seems those two things are going to be heavily blocked moving forward

yea, that makes sense, GLM though I think has less guardrails

ya, but it's not reasonable to run locally

need like 4x RTX6000 (so 60-70k) or the GB300 workstation (>80k)

Modified by: swiftraccoon on 2026-06-26 02:02:16 PM

not spending house money on hardware that will be obsolete very quickly

esp at these inflated prices

Maybe soon though....https://www.macrumors.com/2026/06/25/m5-ultra-mac-studio-2026/



Modified by: PitViper on 2026-06-26 02:41:11 PM

Yes, likely buying one if their prices aren’t insane.

Lol with ram prices, that will be insanely priced.

yeah, but they could charge the same as what it would cost AMD/NVIDIA GPU equivalent which would truly be insane

i doubt they will. it'll probably be 20-30k which compared to the greed of AMD/NVIDIA isn't all that insane

certainly could be 40-50k too which gets into needing a return on investment/business justification

seems Qwen has some guardrails as well.

yes, both gemma4 and qwen3.6 do but there are abliterated models. even the defaults are very easy to jailbreak though.

also should note local really only makes sense if you're trying to do the blocked things on OpenAI and Anthropic. if never getting blocked, the $200/m plan is impossible to beat.