To put all the feature classes together in one feature class use the append tool in the same model as the iterator.
Please can you share a snapshot or your actual model to suggest more/alternative approach?
Before the append tool is used, you can look into the Collect Values tool. What this tool does is creates a list of inputs that can then be used as inputs to a successive tool. For example, run your iteration to select for whatever features you need to, and have that model end with the Collect Values tool. This should create a list of inputs that are maintained in memory. You can integrate this model in another and use the append or merge tool in that final model to create a final feature class with all selections.
You can use the Collect Values tool in the same model as the iterator and then this model in another model as the link shared by j.quinn shows.
Or you can use the Append tool directly in the model with the iterator without the Collect Values tool. Both Append and Merge require a multivalue input, but with Append tool you are only appending the table to the target dataset and therefore you can run it as many times as the iterator runs . On each run you will only get one feature class/table being appended to the target thus eliminating the use of Clollect Value tool.
Only you have to do some extra work in creating a target dataset which has the schema containing all possible fields that your feature classes could have.