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

precedence:

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