Musculoskeletal modeling and simulation software available give a non-invasive prediction of joint contact forces. However, the results are very sensitive to optimization techniques and model choice. In this study, muscle activations and knee joint contact forces (KJCF) were predicted for squatting using two different optimization techniques along with two musculoskeletal models. OpenSim 4.2 was used to simulate squatting using motion capture data and ground reaction forces. Electromyography (EMG) signals were filtered, rectified, normalized, and compared with predicted muscle activations. Good agreement was observed between predicted muscle activations using OpenSim and the EMG with few exceptions using both models. KJCF showed a significantly large difference using both models and optimization techniques. The study demonstrates the sensitivity of results to optimization techniques and model choice. The study also shows the incapability of musculoskeletal models to predict KJCF during squatting.