How does Charles calculate latency?
Charles shows a measure of the latency on each request on the Overview tab. Charles calculates latency by measuring the time taken between finishing sending the request and starting to receive the response. Therefore latency includes network latency and server latency, that is the time spent processing the request.
For static requests the server is usually able to respond instantly, unless under load, so the latency figure mostly represents the network latency.
For dynamic requests, or any request for which the server has to do some work, you can subtract an approximate network latency to determine how long the server spent processing the request.