Based on the limited information provided, those results don't seem obviously wrong to me. A 9.25 COP AC chiller seems very high for a full load efficiency, and will likely have even better performance at part-load than your DX baseline systems.

Heat rejection savings are often high in our LEED models as well for two main reasons. 1) HR devices often have variable speed fans and fan affinity laws make a big difference, even compared against 2-speed baseline equipment (if your baseline was system 7). 2) IES for some reason setup DX curves that reports a percentage of energy as heat rejection, with the rest obviously being compressor energy in the "cooling" category. While this is logical, it's not typical in most modeling tools. This "extra" HR energy in the baseline may also contribute to what seems like excessive savings.

I agree with Greg. Your results don't sound unreasonable to me based on comparing an ACC to packaged DX baseline equipment. Your ACC COP seems unrealistically high though, which is contributing to the savings in cooling and heat rejection. The best full load ACC COP i have seen is about 3.5.

The negative pumping savings is because the baseline has no pumping energy, which is part of why comparing an ACC to packaged DX is difficult.