This paper presents the development and evaluation of a Text-to-Image AI Generator web application, leveraging advanced AI techniques, with a particular focus on a custom-built Stable Diffusion model, to transform textual prompts into realistic images. Aimed at enhancing accessibility in image generation. This application is designed with a user-friendly, responsive interface, enabling users with minimal technical knowledge to create high-quality images by simply entering descriptive text. Unlike existing solutions that rely on pre-trained models, our backend system utilizes a Stable Diffusion model implemented from scratch, providing greater flexibility and opportunities for optimization. This custom implementation allows for a deeper understanding of the underlying mechanisms and enables tailored enhancements. The application workflow is streamlined: users enter a text prompt, initiate the generation process, and have the option to download the resulting image. The paper includes sample prompts demonstrating diverse input–output capabilities, providing insights into the model’s interpretative range and generation accuracy. We detail the architectural choices, training process, and key modifications implemented to improve performance and efficiency. The study underscores the potential of AI-driven creative tools in transforming content generation workflows across fields such as digital art, media, and education.