I'm not entirely sure of what your question is asking. But if you know the gene you are interested in, then you should know its size from the sequence. And the size of GFP is also known, so you can simply calculate the predicted size based on the location where the fusion is constructed.
You can easily estimate the size of the combined cDNA (gene + GFP) using any offline/online in silico tools and by keeping into account the joining strategy like, GFP cloning at N-ter or C-Ter of the gene, if there is any linker region in between, and whether there are any targeting signals attached