Take a look at this link to see what is available for GSEA tool by broad for GO terms.
Alternatively you have also take a look into this
The thing is still not clear about your motivation. Usually one does GO enrichment and that will have some sort of enrichment scores based upon the knowledge base and your input gene list. Then you can select the top GO terms based on some threshold of pvalue and enrichment association score. Why do you want to do again GSEA of GO terms. What you can do is taking a GO terms and the genes that you have in your list for that GO term and then perform classical GSEA for the specific molecular function of biological process. That will still give some kind of idea.