Solar models computed with mass loss, microscopic diffusion of helium and heavy elements and with updated physics have been evolved from the pre-main sequence to present days; they are compared to the observational constraints including lithium depletion and to the Solar Reference Model (SRM) of Basu et al. (1996), derived by inversion. Microscopic diffusion significantly improves the agreement with the observed solar frequencies and agree with the SRM within +- 0.2% for the sound velocity and +- 1% for the density, but slightly worsens the neutrino problem. Neither microscopic diffusion nor overshooting explain the observed lithium depletion consistently with helioseismological constraints, while a mass loss process does it. Models computed with OPAL equation of state and opacities are in better agreement with the seimic sound speed. To reach the level of precision of helioseismological observations the accuracy of solar models still needs to be improved by one magnitude; any such improvement will necessitate equation of state and opacity data taking into account of detailed changes in the mixture.