cs.SEMar 31, 2026

The Energy Footprint of LLM-Based Environmental Analysis: LLMs and Domain Products

Authors: Alicia Bao, Jiamian He, Angel Hsu, Diego Manya, Ji, Zhang

Organizations: James · Department of Computer Science, University of North Carolina at Chapel Hill, Chapel Hill, NC 27516 · Arboretica, Rotterdam, The Netherlands · Department of Public Policy, University of North Carolina at Chapel Hill, Chapel Hill, NC 27516 · 

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

As large language models (LLMs) are increasingly used in domain-specific applications, including climate change and environmental research, understanding their energy footprint has become an important concern. The growing adoption of retrieval-augmented (RAG) systems for climate-domain specific analysis raises a key question: how does the energy consumption of domain-specific RAG workflows compare with that of direct generic LLM usage? Prior research has focused on standalone model calls or coarse token-based estimates, while leaving the energy implications of deployed application workflows insufficiently understood. In this paper, we assess the inference-time energy consumption of two LLM-based climate analysis chatbots (ChatNetZero and ChatNDC) compared to the generic GPT-4o-mini model. We estimate energy use under actual user queries by decomposing each workflow into retrieval, generation, and hallucination-checking components. We also test across different times of day and geographic access locations. Our results show that the energy consumption of domain-specific RAG systems depends strongly on their design. More agentic pipelines substantially increase inference-time energy use, particularly when used for additional accuracy or verification checks, although they may not yield proportional gains in response quality. While more research is needed to further test these initial findings more robustly across models, environments and prompting structures, this study provides a new understanding on how the design of domain-specific LLM products affects both the energy footprint and quality of output.

Figures & tables

Figure 1: Design of ChatNetZero Hsu et al. (2024) .
Figure 2: Architectures of (a) ChatNetZero and (b) ChatNDC.
Question typeExample questions
Knowledge“Has Marathon Petroleum made any climate pledges?”; “Does Foxconn have any climate strategy?”; “What is the total emissions reduction target of Malaysia?”; “Does Taiwan have a carbon trading scheme or policy under its latest NDC?”
Comprehension“Interpret how ESG regulations influence business strategy and investor decisions”; “Explain the relationship between local air pollution and public health outcomes.”
Application“What are Scope 3 emissions and what should be included in a net-zero target?”; “If an entity doesn’t control out-of-boundary emissions, how can it credibly set a net-zero target?”
Analysis“Are Costco’s or Walmart’s net zero target 1.5 degree aligned?”; “How does Walmart’s climate goals compare with other large retail stores?”; “How do England and Scotland NDCs compare to the UK’s?”
Evaluation“Is Apple’s net zero target credible?”; “How is supermarket chain Jeronimo Martins doing on their commitment to reduce food waste by 50% by 2025?”
Creation“Propose an ESG audit model that integrates social, environmental, and ethical performance indicators into one scorecard.”; “Develop a set of innovative KPIs to measure corporate progress toward net-zero goals beyond traditional carbon metrics.”
Table 1: Representative questions from the evaluation dataset.
Figure 3: Median answer token size by each model.
ModelSizeGPUGGDgpuD_{\mathrm{gpu}}DnongpuD_{\mathrm{nongpu}}NNBB
GPT-4oLargeH10080.60.588
GPT 4o-MiniMediumA10041.20.588
Table 5
ModelUGPUtotalU_{\mathrm{GPUtotal}}UGPUU_{\mathrm{GPU}}PGPUP_{\mathrm{GPU}}Pnon−GPUP_{\mathrm{non-GPU}}WtotalW_{\mathrm{total}}PUEW⋅PUEW\cdot\mathrm{PUE}
GPT-4o0.420.28755.64.60.70751.120.7924
GPT 4o-Mini0.240.1031253.23.30.3431251.120.3843
Table 6
Figure 4: Median energy use of models per query.
Figure 5: Energy use across query steps: (a) Mean energy use of models by each step of the query; (b) Proportion of energy use by step of the query.
Figure 6: Output token vs Energy Usage by model.
Figure 7: a) Mean factual score by model. b) Mean embellishment score by model.
Figure 8: a) Energy use vs. factual score of questions, by model. b) Energy use vs. embellishment score of questions, by model.
Figure 9: a) Output token size vs. factual score of questions, by model. b) Output token size vs. embellishment score of questions, by model.
More in this topic

LLM Inference Efficiency

Large language model inference efficiency reduces the time, memory, energy, and cost of generating useful outputs.

See all 351 papers

Explore similar work

CardsList
Mar 16, 2026

This Is Taking Too Long -- Investigating Time as a Proxy for Energy Consumption of LLMs

The energy consumption of Large Language Models (LLMs) is raising growing concerns due to their adverse effects on environmental stability and resource use. Yet, these energy costs remain largely opaque to users, especially when models are accessed through an API -- a black box in which all information depends on what providers choose to disclose. In this work, we investigate inference time measurements as a proxy to approximate the associated energy costs of API-based LLMs. We ground our approach by comparing our estimations with actual energy measurements from locally hosted equivalents. Our results show that time measurements allow us to infer GPU models for API-based LLMs, grounding our energy cost estimations. Our work aims to create means for understanding the associated energy costs of API-based LLMs, especially for end users.

Lars Krupp, Daniel Geißler, Francisco M. Calatrava-Nicolas +3

Embedded Intelligence DFKI and RPTU Kaiserslautern, Germany · Centre for Applied Autonomous Sensor Systems (AASS) Örebro University Örebro, Sweden

Energy-Aware LLM InferenceLLM InferenceEnergy-Efficient ML
AskPDF
Jul 29, 2026

From Tokens to Watt-hours: Analytical Energy Estimation for LLM Inference on Modern GPUs

The operational energy consumption of large language model (LLM) inference is becoming an increasingly important component of the environmental footprint of deployed AI systems. However, direct measurement of inference energy often requires hardware telemetry, power instrumentation, or infrastructure-specific monitoring, limiting its applicability in comparative studies, early-stage system design, and sustainability reporting. This report presents an analytically structured, empirically calibrated, GPU-level methodology for estimating LLM inference energy on NVIDIA H100-class accelerators without direct runtime measurement. The proposed estimator combines parameter-scaled transformer FLOP accounting, calibrated memory-traffic factors, and hardware-specific energy coefficients for FP16/BF16 tensor-core computation and high-bandwidth-memory movement. It explicitly separates prompt prefill from autoregressive decoding, enabling energy estimates for input tokens, output tokens, and complete inference requests. The methodology further decomposes total energy into compute, parameter-access, key-value-cache write, and attention-read components, allowing the scaling behavior with model size, context length, and generated-token count to be analyzed. The resulting estimates are not intended to replace physical power measurements; rather, they provide transparent, reproducible, and assumption-explicit approximations suitable for model comparison, green-coding analysis, and design-time evaluation of LLM inference workloads.

Tina Vartziotis, Rodopi Kosteli, Elli Vartziotis +5

1National Technical University of Athens, Patission Complex 42, 10682 Athens, Greece · 2Harvard University, 1350 Massachusetts Avenue, 02138 Cambridge, MA, USA · 3TWT GmbH Science & Innovation, Industriestraße 6, 70565 Stuttgart, DE +4

LLM Inference EfficiencyEnergy-Efficient MLLLM InferenceEnergy-Aware LLM Inference
AskPDF
Jun 20, 2026

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

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.

Naihao Deng, Alissa Shen, Yiming Feng +5

University of Michigan · University of Cambridge · University of Aberdeen

LLM Inference EfficiencyMultilingual Language ModelsMultilingual Language Model EvaluationLow-Resource Language Processing
AskPDF