Response time (RT) models traditionally consider positive continuous distributions, for instance, the lognormal, gamma, exponential, and Weibull distributions, with support < 0, ∞ >. However, usually in Assessment, the time that an examinee takes to complete a test is limited and not infinite as the previous models assume. By considering this fact, the purpose of this article is to model RT following a bounded distribution and then in combination with response accuracy to obtain a joint model. Specifically, the use of the simplex distribution is proposed to model RT adopting the Bayesian inference. Performance of the proposed model is evaluated in a simulation study and the PISA 2015 computer-based reading data is used to apply the model.