The past two years have witnessed the evolution of large language model (LLM)-based multi-agent systems from labor-intensive manual design to partial automation (\textit{e.g.}, prompt engineering, communication topology) and eventually to fully automated design. However, existing agentic automation pipelines often lack LLM heterogeneity and focus on single-objective performance optimization, limiting their potential to combine weaker models for more customized and cost-effective solutions. To address this challenge, we propose EvoFlow, a niching evolutionary algorithm-based framework to automatically search a population of heterogeneous and complexity-adaptive agentic workflows, rather than a single homogeneous, complex workflow. Technically, EvoFlow performs \textit{(1) tag-based retrieval} to extract parent workflows from an agentic population, evolves new workflows through \textit{(2) crossover} and \textit{(3) mutation}, and employs \textit{(4) niching-based selection} to maintain population diversity and quality. Extensive evaluations across seven benchmarks demonstrate that EvoFlow is: \textbf{(I) diverse}, evolving a population of workflows ranging from simple I/O tasks to complex multi-turn interactions; \textbf{(II) high-performing}, outperforming previous handcrafted and automated workflows by ; \textbf{(III) economical}, surpassing powerful \llmname{o1-preview} at of its inference cost using weaker open-source models.
View on arXiv@article{zhang2025_2502.07373,
title={ EvoFlow: Evolving Diverse Agentic Workflows On The Fly },
author={ Guibin Zhang and Kaijie Chen and Guancheng Wan and Heng Chang and Hong Cheng and Kun Wang and Shuyue Hu and Lei Bai },
journal={arXiv preprint arXiv:2502.07373},
year={ 2025 }
}