SVM is a fundamentally $2-$class classifier, but if we have more than two classes, we can overcome this problem. Hence, using SVM to perform a k$-$class classification by using:

$$ One$-$versus$-$rest

One$-$versus$-$one

Both one$-$versus$-$rest and one$-$versus$-$one

None of the above as SVM is stricly a $2-$class classifier