(opinions)
Vero29 offers the best efficiency to $/PAR W compromise, making it most suited for commercial LED grow lamps and has the advantage of better availability
The CXB could be geared more toward those who benefit from high efficiency (high electric cost, hot climates, stealth requirements etc) and is appealing for DIY
XML2/XPL is out of the race, no high bin 3K that I have seen, but the 4K is perfect for 3.6V headlamps
Value:
Vero V2.0 29 3K @ 2.1A (80W) = 40% for $0.90/PAR W (bingo!)
CXB3070 3K AD @ 2.1A (80W) = 42% for $1.20/PAR W
CXB3590 3K CB @ 1.4A (104W) = 40% for $1.25/PAR W
Value and Spread:
CXB2530 3K U2 @ .8A (30W) = 44.2% for $1.10/PAR W
Vero10 V2.0 3000K @ .35A (9W) = 37% for $1.20/PAR W
Spread:
XPL 3900K V4 @ 2.1A (7W) = 37% for $2.00/PAR W
Efficiency and Spread:
CXB2530 3K U2 @ .35A (12W) = 52.4% for $2.20/PAR W
CXB3070 3K AD @ 1.05A (36W) = 51.3% for $2.10/PAR W
Max Efficiency and Spread
CXB3070 3K AD @ .7A (23W) = 55.4% for $3.00/PAR W
CXB3590 3K CB @ .35A (23W) = 58.7% for $3.80/PAR W
(caveat)
The figures above are assuming 4.33% temp droop at Tj 50C for the Veros, per the datasheet. But my measurements show only 2% temp droop, repeatably. Assuming there are no systemic errors in my test, typical efficiency for the Vero29 3K V1.2 would be about 38.3% and for the Vero 29 3K V2.0 @ 2.1A would be about 41%. I think these are the efficiencies we will be dealing with in the most competitively priced commercial LED lamps.