There are already several wonderful monographs mentioned above, and I may mention a couple more in another answer too (below this one). However, if there was just one to choose, I would select one here without any hesitation, be it on number theory or from all mathematical books.
This book is not too well spread around because there is only a limited number of copies of it. (Also, it has a strange peculiarity which I will mention in a moment). The special (for me) book is
Emil Artin, Theory of Algebraic Numbers
The book is based on Artin's notes taken by Gerhard Wurges, 1956/7. It was translated into English and distributed by George Striker. Both deserve gratitude for their contribution.
The impressive thing (one of them) is that the text is completely void of any public relation chat. Any introductory remarks to the consecutive segments of the text are minimal, almost non-existent. And still, the text rolls smoothly and naturally. Of course all experts can appreciate the beautiful construction, discretely axiomatic, with axioms taken almost for granted.
Now about peculiarity. You may see that at one moment a statement is called a theorem where this is really not any theorem to be called by such a proud name. And then soon you can see an unannounced formulation which actually is a theorem. Well, at least at one time the note taker. as well as the translator, didn't concentrate too well. Despite a somewhat imperfect medium, the monograph is beautiful all the same.
There was just one not strictly mathematical omission in the book, of not crediting at all Stanisław Mazur for his theorem about the division Banach algebras. But then, Mazur was hardly ever fully credited for thishis wonderful result anyway.