Autonomous Agents Based on Large Language Models: Reasoning Strategies and Multi-Agent Interaction
DOI:
https://doi.org/10.63808/tcs.v2i2.519Keywords:
Large language model, Autonomous agent, Reasoning strategy, Multi-agent interaction, Cost-aware evaluationAbstract
Language-model agents acquire capability along two routes that published work almost always varies together, one enlarging the computation a single agent performs and the other enlarging the number of agents and the channels along which they pass partial results, which leaves the contribution of each route indistinguishable. This paper separates an agent system into a reasoning operator and an interaction topology, holding the number of agents fixed at three across all multi-agent topologies, together with the number of rounds, the message cap, the prompt skeleton, and the memory settings for all combinations, and evaluates a full factorial design of four reasoning levels and four topologies on three public test sets totalling 484 instances under five random seeds, giving 38,720 task executions. Comparisons are paired at the instance level, and effects are estimated from logistic mixed-effects models fitted to each test set separately. The reasoning factor spans 18.0 points on the static suite whereas the topology factor spans 7.8 points on the interactive one, a reversal indicating that the two mechanisms act on different failure modes. The topology span is taken across all four topology levels and therefore includes the step from one agent to three, against a structure-only span of 2.8 points among the multi-agent topologies. The two factors also depart from additivity in a localised manner. A mechanism-motivated single-degree-of-freedom contrast on the only cell in which both factors contribute a corrective pass is significant at p = 0.001 on the static suite, whereas the omnibus nine-degree-of-freedom interaction test is not (p = 0.094), and in that cell the observed 50.0 percent lies 5.8 points below the additive expectation. After tokens are counted, seven of the sixteen configurations remain undominated on the static suite and five on the operating-system suite, and returns saturate at 12,400 tokens per instance on the static suite under the present design.
References
[1] Besta, M., Blach, N., Kubicek, A., Gerstenberger, R., Podstawski, M., Gianinazzi, L., Gajda, J., Lehmann, T., Niewiadomski, H., Nyczyk, P., & Hoefler, T. (2024). Graph of Thoughts: Solving elaborate problems with large language models. Proceedings of the AAAI Conference on Artificial Intelligence, 38(16), 17682–17690. https://doi.org/10.1609/aaai.v38i16.29720
[2] Du, Y., Li, S., Torralba, A., Tenenbaum, J. B., & Mordatch, I. (2024). Improving factuality and reasoning in language models through multiagent debate. In Proceedings of the 41st International Conference on Machine Learning (Vol. 235, pp. 11733–11763). PMLR. https://proceedings.mlr.press/v235/du24e.html
[3] Hong, S., Zhuge, M., Chen, J., Zheng, X., Cheng, Y., Zhang, C., Wang, J., Wang, Z., Yau, S. K. S., Lin, Z., Zhou, L., Ran, C., Xiao, L., Wu, C., & Schmidhuber, J. (2024). MetaGPT: Meta programming for a multi-agent collaborative framework. In Proceedings of the 12th International Conference on Learning Representations. OpenReview. https://openreview.net/forum?id=VtmBAGCN7o
[4] Li, G., Hammoud, H., Itani, H., Khizbullin, D., & Ghanem, B. (2023). CAMEL: Communicative agents for “mind” exploration of large language model society. Advances in Neural Information Processing Systems, 36, 51991–52008. https://proceedings.neurips.cc/paper_files/paper/2023/hash/a3621ee907def47c1b952ade25c67698-Abstract-Conference.html
[5] Liu, X., Yu, H., Zhang, H., Xu, Y., Lei, X., Lai, H., Gu, Y., Ding, H., Men, K., Yang, K., Zhang, S., Deng, X., Zeng, A., Du, Z., Zhang, C., Shen, S., Zhang, T., Su, Y., Sun, H., & Tang, J. (2024). AgentBench: Evaluating LLMs as agents. In Proceedings of the 12th International Conference on Learning Representations. OpenReview. https://openreview.net/forum?id=zAdUB0aCTQ
[6] Park, J. S., O’Brien, J., Cai, C. J., Morris, M. R., Liang, P., & Bernstein, M. S. (2023). Generative agents: Interactive simulacra of human behavior. In Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology (Article 2). Association for Computing Machinery. https://doi.org/10.1145/3586183.3606763
[7] Qian, C., Liu, W., Liu, H., Chen, N., Dang, Y., Li, J., Yang, C., Chen, W., Su, Y., Cong, X., Xu, J., Li, D., Liu, Z., & Sun, M. (2024). ChatDev: Communicative agents for software development. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (pp. 15174–15186). Association for Computational Linguistics. https://doi.org/10.18653/v1/2024.acl-long.810
[8] Schick, T., Dwivedi-Yu, J., Dessì, R., Raileanu, R., Lomeli, M., Hambro, E., Zettlemoyer, L., Cancedda, N., & Scialom, T. (2023). Toolformer: Language models can teach themselves to use tools. Advances in Neural Information Processing Systems, 36, 68539–68551. https://proceedings.neurips.cc/paper_files/paper/2023/hash/d842425e4bf79ba039352da0f658a906-Abstract-Conference.html
[9] Shinn, N., Cassano, F., Gopinath, A., Narasimhan, K., & Yao, S. (2023). Reflexion: Language agents with verbal reinforcement learning. Advances in Neural Information Processing Systems, 36, 8634–8652. https://proceedings.neurips.cc/paper_files/paper/2023/hash/1b44b878bb782e6954cd888628510e90-Abstract-Conference.html
[10] Srivastava, A., Rastogi, A., Rao, A., Shoeb, A. A. M., Abid, A., Fisch, A., Brown, A. R., Santoro, A., Gupta, A., Garriga-Alonso, A., Kluska, A., Lewkowycz, A., Agarwal, A., Power, A., Ray, A., Warstadt, A., Kocurek, A. W., Safaya, A., Tazarv, A., … Wu, Z. (2023). Beyond the imitation game: Quantifying and extrapolating the capabilities of language models. Transactions on Machine Learning Research. https://openreview.net/forum?id=uyTL5Bvosj
[11] Suzgun, M., Scales, N., Schärli, N., Gehrmann, S., Tay, Y., Chung, H. W., Chowdhery, A., Le, Q., Chi, E., Zhou, D., & Wei, J. (2023). Challenging BIG-Bench tasks and whether chain-of-thought can solve them. In Findings of the Association for Computational Linguistics: ACL 2023 (pp. 13003–13051). Association for Computational Linguistics. https://doi.org/10.18653/v1/2023.findings-acl.824
[12] Wang, G., Xie, Y., Jiang, Y., Mandlekar, A., Xiao, C., Zhu, Y., Fan, L., & Anandkumar, A. (2024). Voyager: An open-ended embodied agent with large language models. Transactions on Machine Learning Research. https://doi.org/10.48550/arXiv.2305.16291
[13] Wang, L., Ma, C., Feng, X., Zhang, Z., Yang, H., Zhang, J., Chen, Z., Tang, J., Chen, X., Lin, Y., Zhao, W. X., Wei, Z., & Wen, J.-R. (2024). A survey on large language model based autonomous agents. Frontiers of Computer Science, 18(6), Article 186345. https://doi.org/10.1007/s11704-024-40231-1
[14] Wang, X., Wei, J., Schuurmans, D., Le, Q. V., Chi, E. H., Narang, S., Chowdhery, A., & Zhou, D. (2023). Self-consistency improves chain of thought reasoning in language models. In Proceedings of the 11th International Conference on Learning Representations. OpenReview. https://openreview.net/forum?id=1PL1NIMMrw
[15] Wei, J., Wang, X., Schuurmans, D., Bosma, M., Ichter, B., Xia, F., Chi, E., Le, Q., & Zhou, D. (2022). Chain-of-thought prompting elicits reasoning in large language models. Advances in Neural Information Processing Systems, 35, 24824–24837. https://proceedings.neurips.cc/paper_files/paper/2022/hash/9d5609613524ecf4f15af0f7b31abca4-Abstract-Conference.html
[16] Wu, Q., Bansal, G., Zhang, J., Wu, Y., Li, B., Zhu, E., Jiang, L., Zhang, X., Zhang, S., Liu, J., Awadallah, A. H., White, R. W., Burger, D., & Wang, C. (2024). AutoGen: Enabling next-gen LLM applications via multi-agent conversations. In Proceedings of the First Conference on Language Modeling. OpenReview. https://openreview.net/forum?id=tEAF9LBdgu
[17] Xi, Z., Chen, W., Guo, X., He, W., Ding, Y., Hong, B., Zhang, M., Wang, J., Jin, S., Zhou, E., Zheng, R., Fan, X., Wang, X., Xiong, L., Zhou, Y., Wang, W., Jiang, C., Zou, Y., Liu, X., … Gui, T. (2025). The rise and potential of large language model based agents: A survey. Science China Information Sciences, 68(2), Article 121101. https://doi.org/10.1007/s11432-024-4222-0
[18] Yao, S., Chen, H., Yang, J., & Narasimhan, K. (2022). WebShop: Towards scalable real-world web interaction with grounded language agents. Advances in Neural Information Processing Systems, 35, 20744–20757. https://proceedings.neurips.cc/paper_files/paper/2022/hash/82ad13ec01f9fe44c01cb91814fd7b8c-Abstract-Conference.html
[19] Yao, S., Yu, D., Zhao, J., Shafran, I., Griffiths, T. L., Cao, Y., & Narasimhan, K. (2023). Tree of Thoughts: Deliberate problem solving with large language models. Advances in Neural Information Processing Systems, 36, 11809–11822. https://proceedings.neurips.cc/paper_files/paper/2023/hash/271db9922b8d1f4dd7aaef84ed5ac703-Abstract-Conference.html
[20] Yao, S., Zhao, J., Yu, D., Du, N., Shafran, I., Narasimhan, K., & Cao, Y. (2023). ReAct: Synergizing reasoning and acting in language models. In Proceedings of the 11th International Conference on Learning Representations. OpenReview. https://openreview.net/forum?id=WE_vluYUL-X
[21] Zhang, G., Yue, Y., Li, Z., Yun, S., Wan, G., Wang, K., Cheng, D., Yu, J. X., & Chen, T. (2025). Cut the crap: An economical communication pipeline for LLM-based multi-agent systems. In Proceedings of the 13th International Conference on Learning Representations. https://proceedings.iclr.cc/paper_files/paper/2025/hash/bbc461518c59a2a8d64e70e2c38c4a0e-Abstract-Conference.html
[22] Zhang, H., Cui, Z., Chen, J., Wang, X., Zhang, Q., Wang, Z., Wu, D., & Hu, S. (2025). Stop overvaluing multi-agent debate: We must rethink evaluation and embrace model heterogeneity (No. arXiv:2502.08788). arXiv. https://doi.org/10.48550/arXiv.2502.08788
[23] Zhuge, M., Wang, W., Kirsch, L., Faccio, F., Khizbullin, D., & Schmidhuber, J. (2024). GPTSwarm: Language agents as optimizable graphs. In Proceedings of the 41st International Conference on Machine Learning (Vol. 235, pp. 62743–62767). PMLR. https://proceedings.mlr.press/v235/zhuge24a.html
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Zhengyuan Wang, Yuyao Wang, Yixin Deng
This work is licensed under a Creative Commons Attribution 4.0 International License.