In the paper Qubic: $4 \times 4 \times 4$ Tic-Tac-Toe, Patashnik proves that even the $4^3$ game is a win for the first player with optimal play; therefore the $4^4$ game should also be a first-player win. The proof is a computer-aided solution, however, which might not be satisfying.
A table in this paper shows that the status of the $5^5$ game was still open at the time (in 1980). It's possible that the problem has been resolved by now, but it makes me skeptical that an "obvious win for the first player" exists; it definitely wasn't obvious to anyone at the time!