How does the Simulink Extended Kalman Filter calculate the Jacobian numerically?

I've been given a Simulink-Model which utilizes the Simulink Extended Kalman Filter Block to estimate the State of my model. Thus I also have access to the State Transition and Measurement function. I then have been tasked to do my own implementation of the EKF without using the Simulink EKF Block. I struggle a bit with the derivation of the right Jacobians for my implementation but the Simulink EKF somehow doesn't require the Jacobians to be specified and can calculate them numerically (as stated in https://de.mathworks.com/help/control/ug/extended-and-unscented-kalman-filter-algorithms-for-online-state-estimation.html).

But how does it do that exactly? Is some kind of numerical differentiation done here?