Let $M$ be an abelian group, and consider a “projective resolution” $$ 0 \to M \to P_0 \to P_1 \to \cdots \to P_n \to 0. $$ Note that this is not a projective resolution in the usual sense since it's going the wrong way! But I claim that there must exist a projective resolution $P_\bullet \to M$ going the right way. The argument is as follows. The exactness of the above sequence exhibits a quasi-isomorphism $M \to P_\bullet$. This arrow must be invertible in the derived category, and since $P_\bullet$ is “cofibrant” (here I am using the boundedness of $P_\bullet$), there must exist a map $P_\bullet \to M$ realizing this inverse. Thus, $P_\bullet$ is a projective resolution of $M$ in the usual sense.
Is this correct? I'm mainly asking as a sanity check since this situation seems so strange to me.