The Collapsar model provides a framework to the well known link between long duration gamma-ray bursts and dying massive stars. According to the model the core of the massive star collapses and form a compact object at the center of the star.This object launches a jet which penetrate the star, accelerates and emits the gamma-ray photons far from the star. It follows that a necessary condition to the formation of a GRB within the framework of the Collapsar model is the successful breakout of the jet from the star. This breakout leaves a distinctive imprint on the distribution of GRB durations: a plateau in the distribution at times shorter than the typical breakout time.
In this talk I will show that such a plateau does exist in the duration distributions of GRBs observed by all major GRB satellites. I will discuss the implications of this plateau on low luminosity GRBs and present a new method to classify Collapsars and Non-collapsar GRB. Finally I will discuss the different physical models for Collapsar jets and show how we can use the breakout time of the jet to distinguish between them. This distinction can provide us some valuable clues to the actual type of the compact object that resides at the center of collapsars and to the launching mechanism of the jet.