Mixed
AG
Cisinski-Deglise paper(s)
Summary in Levine: Smooth motives
I think Deglise said in early 2009 that the problem they had over a general base is the localization thm. So far I think (???) he said they can do it for a closed immersion of smooth schemes, I think over a general regular base.
nLab page on Cisinski-Déglise motives