The native integration in the Filezilla is outdated, and the Filezilla team is procrastinating to update it. So as result it can download either older objects or objects uploaded via Filezilla.
I would recommend to use an Uplink CLI or rclone to download your files back.
Since you using an objects browser on the web-site, you opted-in for server-side encryption, thus you can also use Cyberduck.