News for hackers. Stuff that matters.
Re:DeepSeek v4 by applfanboysbgon on Friday April 24, 2026 @12:33AM (#47885552)
No, but businesses do. Being able to run quality LLMs without your business, or business's private information, being held at the mercy of another corp has a lot of value.
Re:DeepSeek v4 by forrestthewoods on Friday April 24, 2026 @12:52AM (#47885699)
What type of system is needed to self host this? How much would it cost?
Re:DeepSeek v4 by disiplus on Friday April 24, 2026 @01:06AM (#47885778)
Depends how many users you have and what is "production grade" for you but like 500k gets you a 8x B200 machine.
Re:DeepSeek v4 by p1esk on Friday April 24, 2026 @01:01AM (#47885742)
Depends on fast you want it to be. I’m guessing a couple of $10k mac studio boxes could run it, but probably not fast enough to enjoy using it.
Re:DeepSeek v4 by fragmede on Friday April 24, 2026 @01:19AM (#47885881)
One GB200 NVL72 from Nvidia would do it. $2-3 million, or so. If you're a corporation, say Walmart or PayPal, that's not out of the question.

If you want to go budget corporate, 7 x H200 is just barely going to run it, but all in, $300k ought to do it.

Re:DeepSeek v4 by gloflo on Friday April 24, 2026 @01:24AM (#47885922)
How many users can you serve with that?
Re:DeepSeek v4 by fragmede on Friday April 24, 2026 @02:06AM (#47886201)
For the H200, between 150-700. The GB200 gets you something like 2-10k users.
Re:DeepSeek v4 by forrestthewoods on Friday April 24, 2026 @05:37PM (#47896119)
Whoa. How on earth can one system serve 2000 potentially concurrent users?
Re:DeepSeek v4 by CamperBob2 on Friday April 24, 2026 @02:22AM (#47886317)
$20K worth of RTX 6000 Blackwell cards should let you run the Flash version of the model.
Re:DeepSeek v4 by choldstare on Friday April 24, 2026 @12:42AM (#47885628)
Not really - on prem llm hosting is extremely labor and capital intensive
Re:DeepSeek v4 by applfanboysbgon on Friday April 24, 2026 @12:47AM (#47885661)
But can be, and is, done. I work for a bootstrapped startup that hosts a DeepSeek v3 retrain on our own GPUs. We are highly profitable. We're certainly not the only ones in the space, as I'm personally aware of several other startups hosting their own GLM or DeepSeek models.
Re:DeepSeek v4 by wuschel on Friday April 24, 2026 @02:42AM (#47886473)
Why a retrain? What are you using the model for?