We have proposed a new stochastic interpretation of the sudiffusion described by the Sharma-Mittal entropy formalism which generates a nonlinear subdiffusion equation with natural order derivatives. We have shown that the solution to the diffusion equation generated by Gauss entropy (which is the particular case of Sharma-Mittal entropy) is the same as the solution of the Fokker-Planck (FP) equation generated by the Langevin generalised equation, where the 'long memory effect' is taken into account. The external noise which pertubates the subdiffusion coefficient (occurring in the solution of FP equation) according to the formula D α → D α /u, where u is a random variable described by the Gamma distribution, provides us with solutions of equations obtained from Sharma-Mittal entropy. We have also shown that the parameters q and r occurring in Sharma-Mittal entropy are controlled by the parameters α and u , respectively.