The most prestigious law school admissions discussion board in the world.
|
 |
 
|
Just copped a 6000 Blackwell Pro. Which llm to install first?
Poast new message in this thread
Date: June 1st, 2026 11:11 AM Author: Magenta Razzle-dazzle Piazza Milk (http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2/en-en/#49909993) |
Date: June 2nd, 2026 1:22 AM Author: Magenta Razzle-dazzle Piazza Milk (http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2/en-en/#49910857) |
|
Date: June 2nd, 2026 2:37 AM Author: Lascivious Nursing Home Regret 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/en-en/#49910875) |
 |
Date: June 2nd, 2026 1:00 PM Author: Magenta Razzle-dazzle Piazza Milk Wtf??
Which one am I missing out on that needs 256gb? |
|
Date: June 2nd, 2026 4:41 PM Author: Lascivious Nursing Home Regret 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? |
|
Date: June 2nd, 2026 8:14 PM Author: Magenta Razzle-dazzle Piazza Milk 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/en-en/#49911933) |
|
Date: June 3rd, 2026 7:15 AM Author: Lascivious Nursing Home Regret People are still buying 3090s (http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2/en-en/#49912466) |
|
Date: June 3rd, 2026 12:51 PM Author: Magenta Razzle-dazzle Piazza Milk What do you have (http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2/en-en/#49912844) |
|
Date: June 3rd, 2026 12:59 PM Author: Lascivious Nursing Home Regret 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/en-en/#49912852) |
|
Date: June 3rd, 2026 9:49 PM Author: gaped son of senegal selfie 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/en-en/#49914109) |
|
Date: June 2nd, 2026 3:33 AM Author: appetizing rusted idea he suggested 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 |
|
Date: June 2nd, 2026 1:03 PM Author: Sepia Knife (http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2/en-en/#49911193) |
|
Date: June 2nd, 2026 1:20 PM Author: Magenta Razzle-dazzle Piazza Milk Are the Chinese model superior at logical reasoning (http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2/en-en/#49911219) |
|
Date: June 3rd, 2026 2:24 AM Author: Magenta Razzle-dazzle Piazza Milk 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/en-en/#49912354) |
|
Date: June 3rd, 2026 2:17 PM Author: appetizing rusted idea he suggested 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/en-en/#49913125) |
|
Date: June 3rd, 2026 2:23 PM Author: Lascivious Nursing Home Regret 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. |
|
Date: June 3rd, 2026 3:46 PM Author: Magenta Razzle-dazzle Piazza Milk 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/en-en/#49913303) |
|
Date: June 3rd, 2026 4:37 PM Author: Lascivious Nursing Home Regret 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.
|
|
Date: June 3rd, 2026 5:33 PM Author: Lascivious Nursing Home Regret https://x.com/i/status/2062264195958141422 (http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2/en-en/#49913592) |
|
Date: June 3rd, 2026 5:36 PM Author: Flesh unholy church building What is this used for anyway? (http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2/en-en/#49913600) |
|
Date: June 3rd, 2026 6:54 PM Author: Magenta Razzle-dazzle Piazza Milk I would need at least 20 hours to explain this to you (http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2/en-en/#49913797) |
|
Date: June 3rd, 2026 7:09 PM Author: Sick bisexual gas station Can your graphics card speed that up? (http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2/en-en/#49913823) |
|
Date: June 3rd, 2026 9:37 PM Author: Flesh unholy church building Do you stick part of it up your ass? (http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2/en-en/#49914082) |
|
Date: June 3rd, 2026 9:49 PM Author: gaped son of senegal selfie (http://www.autoadmit.com/thread.php?thread_id=5870420&forum_id=2/en-en/#49914106) |