Energy-Aware Data-Driven Model Selection in LLM-Orchestrated AI Systems

2025-11-30

AI Summary

Existing LLM-driven AI collaboration systems suffer from inaccurate model selection, low accuracy, and poor energy efficiency due to the lack of quantitative performance and energy consumption characterization in model descriptions. This paper proposes GUIDE, the first framework to integrate fine-grained performance–energy trade-off modeling into LLM scheduling. GUIDE employs a data-driven approach to dynamically fuse empirically measured accuracy, latency, and power consumption across multiple models, enabling energy-aware real-time model selection. Its key innovation lies in constructing interpretable and generalizable quantitative model representations—explicitly embedded within the LLM’s reasoning chain—to significantly enhance decision quality and energy efficiency. Experiments across diverse multitask scenarios demonstrate that GUIDE improves accuracy by 0.90%–11.92%, reduces energy consumption by up to 54%, and slashes model selection latency from 4.51 seconds to just 7.2 milliseconds.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Planning, Routing, and Scheduling: Planning with Language Models
• Natural Language Processing: (Large) Language Models

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

As modern artificial intelligence (AI) systems become more advanced and capable, they can leverage a wide range of tools and models to perform complex tasks. Today, the task of orchestrating these models is often performed by Large Language Models (LLMs) that rely on qualitative descriptions of models for decision-making. However, the descriptions provided to these LLM-based orchestrators do not reflect true model capabilities and performance characteristics, leading to suboptimal model selection, reduced accuracy, and increased energy costs. In this paper, we conduct an empirical analysis of LLM-based orchestration limitations and propose GUIDE, a new energy-aware model selection framework that accounts for performance-energy trade-offs by incorporating quantitative model performance characteristics in decision-making. Experimental results demonstrate that GUIDE increases accuracy by 0.90%-11.92% across various evaluated tasks, and achieves up to 54% energy efficiency improvement, while reducing orchestrator model selection latency from 4.51 s to 7.2 ms.

Problem

• LLM-based orchestrators lack quantitative model performance data
• Suboptimal model selection reduces accuracy and increases energy costs
• Proposing an energy-aware framework to improve selection efficiency and accuracy

Innovation

• Energy-aware framework for model selection in AI systems
• Incorporates quantitative performance data for better decisions
• Reduces latency and improves energy efficiency significantly