This is going to be due the difference in your environment. Cron has a pretty empty environment by default.

Best thing to do is type "set" on the command line and add "set" to the script and compare the two. There will be one or more values missing from the cron environment which will need adding in the script (export xxx=yyy)

Each LIST request will return something like 1,000 files, with multiple requests needed to fetch details of further files.

As a result, say you have 3,000 files that would require 3 LIST requests to fetch details about each. This is needed on each s3sync invocation to build the server list, for comparison with the local list.

The cost of running s3sync is:

Cost of LIST requests (which depends on the number of files already on the server)
+
Cost of files to upload (new files not on the server)

With the former always applying even if there are no new files to upload.