Gamma ray bursts (GRBs) are usually classified as long or short based on the duration of the burst. They can be completely dominated by gamma rays (classic GRBs), show an abundance of X-rays (X-ray-rich GRBs) or, it appears, even be dominated by X-ray emission with few or no gamma rays detected at all. These latter are now known as X-ray flashes and account for 20-30% of all GRBs.
Although it is not immediately obvious based on the wavelength of the dominant emission, most astronomers believe that X-ray flashes, X-ray-rich GRBs and classic GRBs form a continuous sequence. Like GRBs, the emission observed for X-ray flashes starts quickly and seems to be a one-off occurrence. They are distributed homogeneously in the sky, have durations ranging from 90 to 200 seconds, afterglows at lower energies are detected, and they are located in the star forming regions of galaxies at cosmological distances. In fact the only apparent difference between X-ray flashes and classic GRBs is that their peak energies are lower.
While it is generally agreed that these objects arise in a similar manner to long duration GRBs (i.e. through hypernova explosions), an explanation of the reduced peak energy remains a matter of debate. One currently popular theory suggests that X-ray flashes are simply classic GRBs viewed slightly away from the axis of the beamed radiation. In this scenario the emission is dominated by mildly relativistic material in the wings of the jet, rather than the highly relativistic material beamed along the axis and away from the line of sight.
Another idea, called the dirty fireball model, is that the material ejected in the explosion is heavy with baryonic matter. This would act to slow down the expanding fireball, making it less able to boost photons to gamma ray energies.
In either case, we would expect a hypernova to accompany the X-ray flash. To date there are only a handful of X-ray flashes that have been localised to arcminute accuracy and studied extensively, but at least one of these has shown conclusive evidence for an underlying hypernova.