
You already own the objects. Exact copies are eating space.
file.rocks is the UI. Cleanup finds same ETag or content hash. You keep one copy. The rest leave the bucket for good.
This is Likely duplicates under Settings → Cleanup. Exact match. It is not how to clean up near-duplicate photos in your own bucket (visual similarity). It is not Files row Delete. That hard delete path is how to delete a file from your own bucket.
You need S3-compatible object storage. Not another Drive clone.
If you do not have a bucket hooked up, follow how to connect a Backblaze B2 bucket and come back. For the price table, see the cheapest cloud object storage comparison.
Sign in at file.rocks. Open Settings → Cleanup. Pick the bucket if needed.
Open Likely duplicates (/settings/cleanup/duplicates).
The meta line is Same ETag or content hash. Keep one copy. Guide: Newest is kept by default. Change Keep if you prefer another path.
Empty state: Nothing in this category. Optional Show dismissed / Restore.
The wizard is 1 · Select / 2 · Confirm.
Select groups (Select all / Select none). Per copy you see badge Keep / Delete and button Keeping this copy / Keep this instead. Optional Dismiss selected.
Click Review deletions. Confirm step says Final check before permanent delete. Click Delete N permanently (busy: Deleting…).
The modal asks: Permanently delete N file(s) and reclaim about X? This cannot be undone. Buttons Delete permanently / Cancel.
Toasts: Deleted N file(s) / Deleted with N error(s) / Dismissed N suggestion(s).
This path is permanent delete. Do not invent trash or undo here. Soft index rows may show as deleted after.
This is not visual near-duplicates. Burst shots and lookalikes stay in how to clean up near-duplicate photos in your own bucket.
This is not toolbar or row Delete in Files. That path is how to delete a file from your own bucket.
Need visual near-copies instead? That is how to clean up near-duplicate photos in your own bucket.
Need a single hard delete outside Cleanup? That is how to delete a file from your own bucket.