In this paper, we are investigating the benefits of employing pre-trained models in computer vision as we work toward creating a model for rapid image style transfer. These models, like VGG or MobileNet, have undergone painstaking optimization for efficiency and precision in picture categorization tasks. We want to change photographs quickly while ensuring high-quality outcomes by utilizing their effective designs and transferable features. The training time needed to create a style transfer model from scratch is considerably reduced by the ability to reuse layers from existing models trained with high computational power on big datasets. We are actively adjusting the models to further tailor them to particular style transfer requirements and aesthetic preferences, allowing for more flexibility and creative control. The methodology we use is based on extracting layers of the pre-trained models with Imagenet dataset weights to create smaller style transfer-capable systems that can be tuned and applied in less than a minute on a particular pair of content and style images. The availability of pre-trained models within well-known deep-learning frameworks, such as TensorFlow and PyTorch, speeds up the implementation of style transfer capabilities and makes the development process easier. Although we are aware of the drawbacks of pre-trained models, we are committed to overcoming them and developing an effective style transfer solution.