This thesis introduces additions to the existing R package GMNL which estimates multinomial logit models with unobserved heterogeneity across individuals for cross-section and panel data. The use of multinomial logit models has become quite popular due to their interpretable nature and they can be applied in a variety of fields. The models supported by gmnl are the multinomial/conditional logit, mixed multinomial logit, scale heterogeneity multinomial logit, generalized multinomial logit, latent class logit, and mixed-mixed multinomial logit. New features in the package include the option for utility models to be specified in the willingness to pay space, robust confidence intervals in the willingness to pay space, elasticities, robust standard errors and statistics to evaluate model fit. This work highlights improvements to the currently published portions of the package such as added documentation and improvement of logical evaluation statements as well as examples highlighting the applications of the added functionality.