@inproceedings{gao-etal-2026-agent,
title = "Can Your Agent Read the Source Faithfully? Benchmarking {MLLM} Agents on Source-Grounded Multimodal Retrieval",
author = "Gao, Wentao and
Huang, Shizhou and
Zhuang, Junling and
Garg, Advitya and
Rahman, Mohammad Hasibur and
Zhang, Mengliang and
Qin, Qin and
Yu, Yu",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.1855/",
pages = "42175--42194",
ISBN = "979-8-89176-459-0",
abstract = "Multimodal agents are increasingly evaluated on their ability to search the web, invoke tools, and solve complex tasks autonomously. Yet no existing benchmark systematically evaluates the capability of autonomously searching, parsing, and extracting precise answers from multimodal raw media files as a unified, end-to-end agentic task. We introduce RDQA - Raw Data Question Answering, a benchmark of 1,350 questions across three source modalities (document, video, audio), and 20 real-world task scenarios, where every answer is designed to require source access rather than model knowledge alone. RDQA provides capability-stratified evaluation through 9 capability families organized by modality and cognitive complexity, and introduces the Parametric Leakage Rate (PLR), a process-aware metric that flags correct answers produced without genuine evidence retrieval. We evaluate 12 agent systems spanning 6 backbone models and 2 scaffoldings. Our results show that the the best generic agent reaches only 24.0{\%} adjusted accuracy, while a process-aware baseline (RDQA-Skill) reaches 41.1{\%}. PLR reaches as high as 96.0{\%}, indicating that many ostensibly correct answers are not grounded in retrieved evidence. We release the full dataset, evaluation framework, and a baseline to support future research."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="gao-etal-2026-agent">
<titleInfo>
<title>Can Your Agent Read the Source Faithfully? Benchmarking MLLM Agents on Source-Grounded Multimodal Retrieval</title>
</titleInfo>
<name type="personal">
<namePart type="given">Wentao</namePart>
<namePart type="family">Gao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shizhou</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Junling</namePart>
<namePart type="family">Zhuang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Advitya</namePart>
<namePart type="family">Garg</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mohammad</namePart>
<namePart type="given">Hasibur</namePart>
<namePart type="family">Rahman</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mengliang</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qin</namePart>
<namePart type="family">Qin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yu</namePart>
<namePart type="family">Yu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Multimodal agents are increasingly evaluated on their ability to search the web, invoke tools, and solve complex tasks autonomously. Yet no existing benchmark systematically evaluates the capability of autonomously searching, parsing, and extracting precise answers from multimodal raw media files as a unified, end-to-end agentic task. We introduce RDQA - Raw Data Question Answering, a benchmark of 1,350 questions across three source modalities (document, video, audio), and 20 real-world task scenarios, where every answer is designed to require source access rather than model knowledge alone. RDQA provides capability-stratified evaluation through 9 capability families organized by modality and cognitive complexity, and introduces the Parametric Leakage Rate (PLR), a process-aware metric that flags correct answers produced without genuine evidence retrieval. We evaluate 12 agent systems spanning 6 backbone models and 2 scaffoldings. Our results show that the the best generic agent reaches only 24.0% adjusted accuracy, while a process-aware baseline (RDQA-Skill) reaches 41.1%. PLR reaches as high as 96.0%, indicating that many ostensibly correct answers are not grounded in retrieved evidence. We release the full dataset, evaluation framework, and a baseline to support future research.</abstract>
<identifier type="citekey">gao-etal-2026-agent</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1855/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>42175</start>
<end>42194</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Can Your Agent Read the Source Faithfully? Benchmarking MLLM Agents on Source-Grounded Multimodal Retrieval
%A Gao, Wentao
%A Huang, Shizhou
%A Zhuang, Junling
%A Garg, Advitya
%A Rahman, Mohammad Hasibur
%A Zhang, Mengliang
%A Qin, Qin
%A Yu, Yu
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F gao-etal-2026-agent
%X Multimodal agents are increasingly evaluated on their ability to search the web, invoke tools, and solve complex tasks autonomously. Yet no existing benchmark systematically evaluates the capability of autonomously searching, parsing, and extracting precise answers from multimodal raw media files as a unified, end-to-end agentic task. We introduce RDQA - Raw Data Question Answering, a benchmark of 1,350 questions across three source modalities (document, video, audio), and 20 real-world task scenarios, where every answer is designed to require source access rather than model knowledge alone. RDQA provides capability-stratified evaluation through 9 capability families organized by modality and cognitive complexity, and introduces the Parametric Leakage Rate (PLR), a process-aware metric that flags correct answers produced without genuine evidence retrieval. We evaluate 12 agent systems spanning 6 backbone models and 2 scaffoldings. Our results show that the the best generic agent reaches only 24.0% adjusted accuracy, while a process-aware baseline (RDQA-Skill) reaches 41.1%. PLR reaches as high as 96.0%, indicating that many ostensibly correct answers are not grounded in retrieved evidence. We release the full dataset, evaluation framework, and a baseline to support future research.
%U https://aclanthology.org/2026.emnlp-main.1855/
%P 42175-42194
Markdown (Informal)
[Can Your Agent Read the Source Faithfully? Benchmarking MLLM Agents on Source-Grounded Multimodal Retrieval](https://aclanthology.org/2026.emnlp-main.1855/) (Gao et al., EMNLP 2026)
ACL
- Wentao Gao, Shizhou Huang, Junling Zhuang, Advitya Garg, Mohammad Hasibur Rahman, Mengliang Zhang, Qin Qin, and Yu Yu. 2026. Can Your Agent Read the Source Faithfully? Benchmarking MLLM Agents on Source-Grounded Multimodal Retrieval. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 42175–42194, Budapest, Hungary. Association for Computational Linguistics.