>>539071307
Agreed, I think around 100k feels pretty standard for most high quality booth bases + clothing.

Once you start getting into the 200-300k tris range you better have a very good excuse for it, or have toggles so it's never displaying it all at once.
Anything above 300k you should start feeling shameful about even with toggles. If you don't have toggles and all 300k+ are showing at once, kindly kill yourself or even better, just fix your shit.

I'm sure some people will even get on my shit for saying something like 200-300k tris is fine with toggles. But if you take those toggles in to account you may only be seeing like 50k-100k tris at once on the avatar while the rest is hidden, so a 300k tri avatar may not really be as impactful on performance as we might think if we're just talking strictly about tris/polygon count and its affect on frame rate (if we're taking toggles in to account).
Or it might truthfully be a fully 300k tri avatar where all 300k tris are showing at once and it's fucking horrible on performance.
VRChat avatar ratings makes no distinction between an avatar that has 300k+ polys w/ toggles (so you're never seeing all 300k at the same time) and an avatar that has 300k+ polys all shown at all times. They're both very poor.
And an avatar that is only 70,001 tris is equally rated as very poor to the avatar which is 700,001 tris.

tl;dr: VRchat's ranking system is dogshit. A value of around 100k tris being shown at once seems like a fair enough limit going into 2026, and if you have something like 300k tris because you have multiple high poly hair meshes or something, as long as you set up toggles correctly so that only around 100k tris are ever shown depending on what hair you have active for example, then you really shouldn't be worrying much about the affect of tris/polygons on performance.
Things like texture size and amount of materials are probably more important to worry about since VRAM is such an important factor.