How to combine 3 pytorch model?

I have 2 dataset.
1 dataset is preprocess and the other is actual.
The preprocess data will go through model 1 and being classify as one of 2 classes.
for example the class name for model 1 is A and B.
If image 1 classify as Class A it will go through model 2.
However I want only image successful being classify as class A go through model 2.
Class B will go through model 3 .
In summary input( image) feed to model 1 and 3 is preprocess.
Input for model 2 is the actual one.
Can someone suggest or had any idea to do it?

It’s easy to get answers by following your thoughts and continuing to write…

I’m not fully understanding the use case.
How would you know, if model1 correctly predicted classA during the validation and test phase, where no labels are available?