AgentCPM-Report: Interleaving Drafting and Deepening for Open-Ended Deep Research

TL;DR

This paper introduces AgentCPM-Report, a local, lightweight deep research system that mimics human writing by dynamically revising outlines and alternating between drafting and deepening, achieving superior results without relying on large online models.

Contribution

It proposes a novel framework with a Writing As Reasoning Policy and a multi-stage training strategy, enabling small models to perform deep research tasks effectively.

Findings

Outperforms leading closed-source systems on benchmark datasets.

Achieves substantial gains in insight quality.

Demonstrates effective information acquisition and knowledge refinement.

Abstract

Generating deep research reports requires large-scale information acquisition and the synthesis of insight-driven analysis, posing a significant challenge for current language models. Most existing approaches follow a plan-then-write paradigm, whose performance heavily depends on the quality of the initial outline. However, constructing a comprehensive outline itself demands strong reasoning ability, causing current deep research systems to rely almost exclusively on closed-source or online large models. This reliance raises practical barriers to deployment and introduces safety and privacy concerns for user-authored data. In this work, we present AgentCPM-Report, a lightweight yet high-performing local solution composed of a framework that mirrors the human writing process and an 8B-parameter deep research agent. Our framework uses a Writing As Reasoning Policy (WARP), which enables models to dynamically revise outlines during report generation. Under this policy, the agent alternates between Evidence-Based Drafting and Reasoning-Driven Deepening, jointly supporting information acquisition, knowledge refinement, and iterative outline evolution. To effectively equip small models with this capability, we introduce a Multi-Stage Agentic Training strategy, consisting of cold-start, atomic skill RL, and holistic pipeline RL. Experiments on DeepResearch Bench, DeepConsult, and DeepResearch Gym demonstrate that AgentCPM-Report outperforms leading closed-source systems, with substantial gains in Insight.