A container mode can be dynamically selected when an application is launched on an end user computing device. When an application is deployed to the end user computing device, a container configurator can collect information about the application and share it with a machine learning solution to receive an application score for the application. When the application is launched on the end user computing device, the container configurator can provide the application score, capabilities of the end user computing device, current resource utilization and admin preferences to the machine learning solution. The machine learning solution can then dynamically select a container mode based on this information and provide the selection to the container configurator. The container configurator can then cause the application to be launched within a container that matches the selected container mode.