DualSpec: Accelerating Deep Research Agents via Dual-Process Action Speculation

TL;DR

DualSpec introduces a dual-process speculation framework for deep research agents, leveraging action heterogeneity to significantly reduce latency while maintaining accuracy in long-horizon tasks.

Contribution

It proposes a novel heterogeneous speculation approach that differentiates between Search and Visit actions, improving speed and robustness in deep research agents.

Findings

Achieves up to 3.28× speedup in end-to-end inference

Maintains accuracy comparable to fully reasoning agents

Validates effectiveness across multiple models and benchmarks

Abstract

Large language model-based deep research agents have been increasingly popular for addressing long-horizon information-seeking tasks, but they often incur high end-to-end latency due to extensive reasoning and frequent tool use. Speculation frameworks aim to reduce latency by overlapping action execution with reasoning; however, existing approaches typically rely on uniform speculation strategies and strict action matching, which limits inference speedups and robustness. In this work, we revisit the speculate-verify paradigm for deep research agents through the lens of action heterogeneity. We show that \textit{Search} and \textit{Visit} actions exhibit fundamentally different reasoning and model capacity requirements: entropy-based analysis reveals that Search decisions have higher uncertainty and benefit significantly from explicit reasoning, whereas Visit decisions have lower entropy and depend primarily on model capacity. Motivated by this dual-process characteristic, we propose DualSpec, a heterogeneous speculation framework equipped with a lightweight, confidence-based semantic verifier. Experiments across multiple models and benchmarks demonstrate that DualSpec achieves up to 3.28$\times$ end-to-end speedup while maintaining accuracy comparable to fully reasoning agents.