Usage is proportional to the number of marker updates. Every time you update map markers, some computation is required to calculate the optimal position of the markers. The compute is provided by an API endpoint which is used by the library internally.

In case of a very large number of markers, the library will perform the computation in batches from the markers with the highest priority to the lowest priority. The usage will therefore be potentially less than the total number of markers added to the map.

Usage does not increase on cache hits. For example if you use the same markers on your map multiple times, subsequent compute requests will hit the cache and the usage is not increased. The cache is invalidated after a certain time.