It sounds like you aren't aware of the scale of what we know about what has been lost. Even if you're just looking at the Greco-Roman context, Diophantos is a drop in a veritable ocean of known lost works. Everyone's going to have their own preference. A historian of mathematics and astronomy might well put the highest price on Diophantos, or lost works of Babylonian astronomy (my own pick would be Hipparchos); a literary historian might go for lost tragedians, or Antimachos, or the missing bits of Gilgamesh, or Ennius' Annals; for political and military history there are literally thousands to choose from (my own pick, again thinking mainly of the Greco-Roman context, would be the works of Ephoros, Timaios, and Quintus Fabius Pictor).

But if you started thinking about this just because of Diophantos, then I think the core of the answer has to be that we know of enormously many more lost works than you expected. Like, thousands and thousands of them. And those are just the ones we know of. Any of them would be a treasure if an intact copy were found.