At 2.5 mg/mL a 0.25 mg draw is 10 units on a U-100 barrel and a 2.4 mg draw is 96. Those two numbers decide it, because the concentration is only sensible relative to the smallest and largest volumes you will actually measure with it. Both land on a readable part of a U-100 barrel, which is the entire point of choosing the diluent volume deliberately rather than pouring in a round number. The other consideration is time: a vial you will finish in a fortnight can be concentrated, and a vial you will draw from for months should be split at reconstitution instead.
This is the one decision in the whole preparation sequence that cannot be revised later, which is why it is worth thirty seconds of arithmetic.
The other direction: 10 mg in 3 mL is 3.33 mg/mL, and a 0.5 mg dose becomes 0.15 mL, or 15 units. More barrel, easier reading, and a larger fraction of the vial volume lost to dead space across the same number of draws.
Vial headspace is the hard constraint. A nominal 2 mL vial typically holds a little over 2 mL to the shoulder; adding 3 mL is not an option and attempting it wastes the lot.
Nothing here is medical advice, and research-use material is not approved for human use.
Check the vial can physically hold the volume before you draw it up.