The increasing performance of static converters leads to more severe operating conditions. Because of the electrical waveforms used or the proximity of heat sources, magnetic components are exposed to high temperatures which can degrade the performance of the magnetic material. That's why predicting the temperature behaviour of magnetic components is a key step in the conception of a new generation of converters. In power electronics, nanocrystalline materials are increasingly used because of their low losses with high frequency and their good temperature stability. The aim of our work is to model the magnetic behaviour of a nanocrystalline core with increasing temperature and frequency, in order to implement it in a circuit simulator in the future. Thus the model should be as accurate as possible while remaining simple and fast. Our model is based on the separation of static and dynamic losses. Given the thin static hysteresis of this material, a reversible mathematical function is used as static law. Moreover, in the intended applications, frequencies can be up to 100MHz. In this context, taking the magnetic diffusion into account is essential. That is why we have implemented a model using a fractional derivation, requiring few parameters and with short computation time. The variation of temperature is taken into account in the model parameters.