Malignant lung tumors are the major cause of mortality and morbidity associated with cancer around the globe. Recently, there has been an increase in the frequency of lung cancer. Lung cancer must be diagnosed histopathologically in order for the patient to receive effective treatment. The present research focuses on the determination of histological images to demonstrate the application of the deep learning approach in the categorization of squamous-cell carcinoma and adenocarcinoma within the lung and colon. We proposed a multi-model architecture where the features of ResNet50V2 and VGG19 are merged by merging their properties. This multi-model is then applied using a number of fully connected layers, which are then modified for a specific classification objective. It is suitable for jobs where good classification performance demands the extraction and integration of many different kinds of complex visual data. To validate the suggested approach, the LC25000 dataset was employed. In the research we conducted on the LC25000 dataset, the proposed technique outperformed previously used convolutional neural network methods with regard to accuracy (99.00%), precision (99.3%), recall (99.3%), and FI_score (99.2%).