That's not really much to go on. How are you changing the standard examples?
Firstly i have tried using the default single CW tone signal (with sampling time=1.3020e-07) and then generating my own custom signals (with varied frequencies and sampling times) and in all my cases i am detecting the input signal at the output of TX and there is no mixing of local oscillator frequency signal
The models don't really use the mixer this way. As they use the LO frequency parameter as a lookup for specific noise characteristics of measurements from physical hardware at those frequencies. Look under the block masks for RF_TX block and you will see this.