Looks very good, but agree with Teyon's comments about the trapezius, and the head may be just a little bit small.

I would also like to expand a bit on Teyon's point about modelling in one pose, but morphing to another (T) pose for the final mesh. I think this idea is also important from the standpoint of UV mapping. But first let me come clean, I have never modeled anything more complicated than a sphere myself. So what follows is just off the top of the head stuff from someone who does not really understand the modelling process.

I remember when the Antonia figure was being developed, it ran into an issue where the texture on the eyelids stretched a lot when the eyes were closed. As eyes-closed and eyes-open are almost at the two ends of the possible range it would seem sensible to me, to model and UV map the figure with the eyes half closed, as this requires less stretching of the mesh and UVs to reach the extreme end of the range.

On the same logic, perhaps the figure should be modeled with the mouth partially opened, the thighs slightly spread, and bent forwards, say 30º, the knees bent back 30º, the upper arms bent forwards slightly, the forearms bent inwards, and the fingers slightly curled in.

The topology and UV mapping would be finalized in this "median pose". The mesh would then be straightened to a T pose with the mouth shut and the eyes open to a normal degree.

Of course this is all very easy for me to write, but may be fiendishly difficult to do in practice. Which is probably why it is not done as a matter of course.