When we get the offspring and fitness so why we prefer to apply mutation in genetic algorithm?
Mutation is (usually) needed to introduce new genes not found in the population. Suppose that you have 4 possible genes $A$, $B$, $C$, and $D$, and that your chromosomes have a non-binary encoding. In that case, if no member of your population has gene $D$, then no amount of crossover operations will result in the introduction of that gene.
I say that new genes usually need mutations to be introduced. However, if your chromosomes use a binary encoding, then new genes could be introduced as a side effect of crossover operations. But it is always safer to have some kind of mutation to ensure that all genes can be accessed.
Mutation is used to maintain diversity in the solutions. Crossover alone cannot do this.