We review a series of models of sociophysics introduced by Galam and Galam et al. in the last 25 years. The models are divided into five different classes, which deal respectively with democratic voting in bottom-up hierarchical systems, decision making, fragmentation versus coalitions, terrorism and opinion dynamics. For each class the connexion to the original physical model and techniques are outlined underlining both the similarities and the differences. Emphasis is put on the numerous novel and counterintuitive results obtained with respect to the associated social and political framework. Using these models several major real political events were successfully predicted including the victory of the French extreme right party in the 2000 first round of French presidential elections, the voting at fifty–fifty in several democratic countries (Germany, Italy, Mexico), and the victory of the "no" to the 2005 French referendum on the European constitution. The perspectives and the challenges to make sociophysics a predictive solid field of science are discussed.