@hatsunama: Best new local Ai . Works all the way down to 12gb!! #ai #news #uncensored #qwen #local

jeebs :

Man i stg, Ive been building a qLora model with RAG for a security harness/tools etc. now I gotta pull that lol

Jay :

Hi. I’m going back to school soon for network engineering and just recently started getting back into the tech space. I see everyone talking about the new qwen 3.8. Only have 16gb vram and 32gb system ram, could you recommend a couple models While I plot out my upgrades😅

CoveredInBlood :

why abliterated for cybersecurity? thats kinda NOT good.

GermanGuy :

is this the jailbroken one?

Deus :

Wish I had the hardware to run this

QuantumTruth :

definitely one to add to the stack 👍

Velatria + Kairo 🖤🔥💙 :

I just tuned Qwen3.5:27B. Tech moves too fast. 😂😭😂

OpenclawGateway :

Just give me an Abliterated model and I'll finally feel like i can sleep at night.

Scroatal :

no way it works on 12gh of vram

RUMP42 :

Damn! Thanks brother! I was just thinking this is what I needed. Awesome

Smitty :

thanks Hatsu. would you mind advising system requirements to run something like that locally?

mordwyn :

Blackfrost is a relatively new contributer to the space, nonetheless, I immediately downloaded the ~55 GB of 16 bit safetensors and several quants. That said, I'd believe more in pliny's obliterations (or use his obliteratus project). I'm testing orca's obliteration. It's an art sometimes to get right. I'd trust, download, back up off line, and test, configuring with a list of model prompts. 🙏

🐦‍⬛🖤 Rook :