So probably never ?
It really depends on how much demand there is for this. Aurora performance is very heavily dependent on tuning for particular memory sizes as this is key to the level of asynchronous I/O Aurora performs. So supporting different memory sizes, in addition to different instance families, is very resource intensive. That also means each new release has to be re-checked on each instance type to make sure performance has not been degraded, making it an ongoing resource drain rather than a one time thing. As a result supporting instance types for which there is little apparent demand loses out to applying those resources to higher priority customer requirements. Bottom line, if they are hearing a lot of customer requests for 8GiB they'll do it, but otherwise probably not.
Even if this is in the plans, AWS rarely publicly discloses those plans.
want to activate Disk Usage and Ram Usage metricsAccepted Answerasked 9 months ago
Transit Gateway shared with AWS Resource Access Manager (AWS RAM) identify all accounts as externalAccepted Answerasked 3 years ago
change EC2 parameters (cores, RAM) after I have installed a GUI and some softwaresasked 9 months ago
Why AZStd::vector do not cleared RAM?Accepted Answerasked 5 years ago
No instance with 8GiB RAM ?asked 3 years ago
How to monitor RAM memory usage of an EC2 Ubuntu instanceAccepted Answerasked 9 months ago
AWS OpenSearch Service recommended JVM heap size is often set to 50% of system RAM. If JVM heap size is limited to 32GB, does that mean there is no point in using node types with more than 64GB RAM?Accepted Answerasked 3 months ago
Is it possible to access/create other instance sizesasked 2 months ago
Instance Type to Useasked a month ago
3D RAM as registerasked a year ago