Background Musculoskeletal modeling has paved the way of measuring kinematic and kinetic variables during motions. Nonetheless, since the commonly-used generic models are created based on averaged data; thus, they cannot accurately mimic subjects with skeletal deformities. To overcome this obstacle, one can build personalized models based on subject’s MRI or CT scan data, which is both time and money consuming. The other promising way is to manipulate generic models and create semi-personalized models to match with the individual’s skeletal system at the joint of interest. Research Question Can a semi-personalized model reduce marker error in gait analysis? How a semi-personalized model differentiates the ROM of the lower limb joints and muscle activation pattern while having varus deformity? Method We developed the varus-valgus tool (freely available on: https://simtk.org/projects/var-val-tool) in MATLAB using OpenSim Application Programming Interface (API) to incorporate varus-valgus deformity in the generic OpenSim models. A 36-year-old female subject with a complaint of knee pain participated in our study. The subject had 6.5 and 11.9 degrees of varus in the right and left leg, respectively. A semi-personalized model of the subject was first created using generic OpenSim models. Then, markers’ error during Inverse Kinematic (IK), joints Range of Motion (ROM) and the activation of Tensor Fasciae Latae (TFL), a knee adductor, and Gracilis, a knee abductor, were calculated and compared between a semi-personalized model and a generic model. Results Significant difference was observed in markers’ error during IK between generic and semipersonalized models (p<0.05). Substantial alterations were found in the ROM of the hip, knee and ankle joints while using semi-personalized model. Moreover, the activation pattern of TFL experienced a dramatic rise whereas Gracilis saw a fall during each gait cycle in semi-personalized models. Significance Implementing varus-valgus deformity in the generic models substantially reduces markers’ error which leads to more accurate results. It was observed that semi-personalized models showed different ROM compared to generic ones.