Small Language Models for Real-World Task Automation in Resource-Constrained Environments

• Hugo R. B. Gonçalves UFMS
• Pedro B. G. Neto UFMS
• Guilherme Shinohara UFMS
• Jonathan de Andrade Silva UFMS
• Brivaldo A. da Silva Jr UFMS

DOI: https://doi.org/10.5753/eniac.2026.28957

Resumo

This work investigates the viability of Small Language Models (SLMs) as cost-effective alternatives to Large Language Models (LLMs) for automation tasks in resource-constrained environments. We implement AI agents using the low-code n8n platform and lightweight, self-hosted models served through the Ollama engine, thereby reducing operational costs and computational overhead. We evaluate the semantic similarity between the outputs of these SLMs and those of high-capacity LLMs, including Gemini 3 Flash and Claude Sonnet 4, using BERTScore. The results show that the best-performing SLM achieved F1 scores close to 80%, indicating strong semantic alignment despite limited computational resources and network access.

Referências

Australian Institute for Machine Learning (AIML) (2024). Beyond Bigger Models: The Evolution of Language Model Scaling Laws. [Online; accessed 25-05-2026].

Barcelar, R., Garcia, L., Heleno, A., Ventura, T., and Oliveira, A. (2025). Knowledge Distillation in Compact Models: An Approach Applied to Text Processing for Public Security. In Anais do XVI Simpósio Brasileiro de Tecnologia da Informação e da Linguagem Humana, pages 42–51, Porto Alegre, RS, Brasil. SBC.

Chen, Y., Han, Y., and Li, X. (2025). FASTNav: Fine-Tuned Adaptive Small-Language-Models Trained for Multi-Point Robot Navigation. IEEE Robotics and Automation Letters, 10(1):390–397.

Fernandes, R., Biedenkapp, A., Hutter, F., and Awad, N. (2025). A Llama Walks into the ‘Bar’: Efficient Supervised Fine-Tuning for Legal Reasoning in the Multi-state Bar Exam. [Online; accessed 25-05-2026].

Li, Y., Bubeck, S., Eldan, R., Giorno, A. D., Gunasekar, S., and Lee, Y. T. (2023). Textbooks Are All You Need II: phi-1.5 Technical Report. [Online; accessed 25-05-2026].

Licardo, J. T. and Tankovic, N. (2025). Performance Trade-offs of Optimizing Small Language Models for E-Commerce. [Online; accessed 25-05-2026].

n8n-io (2025). n8n: Secure Workflow Automation for Technical Teams. GitHub. [Online; accessed 25-05-2026].

Nguyen, C. V. et al. (2024). A Survey of Small Language Models. [Online; accessed 22-08-2026].

Rang, M., Bi, Z., Zhou, H., Chen, H., Xiao, A., Guo, T., Han, K., Chen, X., and Wang, Y. (2025). Revealing the Power of Post-Training for Small Language Models via Knowledge Distillation. [Online; accessed 25-05-2026].

Russell, S. and Norvig, P. (2010). Artificial Intelligence: A Modern Approach. Prentice Hall.

Sakib, T. H., Hosain, M. T., and Morol, M. K. (2025). Small Language Models: Architectures, Techniques, Evaluation, Problems and Future Adaptation. [Online; accessed 25-05-2026].

Sharma, A., Ene, T.-D., Kunal, K., Liu, M., Hasan, Z., and Ren, H. (2024). Assessing Economic Viability: A Comparative Analysis of Total Cost of Ownership for Domain-Adapted Large Language Models versus State-of-the-Art Counterparts in Chip Design Coding Assistance. [Online; accessed 25-05-2026].

Vakili, Y. Z., Fallah, A., and Sajedi, H. (2024). Distilled BERT Model in Natural Language Processing. In 2024 14th International Conference on Computer and Knowledge Engineering (ICCKE), pages 243–250.

Wang, F., Zhang, Z., Zhang, X., Wu, Z., Mo, T., Lu, Q., Wang, W., Li, R., Xu, J., Tang, X., He, Q., Ma, Y., Huang, M., and Wang, S. (2025). A Comprehensive Survey of Small Language Models in the Era of Large Language Models: Techniques, Enhancements, Applications, Collaboration with LLMs, and Trustworthiness. ACM Trans. Intell. Syst. Technol., 16(6).

Wooldridge, M. J. (2001). An Introduction to Multiagent Systems. John Wiley & Sons, Ltd.