This paper presents a study on the application of compact Large Language Models (LLMs) in summarizing Chinese medical dialogues, with a strong emphasis on developing a solution that is not only cost-effective but also highly practical. The research conducted a thorough comparison between fine-tuned models possessing less than 2 B parameters and existing proprietary and open-source models available in the market, ultimately identifying Qwen1.5-1.8B as the optimal small model. Following fine-tuning, this model exhibited exceptional performance in the summarization of medical dialogues, matching the capabilities of some of the most advanced proprietary Chinese models and striking a favorable balance between deployment cost and the quality of summarization. The fine-tuning dataset was derived from genuine doctor-patient conversations on a medical website, which not only endowed the model with substantial practical value but also ensured its applicability in real-world scenarios. By open-sourcing the fine-tuned Qwen1.5-1.8B model, this study not only offers valuable insights into the cost-effective deployment of LLMs for medical information processing but also stimulates further research and innovation in this domain. Interestingly, despite having only 0.2 B parameters, the ChatLM-mini-Chinese model achieved the highest scores in the R-1 and R-2 metrics among all the evaluated models, showcasing its significant research potential https://www.modelscope.cn/models/tdtrgxhg/Chinese-medi cal-dialogue-summarization-Model .