Video manipulation is the method to edit the video for various purposes. The deepfake is the manipulated media created using modern deep learning techniques which are the new AI-based methods for video manipulation, Deepfakes are generated for various good and bad purposes. The bad use of Deepfakes has been rising prominently. Researchers worldwide have been developing methods for Deepfake creation and Deepfake detection. The rate of Deepfake Creation has been rising owing to the opportunity for adversaries for financial fraud. This work is toward the detection of Deepfakes using various AI-based methods. The AI-based methods are majorly based on Deep learning models which are trained for Deepfake Detection through the form of AI Inferencing. Over the last decade, many AI models have been designed for various real-world computer vision problems starting from AlexNet followed by new popular models for instance VGGNet, ResNet, InceptionNet, and many more. EfficientNets and Transformers are the recent state of the art in Deep learning models. For the purpose of Deepfake Detection, we have used these newer architectures. We have trained the model for Deepfake Detection using the publicly available Dataset of DFDC Faceforensics++ and CelebDFV2. We have achieved comparable results of 84% accuracy on the popular Deepfake Detection challenge test set.