Thermal emission in the prompt phase of gamma-ray bursts

I discuss the interpretation of the prompt phase in gamma-ray bursts as being dominated by quasi-thermal emission, rather than by synchrotron emission. Such an interpretation gives a more natural explanation of (i) the observed variety of spectral shape and spectral evolution, (ii) the observed narrowness of the distribution of peak energies, as well as (iii) the observed correlations between peak energy and luminosity. However, the physical setting that could produce such a scenario is not yet clear