Jyoti Amit Kumar Dhamecha,
Tulsi Ashokbhai Javandhra,
- Assistant Professor, Department of Computer Science and Engineering, Sardar Patel College of Administration and Management (SPEC Campus), Gujarat, India
- Student, Department of Computer Science and Engineering, Sardar Patel College of Administration and Management (SPEC Campus), Gujarat, India
Abstract
The rapid growth of agentic AI has led to a diverse ecosystem of open-source frameworks for developing LLM-based agents. These frameworks differ in their approaches to state management, multi-agent coordination, tool orchestration, memory, and workflow execution. This paper presents a comparative analysis of prominent open-source AI agent frameworks, including LangChain, LangGraph, CrewAI, AutoGen, Semantic Kernel, and related systems. The study synthesizes a common architectural model to identify the core components and design patterns shared across these frameworks while highlighting their architectural differences. It further examines reported performance characteristics using findings from 2026 benchmark studies, with particular attention to latency, token efficiency, execution overhead, and orchestration costs across standardized tasks. The analysis discusses how architectural choices influence framework performance, scalability, flexibility, and suitability for different agentic AI applications. In addition, the paper examines practical design trade-offs involving centralized versus distributed orchestration, state persistence, agent communication, and tool integration. Finally, it identifies key challenges that remain in the development and evaluation of agentic AI frameworks, including inconsistencies in benchmarking methodologies, limited cross-framework interoperability, reproducibility concerns, production governance, and the absence of widely accepted evaluation standards. The study provides a structured foundation for understanding the evolving landscape of open-source agent frameworks and identifies directions for future research toward interoperable, efficient, and reliable agentic AI systems.
Keywords: AI agents, open-source software, LangChain, AutoGen, CrewAI, multi-agent systems, benchmarking.
[This article belongs to Journal of Open Source Developments ]
References
- Yao S, Zhao J, Yu D, Du N, Shafran I, Narasimhan K, et al. ReAct: synergizing reasoning and acting in language models [Preprint]. 2022. arXiv:2210.03629.
- Wang L, Ma C, Feng X, Zhang Z, Yang H, Zhang J, et al. A survey on large language model based autonomous agents. Front Comput Sci. 2024;18(6):186345. doi:10.1007/s11704-024-40231-1.
- Xi Z, Chen W, Guo X, He W, Ding Y, Hong B, et al. The rise and potential of large language model based agents: a survey. Sci China Inf Sci. 2025;68(2):121101. doi:10.1007/s11432-024-4222-0.
- (2024). LangChain documentation [Online]. LangChain. Available from: https://python.langchain.com/docs/introduction/.
- (2026). CrewAI documentation: agents, crews and flows [Online]. CrewAI. Available from: https://docs.crewai.com/.
- Wu Q, Bansal G, Zhang J, Wu Y, Li B, Zhu E, et al. AutoGen: enabling next-gen LLM applications via multi-agent conversation [Preprint]. 2023. arXiv:2308.08155.
- (2026). Semantic Kernel Agent Architecture [Online]. Microsoft Learn. Available from: https://learn.microsoft.com/en-us/semantic-kernel/frameworks/agent/agent-architecture
- Derouiche H, Brahmi Z, Mazeni H. Agentic AI frameworks: architectures, protocols, and design challenges [Preprint]. 2025. arXiv:2508.10146.
- Orogat A, Rostam A, Mansour E. Understanding multi-agent LLM frameworks: a unified benchmark and experimental analysis [Preprint]. 2026. arXiv:2602.03128.
- Liu D, Upadhyay K, Chhetri V, Siddique AB, Farooq U. A large-scale study on the development and issues of multi-agent AI systems. 2025 IEEE International Conference on Big Data (BigData), Macau, China. 2025. p.7785–92. doi:10.1109/BigData66926.2025.11402104.
- Ahmad MSJ, Marathe K. Benchmarking multi-agent frameworks: LangGraph vs. CrewAI vs. AutoGen. J Acad Trends Innov Res. 2026;2:719–731.
- Chowa SS, Alvi R, Rahman SS, Rahman MA, Raiaan MAK, Islam MR, et al. From language to action: a review of large language models as autonomous agents and tool users. Artif Intell Rev. 2026;59(2):71. doi:10.1007/s10462-025-11471-9.
- Du S, Zhao J, Shi J, Xie Z, Jiang X, Bai Y, et al. A survey on the optimization of large language model-based agents. ACM Comput Surv. 2026;58(9):1–37. doi:10.1145/3789261.
- Dennis S, Patil R, Shabahang K, Guo H. Compiling agentic workflows into LLM weights: near-frontier quality at two orders of magnitude less cost [Preprint]. 2026. arXiv:2605.22502.
- Gupta A, Goyal M. A comprehensive review of Microsoft Semantic Kernel: Architecture, capabilities, and research directions. 2025 IEEE DELCON – International Conference on Recent Smart Technologies in Engineering for Sustainable Development, New Delhi, India. 2025. p. 1–6. doi:10.1109/DELCON68055.2025.11400282.
Journal of Open Source Developments
| Volume | 13 | |
| Issue | 02 | |
| Received | 29/08/2026 | |
| Accepted | 26/09/2026 | |
| Published | 28/09/2026 | |
| Publication Time | 30 Days |