Value iteration: 

(i) Is a model-free method for finding optimal policies. 

(ii) Is sensitive to local optima. 

(iii) Is tedious to do by hand. 

(iv) Is guaranteed to converge when the discount factor satisfies 0 < γ < 1.