A self-extracting archive opened from the filesystem now reads itself. An archive saved without a recovery payload, which is the default, could not be opened from disk at all; Firefox performs the read that makes it work.
An iframe holding an SVG document keeps its content instead of being saved empty.
A sandboxed frame is saved as it renders, rather than as its source attribute reads.
A page declaring the same font twice embeds only the font it actually uses, so saved pages are smaller.
A font declaration left without a usable source is removed instead of being written out invalid.
Fonts offering several sources keep the best one for the browser again, instead of all of them.
A BMP or GIF image served without a content type is stored with its real extension.
With the mixed content option enabled, resources created by the page itself are no longer blocked.
An image whose responsive sources could not be saved falls back to its main source.
A canvas drawing is no longer stretched or cropped by a page rule fixing the background.
A resource the page could not fetch itself is retried directly again.
Pages inspecting the browser's built-in functions during a save see them unchanged.
Data appended after a self-extracting archive is kept within the window the narrowest ZIP readers scan, so more tools can read the saved file.