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()))))

linear polynomial interpretations on N:

  filter_A(x1,x2,x3) = 1
  filter#_A(x1,x2,x3) = x2
  cons_A(x1) = 1
  cons#_A(x1) = 2
  0_A = 3
  0#_A = 0
  s_A(x1) = 1
  s#_A(x1) = 0
  sieve_A(x1) = 1
  sieve#_A(x1) = 1
  nats_A(x1) = 1
  nats#_A(x1) = 3
  zprimes_A = 1
  zprimes#_A = 4

precedence:

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