If you look a the docstring for defrecord in Clojure and compare to ClojureScript, you will see that ClojureScript doesn't mention that __hash is a reserved field name.
This should be mentioned, as it is used internally in the ClojureScript implementation of defrecord.
As an aside, while __hasheq is reserved in Clojure, there doesn't appear to be a reason to indicate it is reserved in ClojureScript. (The relevant Clojure commit is https://github.com/clojure/clojure/commit/a1c3dafec01ab02fb10d91f98b9ffd3241e860c0)
Also, in both Clojure and ClojureScript, fields documented as reserved for defrecord are also documented as such for deftype. It is not immediately clear to me why this is done, so in my mind it is an open question as to whether __hash should be documented as reserved in the ClojureScript deftype docstring.