Easy, the first one you can buy in quantities as small as 1. The second one you have to buy at least 160 of 'em!
Actually, that works against it, the one being
half the cost of the one you have to special-order 160 of.
More likely they were never very good at making the costlier one in the first place, or it's on NRND and so the price is jacking up to reflect the shrinking supply and reluctance to produce more.
But quantity, and supplier markup, is, in general, the biggest part of the cost and variance you will see at small quantities.
You don't really have representative pricing until you get into the thousands for power devices, or millions for small chips (resistors etc.). Basically, whole pallets at a time. Even then, prices are negotiable in those quantities. At all levels, from manufacturer to distributor to broker.
Quantity may vary; TI doesn't give a shit about a thousand here or there, but ADI does. I once met with an ADI FAE and got a nice price on a quad ADC that undercut the competing NS/TI dual ADC (per part in fact, making it a good deal indeed), and that for quantity under 1k/yr. (The FAE then forwarded the negotiated price to the supplier, DK I think, for future purchasing. You won't get factory-direct unless you're using millions qty.)
Cost need not reflect capability. For semiconductors, one of the biggest costs is making the masks themselves. Once those are written off, the cost drops to mere production cost, of which the package itself is often dominant. For example, the cheapest TO-220 you may find is a fraction of a buck, and the cheapest TO-247 is a buck and a half, even with the same silicon inside both. And the amount of silicon inside both is a tiny square not much bigger than a classic 2N2222 die, say.
The SOT-227 I don't think gets much below $10, that's about your starting point for those parts.
As die sizes shrink and capabilities continue to improve, costs can remain competitive, even against classic parts that are very cheap on the global market.
Tim