Different values between pymc3 and stan

Questions version agnostic
2

Those two models don’t seem equivalent. Is the STAN one a Bernoulli mixture likelihood? In that case you should use a pm.Mixture with Bernoulli components