity of On-Device LLM Prompting
Quick Answer
This study reveals that prompt design significantly impacts energy consumption in on-device LLMs, with specific linguistic features influencing decoding length and total energy usage.
Quick Take
By analyzing real power measurements from smartphones, the research highlights that imperative keywords and instruction structures can serve as effective tools for enhancing energy efficiency during inference.
Key Points
- Prompt wording affects energy consumption in on-device , particularly with imperative keywords.
- Real power measurements from smartphones were used to quantify energy differences across tasks.
- The study shows consistent energy variances across different verbs and tasks.
- Prompt engineering is identified as a lightweight method to enhance energy efficiency.
- Model compression and runtime acceleration have been widely studied, but prompt design remains underexplored.
Paper Resources
📖 Reader Mode
~2 min readAbstract:Large Language Models (LLMs) are increasingly deployed on mobile and embedded devices to improve privacy and reduce network latency. Yet on-device inference faces a fundamental constraint: high energy consumption on battery-powered, resource-limited hardware. While model compression and runtime acceleration have been widely studied, the effect of \emph{prompt design} on energy efficiency remains underexplored. This paper presents an empirical study of the relationship between prompt wording and energy consumption for on-device LLMs. Using real power measurements collected on a smartphone, we quantify how linguistic features, particularly imperative keywords and instruction structure, affect decoding length and total energy. Our results show consistent energy differences across verbs and tasks, indicating that prompt engineering is a lightweight lever for improving energy efficiency.
| Subjects: | Artificial Intelligence (cs.AI); Networking and Internet Architecture (cs.NI) |
| Cite as: | arXiv:2607.22568 [cs.AI] |
| (or arXiv:2607.22568v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2607.22568 arXiv-issued DOI via DataCite |
Submission history
From: Ruiyi Tao [view email]
[v1]
Sun, 31 May 2026 18:29:16 UTC (232 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.AI
See more →HOBA: Hierarchical On-Policy Bidding Agents for Adaptive Online Advertising
HOBA (Hierarchical On-policy Bidding Agents) is a novel hierarchical reinforcement learning framework that enhances online advertising bidding systems by improving adaptability and reducing hyperparameter tuning costs. It utilizes a for hyperparameter inference, a SARSA agent for expert model selection, and a dynamic expert pool for bid execution, achieving a +3.6% increase in target cost during large-scale deployment and outperforming state-of-the-art baselines on AuctionNet.