Gpt and AI Discussions : General Discussion Area

[ViperDiscord]

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

[ViperDiscord]

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

[ViperDiscord]

ya, but it's not reasonable to run locally

[ViperDiscord]

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

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

[ViperDiscord]

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

[ViperDiscord]

esp at these inflated prices

[ViperDiscord]

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

[ViperDiscord]

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

[ViperDiscord]

Lol with ram prices, that will be insanely priced.

[ViperDiscord]

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

[ViperDiscord]

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

[ViperDiscord]

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

[ViperDiscord]

seems Qwen has some guardrails as well.

[ViperDiscord]

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

[ViperDiscord]

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.