SVM regression with logistic link function?

I'm using an elastic net classifier for a model that requires probabilistic outputs between 0 and 1, and I've found that SVMs give much better classification accuracy than the glmnet model (not too