Generally,foundry provides G-P or VBIC model for GaAs HBT. But the measured results show both models can not predict IMD performance accurately. So the design can not be guaranteed especially in designing high linearity amplifier with HBT process. Most of such amplifiers employ predistortion technique, so the model should have the capability of predicting nonlinear performance.G-P or VBIC usually can not model properly the variation of the delay time with current increasing. Agilent HBT model seems to do it,but I'm not sure. Would anyone like to share his/her experience?