That has to be about the biggest effect size on top of what I assume to be standard therapy in diabetes ever.
Yes/no. It is big, but not completely out of family since diabetes treatment's Hba1c effect is strongly a function of initial HBA1c. And the starting hba1c for the trial that produced a drop of 3.4 was extremely high compared to most trials. E.g. I (correctly) predicted that it would be only a drop of 1 to 2 percent in a more typical population which a starting hba1c of mid 8's.
Note: it is still a good treatment paradigm, clearly producing a meaningfully better hba1c drop than straight drug protocol. Just not what 3.4 sounds like.