That's just the HTTP range request used when it requests something from a cloud remote.
This would be specific to a mount and the cache mode doesn't matter.
If you set it high, you might get a little bloat / extra data before it closes out. Specifically how Google calculates that range request to your quota is unknown.
So if I have a file and I want to read say 1KB of it and my default range request is 128M, I start a larger chunk download for 128M but once I get my 1KB, it'll close out and I'll bleed into a little bit extra download.
@VBB was testing with 1MB range requests and that generally seemed to work and on a large library scan, it would have some impact. Without knowing how the quotas work though, it's a guess if it is a material impact or not.
In the logs, you see something like chunksize on the debug as it'll scale up based sequential reading as if it reads a file longer, it'll start to double the ranges to help performance.
This gives a great explanation on it:
New Feature: vfs-read-chunk-size - Howto Guides - rclone forum