# Demystifying deep search: a holistic evaluation with hint-free multi-hop questions and factorised metrics

arXiv: https://arxiv.org/abs/2510.05137 · HTML: https://mlsanity.com/paper/2510.05137/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Agentic RAG, agents and memory, one of 15 directions in Retrieval-augmented generation.
- Builds on 30 of its 41 references that are in the map.
- First posted on arXiv 1 Oct 2025.

## Placed in

- Retrieval-augmented generation / [Agentic RAG, agents and memory](https://mlsanity.com/rag/agentic-rag-agents-and-memory/)
- LLM agents and tool use / [Web, search, and research agents](https://mlsanity.com/agents/web-search-and-research-agents/)

## Builds on (most cited first, 30 of 30)

- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025)
- [Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities](https://mlsanity.com/paper/2507.06261/) (2025)
- [HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering](https://mlsanity.com/paper/1809.09600/) (2018)
- [SWE-bench: Can Language Models Resolve Real-World GitHub Issues?](https://mlsanity.com/paper/2310.06770/) (2023)
- [WebArena: A Realistic Web Environment for Building Autonomous Agents](https://mlsanity.com/paper/2307.13854/) (2023)
- [WebGPT: Browser-assisted question-answering with human feedback](https://mlsanity.com/paper/2112.09332/) (2021)
- [gpt-oss-120b & gpt-oss-20b Model Card](https://mlsanity.com/paper/2508.10925/) (2025)
- [Constructing A Multi-hop QA Dataset for Comprehensive Evaluation of Reasoning Steps](https://mlsanity.com/paper/2011.01060/) (2020)
- [AgentBench: Evaluating LLMs as Agents](https://mlsanity.com/paper/2308.03688/) (2023)
- [MuSiQue: Multihop Questions via Single-hop Question Composition](https://mlsanity.com/paper/2108.00573/) (2021)
- [BrowseComp: A Simple Yet Challenging Benchmark for Browsing Agents](https://mlsanity.com/paper/2504.12516/) (2025)
- [KILT: a Benchmark for Knowledge Intensive Language Tasks](https://mlsanity.com/paper/2009.02252/) (2020)
- [Adaptive-RAG: Learning to Adapt Retrieval-Augmented Large Language Models through Question Complexity](https://mlsanity.com/paper/2403.14403/) (2024)
- [Search-o1: Agentic Search-Enhanced Large Reasoning Models](https://mlsanity.com/paper/2501.05366/) (2025)
- [WebThinker: Empowering Large Reasoning Models with Deep Research Capability](https://mlsanity.com/paper/2504.21776/) (2025)
- [TheAgentCompany: Benchmarking LLM Agents on Consequential Real World Tasks](https://mlsanity.com/paper/2412.14161/) (2024)
- [Agentic Retrieval-Augmented Generation: A Survey on Agentic RAG](https://mlsanity.com/paper/2501.09136/) (2025)
- [DeepResearch Bench: A Comprehensive Benchmark for Deep Research Agents](https://mlsanity.com/paper/2506.11763/) (2025)
- [ZeroSearch: Incentivize the Search Capability of LLMs without Searching](https://mlsanity.com/paper/2505.04588/) (2025)
- [Constructing Datasets for Multi-hop Reading Comprehension Across Documents](https://mlsanity.com/paper/1710.06481/) (2017)
- [From LLM Reasoning to Autonomous AI Agents: A Comprehensive Review](https://mlsanity.com/paper/2504.19678/) (2025)
- [WebShaper: Agentically Data Synthesizing via Information-Seeking Formalization](https://mlsanity.com/paper/2507.15061/) (2025)
- [Multi-hop Question Answering via Reasoning Chains](https://mlsanity.com/paper/1910.02610/) (2019)
- [s3: You Don't Need That Much Data to Train a Search Agent via RL](https://mlsanity.com/paper/2505.14146/) (2025)
- [A Systematic Review of Key Retrieval-Augmented Generation (RAG) Systems: Progress, Gaps, and Future Directions](https://mlsanity.com/paper/2507.18910/) (2025)
- [FanOutQA: A Multi-Hop, Multi-Document Question Answering Benchmark for Large Language Models](https://mlsanity.com/paper/2402.14116/) (2024)
- [HopRAG: Multi-Hop Reasoning for Logic-Aware Retrieval-Augmented Generation](https://mlsanity.com/paper/2502.12442/) (2025)
- [MINTQA: A Multi-Hop Question Answering Benchmark for Evaluating LLMs on New and Tail Knowledge](https://mlsanity.com/paper/2412.17032/) (2024)
- [TRACE the Evidence: Constructing Knowledge-Grounded Reasoning Chains for Retrieval-Augmented Generation](https://mlsanity.com/paper/2406.11460/) (2024)

## Built on by (most cited first, 4 of 4)

- [Failure Modes in Multi-Hop QA: The Weakest Link Effect and the Recognition Bottleneck](https://mlsanity.com/paper/2601.12499/) (2026)
- [When Iterative RAG Beats Ideal Evidence: A Diagnostic Study in Scientific Multi-hop Question Answering](https://mlsanity.com/paper/2601.19827/) (2026)
- [Benchmarking Hybrid Deep Research Across Database Querying and Web Search](https://mlsanity.com/paper/2609.09410/) (2026)
- [PrimeSeeker: Capability-Oriented Supervision for Deep Search Agents](https://mlsanity.com/paper/2609.35816/) (2026)

## Abstract (arXiv)

RAG (Retrieval-Augmented Generation) systems and web agents are increasingly evaluated on multi-hop deep search tasks, yet current practice suffers from two major limitations. First, most benchmarks leak the reasoning path in the question text, allowing models to follow surface cues rather than discover reasoning chains autonomously. Second, evaluation is typically reduced to a single pass rate, which collapses diverse behaviours into one score and obscures whether failures stem from inadequate search, poor knowledge use, or inappropriate refusal. To address these issues, we present WebDetective, a benchmark of hint-free multi-hop questions paired with a controlled Wikipedia sandbox that ensures full traceability of model actions, and a holistic evaluation framework that separates search sufficiency, knowledge utilisation, and refusal behaviour. Our evaluation of 25 state-of-the-art models reveals systematic weaknesses across all architectures: models struggle with knowledge utilisation despite having sufficient evidence and demonstrate near-absent appropriate refusal when evidence is lacking. These patterns expose a fundamental gap: today's systems excel at executing given reasoning paths but fail when required to discover them. We develop an agentic workflow, EvidenceLoop, that explicitly targets the challenges our benchmark identifies, incorporating verification loops and systematic evidence tracking that improve both search and synthesis capabilities. This baseline demonstrates that WebDetective's diagnostic framework can guide concrete architectural improvements, establishing our benchmark as a critical tool for developing genuinely autonomous reasoning systems rather than pattern-following agents.