# Think Big, Search Small: Where Capacity Matters in Hierarchical Search Agents?

arXiv: https://arxiv.org/abs/2607.07548 · HTML: https://mlsanity.com/paper/2607.07548/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Knowledge- and tool-augmented reasoning, one of 17 directions in LLM reasoning.
- Builds on 23 of its 38 references that are in the map.
- First posted on arXiv 8 Jul 2026.

## Placed in

- LLM reasoning / [Knowledge- and tool-augmented reasoning](https://mlsanity.com/lr/knowledge-and-tool-augmented-reasoning/)
- LLM agents and tool use / [Web, search, and research agents](https://mlsanity.com/agents/web-search-and-research-agents/)

## Builds on (most cited first, 23 of 23)

- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning](https://mlsanity.com/paper/2503.09516/) (2025)
- [Kimi K2.5: Visual Agentic Intelligence](https://mlsanity.com/paper/2602.02276/) (2026)
- [The Faiss library](https://mlsanity.com/paper/2401.08281/) (2024)
- [R1-Searcher: Incentivizing the Search Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2503.05592/) (2025)
- [Beyond Ten Turns: Unlocking Long-Horizon Agentic Search with Large-Scale Asynchronous RL](https://mlsanity.com/paper/2508.07976/) (2025)
- [Context Length Alone Hurts LLM Performance Despite Perfect Retrieval](https://mlsanity.com/paper/2510.05381/) (2025)
- [WebSailor-V2: Bridging the Chasm to Proprietary Agents via Synthetic Data and Scalable Reinforcement Learning](https://mlsanity.com/paper/2509.13305/) (2025)
- [SimpleDeepSearcher: Deep Information Seeking via Web-Powered Reasoning Trajectory Synthesis](https://mlsanity.com/paper/2505.16834/) (2025)
- [Deep Research: A Systematic Survey](https://mlsanity.com/paper/2512.02038/) (2025)
- [Adaptation of Agentic AI: A Survey of Post-Training, Memory, and Skills](https://mlsanity.com/paper/2512.16301/) (2025)
- [Reasoning and Tool-use Compete in Agentic RL:From Quantifying Interference to Disentangled Tuning](https://mlsanity.com/paper/2602.00994/) (2026)
- [Small Language Models for Agentic Systems: A Survey of Architectures, Capabilities, and Deployment Trade offs](https://mlsanity.com/paper/2510.03847/) (2025)
- [Advancing General-Purpose Reasoning Models with Modular Gradient Surgery](https://mlsanity.com/paper/2602.02301/) (2026)
- [WideSeek-R1: Exploring Width Scaling for Broad Information Seeking via Multi-Agent Reinforcement Learning](https://mlsanity.com/paper/2602.04634/) (2026)
- [AI-SearchPlanner: Modular Agentic Search via Pareto-Optimal Multi-Objective Reinforcement Learning](https://mlsanity.com/paper/2508.20368/) (2025)
- [QAgent: A modular Search Agent with Interactive Query Understanding](https://mlsanity.com/paper/2510.08383/) (2025)
- [OpenSeeker-v2: Pushing the Limits of Search Agents with Informative and High-Difficulty Trajectories](https://mlsanity.com/paper/2605.04036/) (2026)
- [MAO-ARAG: Multi-Agent Orchestration for Adaptive Retrieval-Augmented Generation](https://mlsanity.com/paper/2508.01005/) (2025)
- [SearchSwarm: Towards Delegation Intelligence in Agentic LLMs for Long-Horizon Deep Research](https://mlsanity.com/paper/2606.09730/) (2026)
- [Terminus-4B: Can a Smaller Model Replace Frontier LLMs at Agentic Execution Tasks?](https://mlsanity.com/paper/2605.03195/) (2026)
- [Retrieval, Reward, and Training Protocols: What Matters in Training Search Agents?](https://mlsanity.com/paper/2605.27881/) (2026)

## Built on by (most cited first, 3 of 3)

- [What Can Component-Replacement Evidence Establish? A Critical Scoping Review of Local Decisions in LLM Agents](https://mlsanity.com/paper/2609.39989/) (2026)
- [Continually Post-training Small Models via Synthetic Self-distillation](https://mlsanity.com/paper/2610.13113/) (2026)
- [Avoiding What Your Teacher Already Knows: Entity Familiarity for Deep-Search Distillation](https://mlsanity.com/paper/2610.16251/) (2026)

## Abstract (arXiv)

Large language model based search agents increasingly adopt multi-agent architectures in which a main agent decomposes a complex question into sub-queries and dispatches them to parallel sub-agents. However, existing systems instantiate all roles from a single model of identical scale, leaving open how model capacity should be distributed across roles. We factorize hierarchical search into three roles: a delegation role responsible for task decomposition, an execution role responsible for retrieval and evidence extraction, and an answer generation role held fixed as a confound control. We then conduct controlled capacity sweeps along the delegation and execution axes on five multi-hop QA benchmarks. The experiments yield three findings. First, role factorization consistently outperforms a single-agent baseline, improving exact match from 4.5 to 8.6 points across six model scales. Second, capacity sensitivity is asymmetric: scaling the delegation backbone improves EM by ~11 points, whereas scaling the execution sub-agent moves EM by only ~2.6 points, identifying decomposition as the capability bottleneck. Third, a 1.7B-parameter executor trained via quality-filtered trajectory distillation matches a frontier sub-agent in accuracy while consuming 37% fewer sub-agent tokens, advancing the Pareto frontier. These results suggest a concrete recipe for building hierarchical search agents: concentrate capacity at delegation and downsize execution without sacrificing accuracy. Our code is available at https://github.com/QinnanCai0115/role-factorized-search.