跳到正文
arXiv cs.AI· Naihao Deng, Alissa Shen, Yiming Feng, Joan Nwatu, Jae-Won Chung, Mosharaf Chowdhury, Yulong Chen, Rada Mihalcea·· 2 小时前AI 评分37

大语言模型多语言推理的能源鸿沟:不同语言能耗差异有多大

The Language-Energy Divide: Measuring Energy Costs of Multilingual LLM Inference

AI 导读

一项被 EMNLP 2026 接收的系统性研究测量了 LLM 多语言推理的能耗差异:每输出 token 的能耗跨语言最多相差 8.3 倍,固定请求集的总能耗从英语的 17.6 kJ 到普什图语的 3,147 kJ 相差最多 179 倍。

正文 · 原文

View PDF HTML (experimental)

Abstract:Large language models (LLMs) are increasingly deployed in multilingual settings, yet the energy costs of serving these models across different languages remain poorly understood. We present a systematic study of inference energy consumption across languages with this http URL framework (Chung et al., 2026). We find striking disparities: energy consumption per output token varies by up to 8.3 times across languages, while total energy for a fixed set of requests varies by up to 179 times between the cheapest (English, 17.6 kJ) and the most expensive (Pashto, 3,147 kJ) languages. Our analysis shows that this disparity is driven by two compounding factors: (1) higher per-token energy costs for languages using complex or rare scripts, and (2) more tokens generated for low-resource languages. Moreover, we find a double cost + performance penalty: languages with the highest energy footprints also tend to achieve the lowest task accuracy. We reveal that the energy divide persists across models, hardware, and tasks, suggesting a systemic energy inequity in multilingual LLM deployment. Finally, we recommend that the community treat energy as a first-class evaluation axis, extend reporting checklists and model cards to include it, and adopt deployment-side mitigations for better energy efficiency.
Comments: Accepted to EMNLP 2026 Main
Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
Cite as: arXiv:2606.21869 [cs.CL]
  (or arXiv:2606.21869v3 [cs.CL] for this version)
  https://doi.org/10.48550/arXiv.2606.21869

arXiv-issued DOI via DataCite

Submission history

From: Naihao Deng [view email]
[v1] Sat, 20 Jun 2026 04:12:29 UTC (2,507 KB)
[v2] Mon, 14 Sep 2026 00:16:05 UTC (2,514 KB)
[v3] Mon, 16 Nov 2026 20:22:44 UTC (2,505 KB)

来源:arXiv cs.AI · arxiv.org