You are talking about Haskell.

The lambda calculus "in its purest form" is not typed.

Common Lisp, at least, is also (dynamically) typed.

Common Lisp, however, has a "top type" T which is a supertype of all types. So one can manipulate lambdas themselves as if they're untyped.

Of course, I personally don't think programming in Lisp has much to do with wielding lambda calculus in any superlatively direct sense.