Probit model is a type of limited dependent variable estimation techniques. It can either be used when the latent dependent variable is binary in nature eg. Yes or no, or ordinal In nature. The underlying distribution assumption of probit is that, it assumed the error term is normally distributed. Which is why it's different from it's counterpart logit model which assumes the error term followed a logistic distribution. You can read Long and frees or microeconometrics with by Trivedi