EVOCHAMBER: Test-Time Co-evolution of Multi-Agent System at Individual, Team, and Population Scales

Yaolun Zhang, Tianyi Xu, Shengyu Dai, Zhenwen Shao, Qingyun Wu, Huazheng Wang

arXiv:2605.11136 · 2026-05-13 공개 · arXiv · PDF

multi-agent-systems qwen3-8b test-time-evolution asymmetric-transfer coevolution team-formation population-dynamics competition-math

Abstract

We argue that multi-agent test-time evolution is not single-agent evolution replicated N times. A single-agent learner can only evolve its own context and memory. A multi-agent system additionally evolves who collaborates, how they collaborate, and how knowledge flows across the population. These components have no single-agent counterpart and can produce phenomena such as emergent specialization. Yet prior test-time methods either confine experiences to individual agents, forfeiting cross-agent learning, or broadcast symmetrically to all agents, erasing the specialization that makes collaboration valuable. We present EVOCHAMBER, a training-free framework that instantiates test-time evolution at three levels over a coevolving agent pool. At its core is CODREAM (Collaborative Dreaming), a post-task protocol triggered on team failure or disagreement, in which agents collaboratively reflect, distill insights, and route them asymmetrically from strong to weak agents on the failed niche, preserving specialization while filling knowledge gaps. Team-level operators assemble niche-conditioned teams and select collaboration structures online. Population-level lifecycle operators fork, merge, prune, and seed agents under performance pressure. On three heterogeneous task streams with Qwen3-8B, EVOCHAMBER reaches 63.9% on competition math, 75.7% on code, and 87.1% on multi-domain reasoning, outperforming the best baseline by 32% relative on math and confirming asymmetric cross-agent transfer as the primary driver in ablation. Starting from several identically initialized agents, four to five stable niche specialists spontaneously emerge, a structural signature of multi-agent evolution that no single-agent learner can express. See our code at: https://github.com/Mercury7353/EvoChamber

한국어 요약

📋 한 줄 요약

**[멀티에이전트 시스템 / 테스트 타임 진화]** 개인·팀·집단 세 수준에서 테스트 타임에 공진화(co-evolution)하는 학습-프리 멀티에이전트 프레임워크 EVOCHAMBER 제안.

🎯 핵심 기여도

💡 핵심 아이디어

멀티에이전트 진화의 핵심은 "누가 협업하고, 어떻게 협업하며, 지식이 어떻게 흐르는가"이다. 모두에게 동일하게 broadcast하면 specialization이 소실되고, 개별 에이전트에 격리하면 cross-agent learning을 잃는다. 비대칭 라우팅이 specialization을 보존하면서 약한 에이전트의 갭을 채운다.

🔬 기술적 접근법

📊 주요 결과

💭 의의 및 한계

**의의**: 단일 모델 self-improvement 한계를 넘어 다중 에이전트 공진화 패러다임의 실증 가능성을 제시. **한계**: 학습-프리이지만 multi-agent 호출 비용이 크며, 평가가 reasoning·math·code에 집중되어 도구 사용·임바디드 환경 일반화는 후속 과제.

🚀 실용적 활용