The evolution of the Large Language model (LLM) has revolutionized the field of natural language processing with its unparalleled ability to generate text, images, code, audio, and video through prompts and the ability to comprehend the contents. Nonetheless, the effectiveness of these LLMs heavily relies on various factors such as the ability to handle huge volumes of context accurately, quality of training data, and relevance in the prompting has an impact on the generation of content appropriate to the context. Within the realm of software engineering, one of the most persistent challenges is the development of automated code to meet the requirements of the programmers. Recent developments of LLM’s powered code generation tools caught the attention of researchers. Although models like LLaMA, Claude, GPT, and Gemini can facilitate code development, access to their advanced iterations is frequently restricted, and their computing requirements are substantial. This study explores effective prompting techniques that maximize code generation accuracy in Python, minimize hallucination, and optimize outputs. Emphasis is placed on choosing between prompting and fine-tuning to achieve the most coherent, effective responses for practical coding tasks.