Best person to ask this question Dr. Jan van Deursen of Mayo clinic. or Janet Rossant of Toronto, I did work with both of them partly or in collaboration. They do regularly KO generation. van Deursen can best address you this if any improved method has come up.
In principle CRISPR or TALEN or zinc finger nuclease methods are more efficient, but you should consider other issues. If you use these methods, and you just induce a cut and a gene disruption by non-homologous end joining, the precision of gene editing is inferior to what you can achieve with homologous recombination. In addition, you have to consider the likelihood of off-target effects (cuts or recombinations at loci other than the ones intended), which is much lower with homologous recombination, due to the usually much larger regions of homology. I guess time will tell if homologous recombination will become redundant one day, for now I think it's too early to tell.