InfoMosaic-Bench: Evaluating Multi-Source Information Seeking in Tool-Augmented Agents
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866918154387587072 |
|---|---|
| author | Du, Yaxin Zhang, Yuanshuo Yang, Xiyuan Zhou, Yifan Wang, Cheng Zou, Gongyi Pang, Xianghe Wang, Wenhao Chen, Menglan Tang, Shuo Li, Zhiyu Xiong, Feiyu Chen, Siheng |
| author_facet | Du, Yaxin Zhang, Yuanshuo Yang, Xiyuan Zhou, Yifan Wang, Cheng Zou, Gongyi Pang, Xianghe Wang, Wenhao Chen, Menglan Tang, Shuo Li, Zhiyu Xiong, Feiyu Chen, Siheng |
| contents | Information seeking is a fundamental requirement for humans. However, existing LLM agents rely heavily on open-web search, which exposes two fundamental weaknesses: online content is noisy and unreliable, and many real-world tasks require precise, domain-specific knowledge unavailable from the web. The emergence of the Model Context Protocol (MCP) now allows agents to interface with thousands of specialized tools, seemingly resolving this limitation. Yet it remains unclear whether agents can effectively leverage such tools -- and more importantly, whether they can integrate them with general-purpose search to solve complex tasks. Therefore, we introduce InfoMosaic-Bench, the first benchmark dedicated to multi-source information seeking in tool-augmented agents. Covering six representative domains (medicine, finance, maps, video, web, and multi-domain integration), InfoMosaic-Bench requires agents to combine general-purpose search with domain-specific tools. Tasks are synthesized with InfoMosaic-Flow, a scalable pipeline that grounds task conditions in verified tool outputs, enforces cross-source dependencies, and filters out shortcut cases solvable by trivial lookup. This design guarantees both reliability and non-triviality. Experiments with 14 state-of-the-art LLM agents reveal three findings: (i) web information alone is insufficient, with GPT-5 achieving only 38.2% accuracy and 67.5% pass rate; (ii) domain tools provide selective but inconsistent benefits, improving some domains while degrading others; and (iii) 22.4% of failures arise from incorrect tool usage or selection, highlighting that current LLMs still struggle with even basic tool handling. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_02271 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | InfoMosaic-Bench: Evaluating Multi-Source Information Seeking in Tool-Augmented Agents Du, Yaxin Zhang, Yuanshuo Yang, Xiyuan Zhou, Yifan Wang, Cheng Zou, Gongyi Pang, Xianghe Wang, Wenhao Chen, Menglan Tang, Shuo Li, Zhiyu Xiong, Feiyu Chen, Siheng Computation and Language Artificial Intelligence Information seeking is a fundamental requirement for humans. However, existing LLM agents rely heavily on open-web search, which exposes two fundamental weaknesses: online content is noisy and unreliable, and many real-world tasks require precise, domain-specific knowledge unavailable from the web. The emergence of the Model Context Protocol (MCP) now allows agents to interface with thousands of specialized tools, seemingly resolving this limitation. Yet it remains unclear whether agents can effectively leverage such tools -- and more importantly, whether they can integrate them with general-purpose search to solve complex tasks. Therefore, we introduce InfoMosaic-Bench, the first benchmark dedicated to multi-source information seeking in tool-augmented agents. Covering six representative domains (medicine, finance, maps, video, web, and multi-domain integration), InfoMosaic-Bench requires agents to combine general-purpose search with domain-specific tools. Tasks are synthesized with InfoMosaic-Flow, a scalable pipeline that grounds task conditions in verified tool outputs, enforces cross-source dependencies, and filters out shortcut cases solvable by trivial lookup. This design guarantees both reliability and non-triviality. Experiments with 14 state-of-the-art LLM agents reveal three findings: (i) web information alone is insufficient, with GPT-5 achieving only 38.2% accuracy and 67.5% pass rate; (ii) domain tools provide selective but inconsistent benefits, improving some domains while degrading others; and (iii) 22.4% of failures arise from incorrect tool usage or selection, highlighting that current LLMs still struggle with even basic tool handling. |
| title | InfoMosaic-Bench: Evaluating Multi-Source Information Seeking in Tool-Augmented Agents |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2510.02271 |