@inproceedings{liu-etal-2026-h,
title = "{H}-{MAS}: Hierarchical Multi-Agent Scheduling for Multi-Tenant {LLM} Serving",
author = "Liu, Yuhan and
Xu, Cong and
Jia, Qi and
Wang, Yihua and
Chen, Feiyu and
Jin, Liang and
Liu, Lu and
Zhao, Yaqian and
Ding, Yuting and
Li, Xiang",
editor = "Liakata, Maria and
Moreira, Viviane P. and
Zhang, Jiajun and
Jurgens, David",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
month = jul,
year = "2026",
address = "San Diego, California, United States",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-acl.1946/",
pages = "39051--39071",
ISBN = "979-8-89176-395-1",
abstract = "Multi-tenant Model-as-a-Service (MaaS) LLM serving must maintain stringent quality of service (QoS) despite heterogeneous requests competing for constrained GPU resources. In practice, MaaS workloads exhibit non-stationarity across multiple time scales, including request bursts, request-composition drift, and persistent workload shifts. Because workloads change across multiple time scales, existing request schedulers often rely on a single fixed policy (e.g., First-Come-First-Served, FCFS) that remains unchanged at runtime, which can lead to unstable QoS. We propose H-MAS, a hierarchical multi-agent scheduler that operates in a layered closed loop: a perception/prediction layer infers lightweight request attributes and cost signals; a feedback layer summarizes runtime metrics into short- and long-horizon QoS states; a hierarchical control layer updates the active scheduling policy over longer horizons and tunes execution parameters over shorter horizons; and an execution layer applies these decisions during inference. Experiments with load scaling and Azure-trace replays show that H-MAS achieves 1.2$\times${--}3.0$\times$ higher Goodput than SGLang and vLLM, and maintains more stable QoS under workload drift, diverse request lengths and heterogeneous SLO targets."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="liu-etal-2026-h">
<titleInfo>
<title>H-MAS: Hierarchical Multi-Agent Scheduling for Multi-Tenant LLM Serving</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yuhan</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Cong</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qi</namePart>
<namePart type="family">Jia</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yihua</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Feiyu</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Liang</namePart>
<namePart type="family">Jin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lu</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yaqian</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yuting</namePart>
<namePart type="family">Ding</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiang</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: ACL 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Maria</namePart>
<namePart type="family">Liakata</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Viviane</namePart>
<namePart type="given">P</namePart>
<namePart type="family">Moreira</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiajun</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">David</namePart>
<namePart type="family">Jurgens</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">San Diego, California, United States</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-395-1</identifier>
</relatedItem>
<abstract>Multi-tenant Model-as-a-Service (MaaS) LLM serving must maintain stringent quality of service (QoS) despite heterogeneous requests competing for constrained GPU resources. In practice, MaaS workloads exhibit non-stationarity across multiple time scales, including request bursts, request-composition drift, and persistent workload shifts. Because workloads change across multiple time scales, existing request schedulers often rely on a single fixed policy (e.g., First-Come-First-Served, FCFS) that remains unchanged at runtime, which can lead to unstable QoS. We propose H-MAS, a hierarchical multi-agent scheduler that operates in a layered closed loop: a perception/prediction layer infers lightweight request attributes and cost signals; a feedback layer summarizes runtime metrics into short- and long-horizon QoS states; a hierarchical control layer updates the active scheduling policy over longer horizons and tunes execution parameters over shorter horizons; and an execution layer applies these decisions during inference. Experiments with load scaling and Azure-trace replays show that H-MAS achieves 1.2\times–3.0\times higher Goodput than SGLang and vLLM, and maintains more stable QoS under workload drift, diverse request lengths and heterogeneous SLO targets.</abstract>
<identifier type="citekey">liu-etal-2026-h</identifier>
<location>
<url>https://aclanthology.org/2026.findings-acl.1946/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>39051</start>
<end>39071</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T H-MAS: Hierarchical Multi-Agent Scheduling for Multi-Tenant LLM Serving
%A Liu, Yuhan
%A Xu, Cong
%A Jia, Qi
%A Wang, Yihua
%A Chen, Feiyu
%A Jin, Liang
%A Liu, Lu
%A Zhao, Yaqian
%A Ding, Yuting
%A Li, Xiang
%Y Liakata, Maria
%Y Moreira, Viviane P.
%Y Zhang, Jiajun
%Y Jurgens, David
%S Findings of the Association for Computational Linguistics: ACL 2026
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, United States
%@ 979-8-89176-395-1
%F liu-etal-2026-h
%X Multi-tenant Model-as-a-Service (MaaS) LLM serving must maintain stringent quality of service (QoS) despite heterogeneous requests competing for constrained GPU resources. In practice, MaaS workloads exhibit non-stationarity across multiple time scales, including request bursts, request-composition drift, and persistent workload shifts. Because workloads change across multiple time scales, existing request schedulers often rely on a single fixed policy (e.g., First-Come-First-Served, FCFS) that remains unchanged at runtime, which can lead to unstable QoS. We propose H-MAS, a hierarchical multi-agent scheduler that operates in a layered closed loop: a perception/prediction layer infers lightweight request attributes and cost signals; a feedback layer summarizes runtime metrics into short- and long-horizon QoS states; a hierarchical control layer updates the active scheduling policy over longer horizons and tunes execution parameters over shorter horizons; and an execution layer applies these decisions during inference. Experiments with load scaling and Azure-trace replays show that H-MAS achieves 1.2\times–3.0\times higher Goodput than SGLang and vLLM, and maintains more stable QoS under workload drift, diverse request lengths and heterogeneous SLO targets.
%U https://aclanthology.org/2026.findings-acl.1946/
%P 39051-39071
Markdown (Informal)
[H-MAS: Hierarchical Multi-Agent Scheduling for Multi-Tenant LLM Serving](https://aclanthology.org/2026.findings-acl.1946/) (Liu et al., Findings 2026)
ACL
- Yuhan Liu, Cong Xu, Qi Jia, Yihua Wang, Feiyu Chen, Liang Jin, Lu Liu, Yaqian Zhao, Yuting Ding, and Xiang Li. 2026. H-MAS: Hierarchical Multi-Agent Scheduling for Multi-Tenant LLM Serving. In Findings of the Association for Computational Linguistics: ACL 2026, pages 39051–39071, San Diego, California, United States. Association for Computational Linguistics.