Refer to PEP 3137 and PEP 358. Bytes objects are for conveying binary data (or encoded strings). Such binary data is customary specified in hex-dump or base64 format in source files. It would be nice to introduce a way in python to do that 'natively' (at lexical analysis time) using x-strings and y-strings (in lieu of f-strings). x-strings: x"0123456789abcdef" (x-prefixed quoted string of even-numbered amount of characters of the set [0-9a-fA-F], whitespace ignored) equivalent to: b"\x01\x23\x45\x67\x89\xab\xcd\xef" (but more readable) y-strings: y"ASNFZ4mrze8=" (y-prefixed quoted string of valid base64, whitespace ignored) equivalent to: b"\x01\x23\x45\x67\x89\xab\xcd\xef" (but shorter) This is not a replacement of the hex/base64 encoding, binascii packages etc. It just gives the programmer more freedom to specify literal bytes objects in the source code.