AgentFugue: Agent Scaling for Long-Horizon Tasks through Collective Reasoning

2026-05-23

AI Summary

This work addresses the challenge of enhancing agent performance in long-horizon tasks without relying on explicit role assignment or workflow orchestration. It proposes AgentFugue, a novel framework that treats horizontal scaling of multiple peer agents as an intrinsic capability augmentation mechanism. AgentFugue introduces a plug-and-play shared reasoning hub, enabling agents to exchange intermediate findings through concise, note-like representations and thereby achieve decentralized collaborative reasoning. The framework is trained via a combination of supervised fine-tuning and end-to-end reinforcement learning. Experimental results across several challenging long-horizon tasks demonstrate that AgentFugue significantly outperforms strong baselines, confirming that collective reasoning can be effectively harnessed to yield measurable performance gains.

Abstract

Recent progress on long-horizon agentic tasks has been driven largely by scaling up individual agents through stronger models, better tools, and more effective scaffolding. In contrast, much less is understood about scaling out: whether multiple peer agents, all targeting the same task, can become an additional source of capability without relying on explicit role specialization or workflow orchestration. We study this question and propose AgentFugue, a collective reasoning framework built around a shared reasoning hub. As peer agents explore the same task in parallel, the hub records concise notes on what each agent has established, attempted, or ruled out, and enables each agent to selectively access what other agents have discovered in a form useful for its current search. This design turns otherwise isolated trajectories into a connected ecology of reusable intermediate reasoning without requiring centralized planning. We instantiate the hub as a plug-in communication layer, trained with supervised fine-tuning and end-to-end reinforcement learning. Across the challenging long-horizon settings we study, AgentFugue improves over strong baselines. Our results suggest that collective reasoning can turn scaling out peer agent systems into a distinct source of capability gains, rather than merely a way of spending more compute.

Problem

• long-horizon tasks
• multi-agent systems
• collective reasoning
• scaling out
• peer agents

Innovation

• collective reasoning
• agent scaling
• shared reasoning hub
• long-horizon tasks
• multi-agent collaboration