aiqbench analysis
This paper introduces a rigorous methodology for quantifying the environmental footprint of LLM inference, specifically isolating energy consumption between prefill and decode phases across open and proprietary models. By leveraging Bayesian regression and Monte-Carlo sampling, it establishes actionable metrics for estimating per-token electricity usage, water consumption, and CO2 emissions to guide sustainability improvements in cloud AI deployments.
Greenpixie's AI Token Methodology: Assessing the Energy, Water and $\mathrm{CO_2\text{-}eq}$ Impact of AI Tokens for Open and Closed Weight Models
arXiv:2609.33965v1 Announce Type: cross Abstract: We describe a methodology for estimating the per-token energy cost of cloud-hosted large language model (LLM) inference, separating between input (prefill) and output (decode) tokens. Graphics processing unit (GPU) energy usage is measured during inference benchmarking with open-weights models on a wide range of text-based tasks. The remaining server energy contribution from non-GPU hardware is estimated from the inference wall time. Bayesian linear regression is used to model the relationship between energy per token and LLM size, request traffic, and hardware deployment configuration. Proprietary frontier LLMs of unknown size and deployment are binned into size buckets based on naming conventions and performance priors, and the space of possible LLM configurations is sampled with Monte-Carlo methods to give a representative average energy per token and uncertainty. We also describe how these energy measurements can be used to estimate the carbon-dioxide equivalent ($\mathrm{CO_2\text{-}eq}$) emissions, both usage and embodied, and water consumed per token of AI inference. This methodology provides actionable data that enables reductions in cost, electricity usage, $\mathrm{CO_2\text{-}eq}$ emitted and water consumed in cloud and Software as a Service (SaaS).
Abstract:We describe a methodology for estimating the per-token energy cost of cloud-hosted large language model (LLM) inference, separating between input (prefill) and output (decode) tokens. Graphics processing unit (GPU) energy usage is measured during inference benchmarking with open-weights models on a wide range of text-based tasks. The remaining server energy contribution from non-GPU hardware is estimated from the inference wall time. Bayesian linear regression is used to model the relationship between energy per token and LLM size, request traffic, and hardware deployment configuration. Proprietary frontier LLMs of unknown size and deployment are binned into size buckets based on naming conventions and performance priors, and the space of possible LLM configurations is sampled with Monte-Carlo methods to give a representative average energy per token and uncertainty. We also describe how these energy measurements can be used to estimate the carbon-dioxide equivalent ($\mathrm{CO_2\text{-}eq}$) emissions, both usage and embodied, and water consumed per token of AI inference. This methodology provides actionable data that enables reductions in cost, electricity usage, $\mathrm{CO_2\text{-}eq}$ emitted and water consumed in cloud and Software as a Service (SaaS).
Computers and Society (cs.CY); Machine Learning (cs.LG)
arXiv:2609.33965 [cs.CY]
(or arXiv:2609.33965v1 [cs.CY] for this version)
https://doi.org/10.48550/arXiv.2609.33965 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 277 words
