Function calling in GPTs vs Assistants

Will there be any difference in how function calling works in GPTs versus its current operation in Assistants? While the retrieval and code interpreter features are useful in Assistants and are expected to be beneficial in GPTs, the advantages of function calling in GPTs are unclear to me if you cannot integrate an actual call.

Perhaps this is designed to function similarly to plugins, or am I misunderstanding something?


Any ideas? I think this is the key question and may greatly affect usability of Assistants/GPTs.