Warm white. I have done the red and blue thang, but here are the problems with that:

If the lights are not perfectly color mixed, not on the lamp, but on the leaves, then photosynthesis is not optimized. They need both at the same time and place.

Red and blue LEDs do not have the photonic output of white LEDs even though the spectrum of white LEDs is not designed for horticulture, but there is no problem with color mixing no matter the focal point as with bi-spectrum LEDs. The LEDs I am using for this experiment run 70-80 lumens per watt or somewhere between CFLS and MH, but the concentration is many times higher. Compare that to the normal 30 lumens per watt for most red and blue LEDs.

Now all of us edumucated folks know that lumens are not the way to talk about plant light usage, but it is a general concept until we get a new standard for PAR.