Our (Lavis lab) advice is to prepare a stock solution in DMSO at whatever concentration is convenient for you, then from that stock, make ~single use aliquots. We do recommend avoiding repeated freeze-thaw cycles on the same sample (hence the “single use” aliquots), but preparing a stock solution is, in and of itself, fine (and doesn’t require immediate, one-time use).
We’ve done a bit of work examining long-term storage and use of JF probes, and they seem to exhibit more degradation (not a lot, but still some… a few %) when the samples are repeatedly freeze-thawed. Oxidation of the dye and a minor amount of azetidine ring opening seem to be contributing decomposition pathways. This is more of an issue for some dyes than others… JF549 and bluer JFs seems quite resistant to such degradation, but JF646 and other red analogs are more prone to these issues.
If you are going to temporarily dissolve the JF dye in a solvent to aliquot and spin down back to solid, I actually recommend using MeOH (for all SNAP-tags, for all JF503 -> JF549 ligands, but NOT for NHS esters) and MeCN (for NHS esters and for JF585+ HaloTags), as it is much easier to remove these solvents. Even after extended time on a speed-vac, there will almost always be a little DMSO in the “dried” material, and DMSO is not the most innocuous solvent.
EDIT: I should also mention that we recommend using fresh, high-quality DMSO whenever possible. We prefer the Hybri-Max DMSO from Sigma, which can be ordered in boxes of 10 x 5 mL ampoules: https://www.sigmaaldrich.com/catalog/product/sigma/d2650?lang=en®ion=US