YES

TRS:

  filter(cons(X),0(),M) -> cons(0())
  filter(cons(X),s(N),M) -> cons(X)
  sieve(cons(0())) -> cons(0())
  sieve(cons(s(N))) -> cons(s(N))
  nats(N) -> cons(N)
  zprimes() -> sieve(nats(s(s(0()))))

max/plus interpretations on N:

  filter_A(x1,x2,x3) = max{0, -8 + x1, 6 + x2, 0}
  filter#_A(x1,x2,x3) = max{15, x1, 11 + x2, -1}
  cons_A(x1) = max{1, 8}
  cons#_A(x1) = max{14, 9}
  0_A = 2
  0#_A = 0
  s_A(x1) = max{2, -1}
  s#_A(x1) = max{14, 7}
  sieve_A(x1) = max{7, 8}
  sieve#_A(x1) = max{10, 7 + x1}
  nats_A(x1) = max{8, -3 + x1}
  nats#_A(x1) = max{15, 13 + x1}
  zprimes_A = 8
  zprimes#_A = 16

precedence:

  zprimes > filter = nats > cons > s > sieve > 0