The proofs and security of block ciphers constructed especially of Luby Rackoff Constructions and Feistel Networks is based on the number of queries and round functions. The security measure is always based on the probability with which the attacker can distinguish the randomness in the cipher text. For example Patarin's proof says the encryption can be distinguished in O($ m^2/2^{(l/2)}$) where m is number of queries and l is length of input bits.
But in general the bigger the key length it is considered as good against brute force and adding tweaks is considered provides additional randomness. Then why is the security measure not involving the key length or tweak length ? or is there a security measure that involves key and tweak ?