@inproceedings{jiang-etal-2026-emtir,
title = "{EMTIR}-{GRPO}: Efficient Multi-Tool Augmented Large Language Models via Reinforcement Learning",
author = "Jiang, Shixin and
Zhu, Zhihao and
Liang, Jiafeng and
Wu, Yang and
Liu, Ming and
Qin, Bing",
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.1388/",
pages = "27877--27894",
ISBN = "979-8-89176-395-1",
abstract = "Tool-integrated reasoning (TIR) enables large language models (LLMs) to invoke external tools for tasks beyond their internal capacity but often suffers from tool overuse.Existing approaches leverage imitation learning or reward shaping to improve efficiency, yet mainly target single-tool scenarios and ignore the varying invocation costs across tools in multi-tool reasoning (MTIR). To address these gaps, we propose EMTIR-GRPO, a simple yet effective RL algorithm for cost-aware MTIR. Built upon GRPO, we introduce a composite reward considering format completeness, answer correctness, and tool efficiency.By incorporating a cost-aware coefficient with group optimal cost estimation, EMTIR-GRPO explicitly models heterogeneous tool costs and encourages more cost-effective tool-use strategies. Experiments on MTIR-QA and MTIR-TC demonstrate significant efficiency gains (e.g., $\Delta$+10.9 on Tool-Star-7B and $\Delta$+3.6 on ReCall-7B) while maintaining or even improving accuracy (e.g., 55.4 vs. 52.0 on Tool-Star-7B). Additional budget-constrained and tool-free evaluations further validate its effectiveness in maximizing cost-efficiency and reducing cognitive offloading."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="jiang-etal-2026-emtir">
<titleInfo>
<title>EMTIR-GRPO: Efficient Multi-Tool Augmented Large Language Models via Reinforcement Learning</title>
</titleInfo>
<name type="personal">
<namePart type="given">Shixin</namePart>
<namePart type="family">Jiang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhihao</namePart>
<namePart type="family">Zhu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiafeng</namePart>
<namePart type="family">Liang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yang</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ming</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bing</namePart>
<namePart type="family">Qin</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>Tool-integrated reasoning (TIR) enables large language models (LLMs) to invoke external tools for tasks beyond their internal capacity but often suffers from tool overuse.Existing approaches leverage imitation learning or reward shaping to improve efficiency, yet mainly target single-tool scenarios and ignore the varying invocation costs across tools in multi-tool reasoning (MTIR). To address these gaps, we propose EMTIR-GRPO, a simple yet effective RL algorithm for cost-aware MTIR. Built upon GRPO, we introduce a composite reward considering format completeness, answer correctness, and tool efficiency.By incorporating a cost-aware coefficient with group optimal cost estimation, EMTIR-GRPO explicitly models heterogeneous tool costs and encourages more cost-effective tool-use strategies. Experiments on MTIR-QA and MTIR-TC demonstrate significant efficiency gains (e.g., Δ+10.9 on Tool-Star-7B and Δ+3.6 on ReCall-7B) while maintaining or even improving accuracy (e.g., 55.4 vs. 52.0 on Tool-Star-7B). Additional budget-constrained and tool-free evaluations further validate its effectiveness in maximizing cost-efficiency and reducing cognitive offloading.</abstract>
<identifier type="citekey">jiang-etal-2026-emtir</identifier>
<location>
<url>https://aclanthology.org/2026.findings-acl.1388/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>27877</start>
<end>27894</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T EMTIR-GRPO: Efficient Multi-Tool Augmented Large Language Models via Reinforcement Learning
%A Jiang, Shixin
%A Zhu, Zhihao
%A Liang, Jiafeng
%A Wu, Yang
%A Liu, Ming
%A Qin, Bing
%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 jiang-etal-2026-emtir
%X Tool-integrated reasoning (TIR) enables large language models (LLMs) to invoke external tools for tasks beyond their internal capacity but often suffers from tool overuse.Existing approaches leverage imitation learning or reward shaping to improve efficiency, yet mainly target single-tool scenarios and ignore the varying invocation costs across tools in multi-tool reasoning (MTIR). To address these gaps, we propose EMTIR-GRPO, a simple yet effective RL algorithm for cost-aware MTIR. Built upon GRPO, we introduce a composite reward considering format completeness, answer correctness, and tool efficiency.By incorporating a cost-aware coefficient with group optimal cost estimation, EMTIR-GRPO explicitly models heterogeneous tool costs and encourages more cost-effective tool-use strategies. Experiments on MTIR-QA and MTIR-TC demonstrate significant efficiency gains (e.g., Δ+10.9 on Tool-Star-7B and Δ+3.6 on ReCall-7B) while maintaining or even improving accuracy (e.g., 55.4 vs. 52.0 on Tool-Star-7B). Additional budget-constrained and tool-free evaluations further validate its effectiveness in maximizing cost-efficiency and reducing cognitive offloading.
%U https://aclanthology.org/2026.findings-acl.1388/
%P 27877-27894
Markdown (Informal)
[EMTIR-GRPO: Efficient Multi-Tool Augmented Large Language Models via Reinforcement Learning](https://aclanthology.org/2026.findings-acl.1388/) (Jiang et al., Findings 2026)
ACL