With the start of Run 3 data taking the standard pileup mitigation technique for the CMS Collaboration is the Pileup Per Particle Identification (PUPPI) algorithm. Recently, the track-vertex association in PUPPI was optimised in order to recover an inefficiency for hadronically decaying $\tau$ leptons at low transverse momentum ($p_T$).
The jet energy scale and resolution are sensitive to many different subdetector systems, making continuous monitoring crucial.
It is shown that Run 3 promptly reconstructed data have an excellent jet energy resolution performance with respect to Run 2.
Furthermore, a flavor-aware $p_T$-regression was derived and calibrated for the first time in the CMS Collaboration, which is expected to improve the jet energy resolution by up to 18\%.
Finally, two new algorithms have been developed to identify jets originating from the decay products of heavy particles. The first is a boosted decision tree that identifies hadronically decaying top quarks within variable-size cone jets over a wide $p_T$ range. The second is a neural network that distinguishes jets initiated by positively charged W bosons from those initiated by negatively charged W bosons.