So you simply assumed it was EOL.
What do they say about assuming things? It makes an Ass of U and Me!
My dilemma is, should I upgrade to openwrt 25 or wait for GL.iNET to release the “official” v25.
I guess this is a mather of preference too.
Heres my take on this, if your interests are in a easy responsive ui then I think the gl inet firmware is still the winner.
But if your aim is more towards more advanced or package stability then you want to use openwrt directly without the gl part.
As far as stability goes especially with the latest openwrt releases I suggest reading their forum release page and the topics there, I know that one of the latest releases had introduced some problems to the mt76 driver or atleast the duid functionality.
But sometimes in these extreme cases they may also backport things from the master branch to fix it.
To me the bugs do no longer exist on master like the slow tcp experience I had mt76 was affected or the DUID problems, for Duid the workaround is very easy by just making the field in global settings empty.
For homelabs and dumbaps I personally think it's easier to build your own images with pre config because you can basically design your own 'sdk' with a minimum installation of all your lab vlans as factory defaults when reset is pressed.
Also I won't specifically do this for security if the gl ui is of importance while it is true vanilla OpenWrt has a higher kernel and vulnerabilities fixed, it doesn't mean GL-iNet can backport them, but then I won't stay on the op24 which had no updates for some time, but I would go back on the MTK SDK which has a older OpenWrt base which they still seem to actively maintain, but you have to trust them on their word without the full visibility you have with open source.
For me the reasoning to go with self build images or openwrt directly is purely because my advancement started to conflict with the gl software and sometimes also their impl ideas and when even all the hacks and workarounds broke, or packages showed broken things my choice was simple for stability but also complexity openwrt was the better choice it will surely be a much better experience if things don't randomly break and the control is yours how behaviours work 
, and by creating your own images you can also build very advanced on top into extremely advanced networks 
Also one of the fewer things which don't work on mtk sdk are multi psk networks, but if I'm honest this also doesn't work 100% on OpenWrt it fails if you do this with fast roaming enabled and often you randomly hop to vlan 1, this is why self compilling is usefull because you can test patches by others which fixes it.
I'd still use gl software on their travel routers, where convience is a must such like the repeater function, I can wireguard into my home network and basically have access to the full lab backend and my own private docker cloud and have all the nice things, i can also moonlight via a tablet connected to the travel router.
I think people just should edit some file to make it look like they running openWRT 25 and they will be happy.
I doubt anyone can name REAL thing that in v25 is better or lacking in v24.
We, enthusiast just have illness that makes us always want to try new versions, to optimize….in short-to waste our life on not important stuff. All that security talks and so on is BS.
Like you running FBI servers connected to flint2. Biggest security risk is - user -administrator. not old binaries