Reinforcement learning (RL) is central to post-training, particularly for agentic models that require specialized reasoning behaviors. In this setting, model merging offers a practical mechanism for integrating multiple RL-trained agents from different tasks into a single generalist model. However, existing merging methods are designed for supervised fine-tuning (SFT), and they are suboptimal to preserve task-specific capabilities on RL-trained agentic models. The root is a task-vector mismatch between RL and SFT: on-policy RL induces task vectors that are highly sparse and heterogeneous, whereas SFT-style merging implicitly assumes dense and globally comparable task vectors. When standard global averaging is applied under this mismatch, RL's non-overlapping task vectors that encode critical task-specific behaviors are reduced and parameter updates are diluted. To address this issue, we propose Reinforced Agent Merging (RAM), a distribution-aware merging framework explicitly designed for RL-trained agentic models. RAM disentangles shared and task-specific unique parameter updates, averaging shared components while selectively preserving and rescaling unique ones to counteract parameter update dilution. Experiments across multiple agent domains and model architectures demonstrate that RAM not only surpasses merging baselines, but also unlocks synergistic potential among agents to achieve performance superior to that of specialized agents in their domains.

翻译：强化学习（RL）是后训练阶段的核心，尤其对于需要专门推理行为的智能体模型而言。在此背景下，模型融合提供了一种实用机制，可将来自不同任务的多个经过RL训练的智能体整合为一个通用模型。然而，现有的融合方法专为监督微调（SFT）设计，在保留RL训练智能体模型的任务特定能力方面效果欠佳。其根源在于RL与SFT之间存在任务向量失配：基于策略的RL产生的任务向量高度稀疏且异质，而SFT式融合方法隐含假设任务向量密集且全局可比。在这种失配情况下应用标准全局平均法时，RL中编码关键任务特定行为的非重叠任务向量会被削弱，参数更新亦被稀释。为解决此问题，我们提出强化智能体融合（RAM），这是一个专为RL训练智能体模型设计的分布感知融合框架。RAM解耦共享参数更新与任务特定独特参数更新，对共享组件进行平均处理，同时选择性保留并重新缩放独特组件以抵消参数更新稀释。跨多个智能体领域和模型架构的实验表明，RAM不仅超越了现有融合基线方法，更能释放智能体间的协同潜力，实现优于各领域专用智能体的性能表现。