BOute: Cost-Efficient LLM Serving with Heterogeneous LLMs and GPUs via Multi-Objective Bayesian Optimization

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Jiang, Youhe, Fu, Fangcheng, Yoneki, Eiko
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918332378120192
author Jiang, Youhe
Fu, Fangcheng
Yoneki, Eiko
author_facet Jiang, Youhe
Fu, Fangcheng
Yoneki, Eiko
contents The rapid growth of large language model (LLM) deployments has made cost-efficient serving systems essential. Recent efforts to enhance system cost-efficiency adopt two main perspectives: (i) An algorithmic perspective that exploits heterogeneous model capabilities to route simpler queries to lower-cost models and complex queries to higher-cost models (i.e., heterogeneous query routing); and (ii) a systems perspective that utilizes heterogeneous GPU resources as cost-effective alternatives to homogeneous high-end GPUs (i.e., heterogeneous model deployment). However, algorithm-system co-design for cost-efficient LLM serving necessitates sophisticated management: (i) Determining optimal query routing strategies under latency and quality requirements, (ii) configuring model deployment across heterogeneous GPUs with appropriate resource allocation and parallelism strategies, and (iii) co-optimizing routing and deployment decisions to maximize overall system performance. To address these challenges, we present BOute, a quality-aware scheduling system that jointly exploits heterogeneous model and GPU capabilities for cost-efficient LLM serving. BOute employs a multi-objective Bayesian optimization (MOBO) framework to co-optimize the routing strategy and model deployment, thereby maximizing the cost-efficiency of the serving system while guaranteeing response quality. Evaluation results demonstrate that BOute outperforms state-of-the-art LLM serving systems by up to 157% and 59% on average under identical cost budgets and quality requirements, or reducing serving costs by 15%-61% (38% on average) while maintaining the same performance targets, validating its effectiveness in achieving cost-efficient LLM serving.
format Preprint
id arxiv_https___arxiv_org_abs_2602_10729
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle BOute: Cost-Efficient LLM Serving with Heterogeneous LLMs and GPUs via Multi-Objective Bayesian Optimization
Jiang, Youhe
Fu, Fangcheng
Yoneki, Eiko
Distributed, Parallel, and Cluster Computing
The rapid growth of large language model (LLM) deployments has made cost-efficient serving systems essential. Recent efforts to enhance system cost-efficiency adopt two main perspectives: (i) An algorithmic perspective that exploits heterogeneous model capabilities to route simpler queries to lower-cost models and complex queries to higher-cost models (i.e., heterogeneous query routing); and (ii) a systems perspective that utilizes heterogeneous GPU resources as cost-effective alternatives to homogeneous high-end GPUs (i.e., heterogeneous model deployment). However, algorithm-system co-design for cost-efficient LLM serving necessitates sophisticated management: (i) Determining optimal query routing strategies under latency and quality requirements, (ii) configuring model deployment across heterogeneous GPUs with appropriate resource allocation and parallelism strategies, and (iii) co-optimizing routing and deployment decisions to maximize overall system performance. To address these challenges, we present BOute, a quality-aware scheduling system that jointly exploits heterogeneous model and GPU capabilities for cost-efficient LLM serving. BOute employs a multi-objective Bayesian optimization (MOBO) framework to co-optimize the routing strategy and model deployment, thereby maximizing the cost-efficiency of the serving system while guaranteeing response quality. Evaluation results demonstrate that BOute outperforms state-of-the-art LLM serving systems by up to 157% and 59% on average under identical cost budgets and quality requirements, or reducing serving costs by 15%-61% (38% on average) while maintaining the same performance targets, validating its effectiveness in achieving cost-efficient LLM serving.
title BOute: Cost-Efficient LLM Serving with Heterogeneous LLMs and GPUs via Multi-Objective Bayesian Optimization
topic Distributed, Parallel, and Cluster Computing
url https://arxiv.org/abs/2602.10729