Recursive Multi-Agent Systems
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914514626150400 |
|---|---|
| author | Yang, Xiyuan Zou, Jiaru Pan, Rui Qiu, Ruizhong Lu, Pan Diao, Shizhe Jiang, Jindong Tong, Hanghang Zhang, Tong Buehler, Markus J. He, Jingrui Zou, James |
| author_facet | Yang, Xiyuan Zou, Jiaru Pan, Rui Qiu, Ruizhong Lu, Pan Diao, Shizhe Jiang, Jindong Tong, Hanghang Zhang, Tong Buehler, Markus J. He, Jingrui Zou, James |
| contents | Recursive 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$\times$-2.4$\times$ end-to-end inference speedup, and 34.6%-75.6% token usage reduction. Code and Data are provided in https://recursivemas.github.io. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_25917 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Recursive Multi-Agent Systems Yang, Xiyuan Zou, Jiaru Pan, Rui Qiu, Ruizhong Lu, Pan Diao, Shizhe Jiang, Jindong Tong, Hanghang Zhang, Tong Buehler, Markus J. He, Jingrui Zou, James Artificial Intelligence Computation and Language Machine Learning Recursive 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$\times$-2.4$\times$ end-to-end inference speedup, and 34.6%-75.6% token usage reduction. Code and Data are provided in https://recursivemas.github.io. |
| title | Recursive Multi-Agent Systems |
| topic | Artificial Intelligence Computation and Language Machine Learning |
| url | https://arxiv.org/abs/2604.25917 |