I'd call newer formats to have poor-man's deblocking, because they literally apply blur to blocky images, instead of reconstructing an optimal gradient.
In newer codecs you can notice that sometimes blocks of 8 or 16 pixels get blurred with a 4px-wide blur, which is barely enough to cover up the edges, but not enough to mask square-ish shapes underneath.
In newer codecs you can notice that sometimes blocks of 8 or 16 pixels get blurred with a 4px-wide blur, which is barely enough to cover up the edges, but not enough to mask square-ish shapes underneath.