@theo I wish the subsidization of Muse Wonder Full was better on their sub I would prefer that over GPT at this point. Ember didn't feel that bad for what tier it is either.
@VydeogamzXD@Morrowmake Does it actually work out to being higher performance though? I know ROCM support is getting better I have a 7900XTX that got me into local AI but I'm struggling to find much info to point to it actually using its better specs. Maybe just need to grab one and work on optimization
@VydeogamzXD@Morrowmake At that point isn't a V100 32GB just better? Better software support and its cheaper as well as lower power? I was seeing V100's around 850 vs 1200 or so.
@alexandr_wang I like using Wonder Full over any GPT model but the usage limit is rough even on the $200. It’s a hard bargain when a model like Opus 5.5 is just subsidized better.
@Morrowmake Yeah I've been running 3.8 Flash Next for a bit and its great especially with the tweaks you've done and myself and others. I know someone was claiming it was on arena already along with Kimi K3.1 but I'm skeptical