Tor follows the traditional client/server model, it just uses a few connection layers to attempt to make it harder for the server (or any listener, of course) to identify the origin of the connection. That means someone must be running that server, which means it's possible to shut it down and punish the operators, and those operators can in turn identify the clients by any means except those directly prevented by the layering model.
Freenet is a distributed content-addressed store, much like IPFS, except that rather than you directly fetching content from other users who have specifically "pinned" that content (outing you both as interested in the material), the request hops over multiple users, leaving it cached (in an encrypted form, for plausible deniability) over their machines, so that it's very hard to tell who actually requested it.
And nobody uses it.