Does every 4-regular graph contain a cycle of length (number of edges in the cycle) $1 (\mod3)$? Are there only finitely many exceptions?

I suspect such cycles exist for most 3-regular graphs but 4-regularity is enough for what I'm investigating.