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

2026-06-20

AI Summary

This study addresses a significant yet overlooked disparity in energy consumption during multilingual large language model (LLM) inference across different languages. Leveraging the ML.Energy framework, the authors systematically quantify this “language-energy gap” by measuring inference energy across diverse models, hardware platforms, and tasks. Their analysis reveals that English exhibits the lowest energy cost (17.6 kJ), while Pashto incurs the highest (3,147 kJ)—a per-token difference of up to 8.3× and an overall gap as large as 179×. Notably, high-energy languages frequently coincide with low accuracy, resulting in a dual penalty. The work further elucidates how script complexity and low-resource characteristics jointly exacerbate energy inefficiency, urging the integration of energy consumption into standard evaluation protocols for multilingual LLMs.

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 ML.Energy 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.

Problem

• energy consumption
• multilingual LLMs
• inference cost
• language disparity
• energy inequity

Innovation

• energy efficiency
• multilingual LLMs
• inference cost
• language disparity
• ML.Energy framework