if you are sure that you are entering the same public keys and required signatures (2 of 3) in both places and also are generating the same type of multisignature address then the only reason why the resulting addresses could be different is if one of the tools is "sorting" the given public keys and the other one doesn't so the hash of the sorted keys is a different one and as a result you get a different address. (read my edit below about sorting if you are using bitcoin core)
try changing the order in the tool (coinb.in) that doesn't sort them to see this.
example of this case with 3 random public keys that i generated using bitaddress.org for this purpose only, and using coinb.in site for the multi sig addresses:
public keys with orders:
038B66DD2C771B58271C995D8ECBF06EE60DEF61BA4B843F4AA9731DD724FDA87C
034B966F86E8DE46A58C456D0695C9B2ABB2BD7F35F0BC62F0DE45A816BD667413
02F87E983C5A1033D2EA39B5985360947A526DDB495417D1FD121B66BCA859220E
address:
3Eoq2jmc66Qv4ziRoUKkHAFkwHncNhj7mP
same pubkeys with different order (changed place of 1 and 3)
02F87E983C5A1033D2EA39B5985360947A526DDB495417D1FD121B66BCA859220E
034B966F86E8DE46A58C456D0695C9B2ABB2BD7F35F0BC62F0DE45A816BD667413
038B66DD2C771B58271C995D8ECBF06EE60DEF61BA4B843F4AA9731DD724FDA87C
address:
3AzS4Wk6xQ2tmKw6pKP4jwwa3eyw7E8tbC
edit:
i checked my unseynced bitcoin core version and it seems like i was wrong, it also doesn't sort anything so you must have entered the keys wrong. the equivalent of the above exampels are the following commands which did return the same exact address:
addmultisigaddress 2 '["038B66DD2C771B58271C995D8ECBF06EE60DEF61BA4B843F4AA9731DD724FDA87C", "034B966F86E8DE46A58C456D0695C9B2ABB2BD7F35F0BC62F0DE45A816BD667413", "02F87E983C5A1033D2EA39B5985360947A526DDB495417D1FD121B66BCA859220E"]'
addmultisigaddress 2 '["02F87E983C5A1033D2EA39B5985360947A526DDB495417D1FD121B66BCA859220E", "034B966F86E8DE46A58C456D0695C9B2ABB2BD7F35F0BC62F0DE45A816BD667413", "038B66DD2C771B58271C995D8ECBF06EE60DEF61BA4B843F4AA9731DD724FDA87C"]'