248 kbps MP3 does exist at least, for LAME encoders? But 256 kbps is normal.
From this we know some high-res files offer little benefit over the compressed low-res equivalent.
And the history regarding SBR is intriguing. For non-engineers this thread is the only place to understand the history of DSEE HX. Although there are many algorithms to upscale audio to CD (early 2000s), to HiRes (since 2013), and now using AI to increase accuracy for different types of audio (2018), I would say DSEE HX is the most famous and the best algorithm to upscale MP3s for the mass market to embrace HiRes products. Its problems include harsh high frequency sounds for highly compressed audio and certain vocals. It can lead to muddy sound and its stereo enhancing effect is experienced differently depending on whether you use IEMs or headphones, and its make/model. And differences can only be noted in very highly compressed (96 kbps) MP3s which have lots of instruments in them, or in the way harsh sounds turn less 'jagged' - smoothed out in a way. When it comes to 64 kbps it is not so good. Since 2018 DSEE HX AI (now DSEE Ultimate) introduced extra tweaks to increase soundstage 'height' - the beating of drums and audience/ambience of live concerts' clapping, so long as the source file still has some of that left over as a hint. It is well known that different Sony products had different DSEE HX sound effects. Some make no difference at all, some give piano sounds extra depth. Even DSEE Ultimate still makes mistakes with vocals. So it is at best a comforting function to enable MP3 users to save eMMC space and still listen to HiRes.