\
  The most prestigious law school admissions discussion board in the world.
BackRefresh Options Favorite

Just copped a 6000 Blackwell Pro. Which llm to install first?

...
unhinged senate headpube
  06/01/26
...
unhinged senate headpube
  06/02/26
lol there's nothing better than Qwen3.6 27b that will fit on...
soggy den degenerate
  06/02/26
Wtf?? Which one am I missing out on that needs 256gb?
unhinged senate headpube
  06/02/26
https://huggingface.co/antirez/deepseek-v4-gguf https://h...
soggy den degenerate
  06/02/26
Are you supposed to get a Mac ultra to run those? I read tha...
unhinged senate headpube
  06/02/26
People are still buying 3090s
soggy den degenerate
  06/03/26
What do you have
unhinged senate headpube
  06/03/26
On one server I have a 5090, a 5080, a 5070 ti, and a 5060 t...
soggy den degenerate
  06/03/26
just do your own paste my man, it's incredibly easy if you k...
Irate Garnet Property Water Buffalo
  06/03/26
take ur pick from any chinese model that fits from https://w...
hideous gay wizard sanctuary
  06/02/26
...
Fragrant Insanely Creepy Menage Roast Beef
  06/02/26
Are the Chinese model superior at logical reasoning
unhinged senate headpube
  06/02/26
Should I keep this or sell it and get a 5090 or M4 ultra? Lo...
unhinged senate headpube
  06/03/26
there might be some M4s available for cheap soon when the M5...
hideous gay wizard sanctuary
  06/03/26
Can you even buy M4 Ultras? I've been looking at Mac Studios...
soggy den degenerate
  06/03/26
GPT/claude said Macs can run larger models but would be so s...
unhinged senate headpube
  06/03/26
Right. I think the ideal Mac has to be the Studio with an Ul...
soggy den degenerate
  06/03/26
https://x.com/i/status/2062264195958141422
soggy den degenerate
  06/03/26
What is this used for anyway?
Misunderstood Cyan Deer Antler Electric Furnace
  06/03/26
I would need at least 20 hours to explain this to you
unhinged senate headpube
  06/03/26
Can your graphics card speed that up?
Dun principal's office
  06/03/26
Do you stick part of it up your ass?
Misunderstood Cyan Deer Antler Electric Furnace
  06/03/26
...
Irate Garnet Property Water Buffalo
  06/03/26


Poast new message in this thread



Reply Favorite

Date: June 1st, 2026 11:11 AM
Author: unhinged senate headpube



(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49909993)



Reply Favorite

Date: June 2nd, 2026 1:22 AM
Author: unhinged senate headpube



(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49910857)



Reply Favorite

Date: June 2nd, 2026 2:37 AM
Author: soggy den degenerate

lol there's nothing better than Qwen3.6 27b that will fit on that card. You won't even use half your VRAM, but anything better requires more than 96gb. You either need 48gb or 256gb, and nothing in between really counts

(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49910875)



Reply Favorite

Date: June 2nd, 2026 1:00 PM
Author: unhinged senate headpube

Wtf??

Which one am I missing out on that needs 256gb?

(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49911186)



Reply Favorite

Date: June 2nd, 2026 4:41 PM
Author: soggy den degenerate

https://huggingface.co/antirez/deepseek-v4-gguf

https://huggingface.co/unsloth/Step-3.7-Flash-GGUF

You could run the second one at Q2 or Q3 but what's the point?

(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49911462)



Reply Favorite

Date: June 2nd, 2026 8:14 PM
Author: unhinged senate headpube

Are you supposed to get a Mac ultra to run those? I read that Mac m4 with max ram is still slow as fuck vs nvidia

(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49911933)



Reply Favorite

Date: June 3rd, 2026 7:15 AM
Author: soggy den degenerate

People are still buying 3090s

(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49912466)



Reply Favorite

Date: June 3rd, 2026 12:51 PM
Author: unhinged senate headpube

What do you have

(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49912844)



Reply Favorite

Date: June 3rd, 2026 12:59 PM
Author: soggy den degenerate

On one server I have a 5090, a 5080, a 5070 ti, and a 5060 ti, but I'm only using the 5090 and 5080 to run my Hermes Agent right now, because I only need 48gb to run Qwen3.6 27b at Q8. I have two 3090s sitting in another server and I use them to do OCR, translation, image gen and shit like that. I just have a bunch of LXC containers running Ollama on the 3090s, and my Hermes Agent connects to whatever Ollama container it needs for a given task. I used to run a separate model for coding tasks, but Qwen3.6 does that well enough now. I only paid about $700 apiece for the 3090s plus another $100 to get them re-pasted by a shop, now they go for like $1200 on ebay with 6-year old paste lol

(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49912852)



Reply Favorite

Date: June 3rd, 2026 9:49 PM
Author: Irate Garnet Property Water Buffalo

just do your own paste my man, it's incredibly easy if you know what you're doing

(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49914109)



Reply Favorite

Date: June 2nd, 2026 3:33 AM
Author: hideous gay wizard sanctuary

take ur pick from any chinese model that fits from https://willitrunai.com/

American OSS models in that range suck, I haven't tried the newer Mistral models but doubt they're competitive either

(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49910878)



Reply Favorite

Date: June 2nd, 2026 1:03 PM
Author: Fragrant Insanely Creepy Menage Roast Beef



(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49911193)



Reply Favorite

Date: June 2nd, 2026 1:20 PM
Author: unhinged senate headpube

Are the Chinese model superior at logical reasoning

(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49911219)



Reply Favorite

Date: June 3rd, 2026 2:24 AM
Author: unhinged senate headpube

Should I keep this or sell it and get a 5090 or M4 ultra? Looks like the price went up $2k again over the weekend

(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49912354)



Reply Favorite

Date: June 3rd, 2026 2:17 PM
Author: hideous gay wizard sanctuary

there might be some M4s available for cheap soon when the M5s get announced and all the openclaw retards want to upgrade their "secure agent hosts"

(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49913125)



Reply Favorite

Date: June 3rd, 2026 2:23 PM
Author: soggy den degenerate

Can you even buy M4 Ultras? I've been looking at Mac Studios for a few weeks, and they only let you spec it with a M3 Ultra or M4 Pro. And they won't let you add more than 96gb of RAM (which to be fair is plenty).

Anyway the Ultra models are never getting cheaper. The M5 is the same design as the M4 and M3, just made using a newer process at the TSMC foundry. You get insane memory bandwidth with all of them, it's just a matter of how many GPU cores you have to crunch on it. A M2 Ultra probably beats M5 Pro for AI just because it has more GPU cores.

(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49913147)



Reply Favorite

Date: June 3rd, 2026 3:46 PM
Author: unhinged senate headpube

GPT/claude said Macs can run larger models but would be so slow that it would practically be unusable. Like waiting an hour for a response or 8 hours to do a full brief draft

(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49913303)



Reply Favorite

Date: June 3rd, 2026 4:37 PM
Author: soggy den degenerate

Right. I think the ideal Mac has to be the Studio with an Ultra -series and 64gb VRAM. That lets you use 48gb comfortably with room to spare. I wouldn't plan on using more than 48gb with a Mac and I can only see 96gb being useful if you want to have two models running concurrently. What I like about my GPU spread is the ease of serving up multiple LLMs concurrently l. My OCR -> vector embedding workflow is pretty ridic, and I have a server that translates media from any language and produces an English transcript.



(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49913438)



Reply Favorite

Date: June 3rd, 2026 5:33 PM
Author: soggy den degenerate

https://x.com/i/status/2062264195958141422

(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49913592)



Reply Favorite

Date: June 3rd, 2026 5:36 PM
Author: Misunderstood Cyan Deer Antler Electric Furnace

What is this used for anyway?

(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49913600)



Reply Favorite

Date: June 3rd, 2026 6:54 PM
Author: unhinged senate headpube

I would need at least 20 hours to explain this to you

(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49913797)



Reply Favorite

Date: June 3rd, 2026 7:09 PM
Author: Dun principal's office

Can your graphics card speed that up?

(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49913823)



Reply Favorite

Date: June 3rd, 2026 9:37 PM
Author: Misunderstood Cyan Deer Antler Electric Furnace

Do you stick part of it up your ass?

(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49914082)



Reply Favorite

Date: June 3rd, 2026 9:49 PM
Author: Irate Garnet Property Water Buffalo



(http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2:#49914106)