The JS API is (and has been) prototyping labels for clusters for some time. We are looking a couple of possibilities. Do you only care about the count in the cluster, or do you want more control over other summary info, like averages, sums, and other statistics in the cluster?
If you have other use cases for cluster labels, then please reply to the poll or in this question.
I would definitely like more control over cluster labels. In my particular case I have an accident layer where the number of fatalities is an attribute. If I have a cluster with 10 points (i.e. 10 accidents), 8 in which one person died, and 2 in which two people died, I want to show the number 12 and not 10. Is this possible?