[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-85583-en":3,"doc-seo-85583-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":13,"seo_description":14,"update_tm":28,"read_time":29},85583,1649267921044,"Ava Thompson","https://us-avatar.wpscdn.com/avatar/1800007509477c92dfb?_k=1782875107921204101",8,"Research & Report","Recursive Multi-Agent Systems","RecursiveMAS extends recursive or looped language modeling from single-model scaling to multi-agent collaboration. It frames the whole multi-agent system as a unified latent-space recursive computation, linking heterogeneous agents through the lightweight RecursiveLink to generate in-distribution latent thoughts and transfer latent states across agents. An inner-outer loop learning algorithm co-optimizes the entire system with shared gradient-based credit assignment. Runtime complexity and learning dynamics analyses show higher efficiency and stable gradients versus standard MAS, with consistent gains across benchmarks.","arXiv :2604 .259 17v2 [ cs .AI] 13 Jul 2026  \nRecursive Multi-Agent Systems  \nJiaru Zou1 †, Rui Pan2 , Ruizhong Qiu2 , Pan Lu1 , Shizhe Diao3 , Jindong Jiang3 , Hanghang Tong2 , Tong Zhang2 , Markus J. Buehler4 , Jingrui He2 B, James Zou1 B  \n1 Stanford University 2University of Illinois Urbana-Champaign 3NVIDIA 4 MIT †Project Lead B Corresponding Authors  \n Project Page: [https://recursivemas.github.io](https://recursivemas.github.io)  \nRecursive or looped language models have recently emerged as a new scaling axis by iteratively refining the same model computation over latent states to deepen reasoning. We extend such scaling principle from a single model to multi-agent systems, and ask: Can agent collaboration itself be scaled through recursion? To this end, we introduce RecursiveMAS, a recursive multi-agent framework that casts the entire system as a unified latent-space recursive computation. RecursiveMAS connects heterogeneous agents as a collaboration loop through the lightweight RecursiveLink module, enabling in-distribution latent thoughts generation and cross-agent latent state transfer. To optimize our framework, we develop an inner-outer loop learning algorithm for iterative whole-system co-optimization through shared gradient-based credit assignment across recursion rounds. Theoretical analyses of runtime complexity and learning dynamics establish that RecursiveMAS is more efficient than standard text-based MAS and maintains stable gradients during recursive training. Empirically, we instantiate RecursiveMAS under 4 representative agent collaboration patterns and evaluate across 9 benchmarks spanning mathematics, science, medicine, search, and code generation. In comparison with advanced single/multi-agent and recursive computation baselines, RecursiveMAS consistently delivers an average accuracy improvement of 8.3%, together with 1.2×–2.4× end-to-end inference speedup, and 34.6%–75.6% token usage reduction.  \n RecursiveMAS Scaling Law  \nCollaboration Patterns  \nFigure 1 | Performance Landscape of RecursiveMAS across Training/Inference Recursion Depths (Top): The lightweight RecursiveMAS with sub-1.5B agents shows a clean scaling trend as recursion deepens. Generalization across Common Collaboration Patterns (Bottom): The Scaled RecursiveMAS with stronger LLM agents (5-10B) seamlessly adapts to diverse multi-agent system structures.  \n[Contact: jiaru@stanford.edu](Contact: jiaru@stanford.edu)  \n1. Introduction  \nTo tackle complex tasks, a single language model often falls short due to limited capacity, myopic generation, or inefficient exploration of the solution space (Li et al., 2025b; Shojaee et al., 2025; Song et al., 2026) . Once intelligence reaches a threshold, a natural direction is to treat individual models as specialized agents and organize them as a collaborative system (Tran et al., 2025; Xu et al., 2025) . A multi-agent system (MAS) (Wang et al., 2025b; Wu et al., 2024) can scale performance by enabling individuals to work together and contribute complementary strengths. Consider a set of heterogeneous agents, each assigned a distinct role and expertise. The system can either arrange agents into a sequential pipeline (Gu et al., 2025; Qian et al., 2024) to progressively decompose and solve a problem, or engage and integrate multiple domain-specialized agents (Babu et al., 2025; Qian et al., 2025; Ye et al., 2025b) for the task.  \nWhile MAS establishes a structural foundation, the next question is how to enable the system to evolve over time and adapt to different scenarios. Prior work has explored prompt-based adaptation (Shenet al., 2025; Zhang et al., 2025b; Zhou et al., 2025), where model interactions are improved through the iterative refinement of shared context. Although these updated prompts can help agents generate more aligned responses to the question, each agent itself cannot improve. A more principled line of work is to optimize agents through learning (Motwani et al., 2024; Subramaniam e","cbCaikx7KeIOWCTm","https://ap.wps.com/l/cbCaikx7KeIOWCTm","pdf",13240657,2,1,36,"English","en",105,"# Introduction\n## Recursive Multi-Agent Collaboration via RecursiveMAS\n## RecursiveLink Module\n## Inner-Outer Loop Training Paradigm\n## Scaling, Patterns, and Benchmark Evaluation","[{\"question\":\"What is the core idea behind RecursiveMAS?\",\"answer\":\"RecursiveMAS treats a multi-agent collaboration as a unified recursive computation in latent space, extending looped language modeling from a single model to an agent system.\"},{\"question\":\"How do agents communicate in RecursiveMAS?\",\"answer\":\"Agents are connected through the lightweight RecursiveLink module, which supports latent thought generation within agents and cross-agent latent state transfer.\"},{\"question\":\"How is RecursiveMAS trained and optimized?\",\"answer\":\"RecursiveMAS uses an inner-outer loop learning algorithm to iteratively co-optimize the whole system, using shared gradient-based credit assignment across recursion rounds.\"}]",1784204739,91,{"code":4,"msg":31,"data":32},"ok",{"site_id":25,"language":24,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"recursive-multi-agent-systems","",{"@graph":36,"@context":85},[37,53,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,47,50],{"item":41,"name":42,"@type":43,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":20},"https://docshare.wps.com/document/","Document",{"item":48,"name":12,"@type":43,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":43,"position":52},"https://docshare.wps.com/document/recursive-multi-agent-systems/85583/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":24,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-25","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What is the core idea behind RecursiveMAS?","Question",{"text":75,"@type":76},"RecursiveMAS treats a multi-agent collaboration as a unified recursive computation in latent space, extending looped language modeling from a single model to an agent system.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How do agents communicate in RecursiveMAS?",{"text":80,"@type":76},"Agents are connected through the lightweight RecursiveLink module, which supports latent thought generation within agents and cross-agent latent state transfer.",{"name":82,"@type":73,"acceptedAnswer":83},"How is RecursiveMAS trained and optimized?",{"text":84,"@type":76},"RecursiveMAS uses an inner-outer loop learning algorithm to iteratively co-optimize the whole system, using shared gradient-based credit assignment across recursion rounds.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":20,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]