CMB data are beginning to put limits on parameters which are directly
relevant for particle physics models. For example there is a limit on
the neutrino contribution
h2 < 0.0076 (95% confidence)
from a combination of WMAP and galaxy clustering data from the
2dFGRS project
[47].
This directly implies a limit on neutrino mass, assuming the usual
number density of fermions which decoupled when they were relativistic.
A combination of the WMAP data with other data-sets gives some hint
of a running spectral index, i.e.,
dn / d ln k
0
[42].
Although this is still far from resolved
[48],
things will certainly improve as new data come in.
A convincing measurement of a non-zero running of the index would be quite
constraining for inflationary models
[49].
One other hint of new physics lies in the fact that the quadrupole
and some of the other low 
modes seem anomalously low compared with the best-fit
CDM model
[32].
This is what might be expected in a universe which has
a large scale cut-off to the power spectrum, or is
topologically non-trivial. However, because of cosmic variance, possible
foregrounds etc., the significance of this feature is still a
matter of debate
[50].
In addition it is also possible to put limits on other pieces of physics [51], for example the neutrino chemical potentials, time variation of the fine-structure constant, or physics beyond general relativity. Further particle physics constraints will follow as the anisotropy measurements increase in precision.
Careful measurement of the CMB power spectra and non-Gaussianity can in principle put constraints on high energy physics, including ideas of string theory, extra dimensions, colliding branes, etc. At the moment any calculation of predictions appears to be far from definitive. However, there is a great deal of activity on implications of string theory for the early Universe, and hence a very real chance that there might be observational implications for specific scenarios.