I'm giving up trying to explain it to you, as no matter how I explain it, you just seem not to understand it giving it just one last go, see if these facts and rational thinking help you learn about it more, so you can be more informed - as I pointed out, there's enough documentation around LLMs across the board, from experts, that is clear about RAM needed for LLMs. Which you're deciding to ignore because you seemingly know better? What's your qualifications to support that you know more?
As for the RAM in the phones, whilst Apple spent years on the AI, I would suspect that Apple thought/aimed to have it be able to run on lower than 8GB when it was ready to launch (bearing in mind the hardware specs for the iPhone 15s were finalised early 2023 and AI isn't launching officially in beta until the Fall), and failed to meet that goal.
Your point about Android is flawed, as Android is not doing as much on device as Apple will be, and they have delays in coming back with answers, etc, as they rely on your internet connection to go out to the cloud and back again. Whereas, as per Apple's demos, they are doing most of it on Device, and keeping the cloud usage for when absolutely needed. It needs that RAM to be able to work that out.
There's nothing to say that maybe once it comes out of Beta, they can retroactively enable it on some of the 6GB devices like the non-pro 15's, we just don't know for sure yet. We only know what devices support it whilst it's in Beta come the Fall.
Also, promising 5 years support =/= every single feature is guaranteed, same as any Goolge/Samsung. With iOS 18, almost all the devices are getting about 90% of what's demoed (give or take), the only thing that's really not guaranteed to all the devices is Apple Intelligence. So you're misrepresenting the facts to try and make a point.