tiny 25ga needle and successfully hand solder a .4mm QFN without the usual bridges.
Hand solder? Hmm, wonder if you could do that easier with the right iron tip. This whole manual paste thing has me confused. Some very experienced people use it, but it doesn't make sense to me, especially on the ICs. (Other than BGA's, which I have no experience with.)
I've seen some very impressive pasting on individual IC pads, but by the time they're done pasting just this one IC, I would have been done placing, soldering, and inspecting the chip. I find the pasting useful for the passives, in particular. If I'm hand-pasting, I'm usually hand soldering the IC's, first, and only pasting the passives, diodes, caps, trannys, etc. If those parts all had gull-wing leads, like an IC, I would never hand paste anything. You can touch a soldering iron tip to a point as fast and easy as you'll ever be able to hand apply solder paste.
Even for assembling UQFN leadless protos, I find it most time-efficient to use regular solder. I flux, tin all the pads and "leads," more flux, put the chip on top and then hit it with the rework nozzle. Until I have an X-ray inspection machine, this will probably remain my preferred method for protos.
FWIW, I find 18-20 gauge is good for all around use. And the main problem with the plastic needles is they take a lot of paste to prime, which doesn't all suck back into the syringe when done. So when you're done pasting, you lose a bit when you toss the tip and put the air tight cap back on. (I've tried plugging the needle with pins and/or hot glue, but the paste always finds a way to dry out). But they're dirt cheap in quantity and decrease drag.