Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

GLM-5.2 is the new leading open weights model on Artificial Analysis

artificialanalysis.ai

Post reply on HN
Print view View original

311–320 of 476 posts

Re: GLM-5.2 is the new leading open weights model on Artificial Analysis

#311

Post by zozbot234 »

program_whiz wrote: post #263
bitmasher9 wrote: post #121

Earlier quoted context omitted.

Nvidia will sell you an entire server rack ready for inference. Or maybe you can roll out your own Blackwell based system. We’re approaching a world where running a primer frontier model is possible on a workstation, probably will have something under $30k that looks like a desktop for Nvidia’s next generation. It sounds expensive, until you look at your Anthropic bill. It’s similar unit economics as could computing…

"approaching" is doing some work there. $30K today will get you 90-144GB usable VRAM with solid system RAM and disk and CPU. A single B200 chip at 180GB is $40K. Unfortunately that is nowhere close to being able to run a 750B param model. For something like that, we're getting closer to 1TB VRAM (8+ H200/B200), and then 1M context KV cache is many more GBs on top of that. That's a $500K-$1M+ rig as of now. That's a l…

> Unfortunately that is nowhere close to being able to run a 750B param model. For something like that, we're getting closer to 1TB VRAM

You don't have to run a model from VRAM, or even from a sizeable amount of RAM. These choices only ever make sense when serving the model at scale, to hundreds of simultaneous users or more.

Re: GLM-5.2 is the new leading open weights model on Artificial Analysis

#312

Post by granra »

eckelhesten wrote: post #153
granra wrote: post #117

Earlier quoted context omitted.

How are you using it? I have the lite plan and I've only ever maxed my weekly usage a few hours before reset. I will concede that I'm not a super heavy LLM user but it's been really good for me. My workflow is usually: - read file. I want to achieve X, how do? Do not implement anything. - I would do a, b and c - sketch a brief implementation of your suggestion - (not writing files yet) - instead of your approach x, w…

Try pointing it to a small codebase, or even ask it to conjure information found online. You'll see that it quickly gives up. Thing is, they seem to count cached hits as if they were the non-cached tokens. I wont be subscribing again thats for sure. I am not paying iPhone money for a Xiaomi.

That's what I've been doing. I use crush normally. While the codebase are by no means huge, they're not tiny either.

Re: GLM-5.2 is the new leading open weights model on Artificial Analysis

#314

Post by phainopepla2 »

johnnyApplePRNG wrote: post #280
cromka wrote: post #258

Earlier quoted context omitted.

> it's significant onboarding friction. It's crazy that apparently writing software without knowing how to edit a single config file is normal now.

It's crazy that apparently doing math without knowing how to do long division by hand is normal now.

Absolutely ludicrous comparison

Re: GLM-5.2 is the new leading open weights model on Artificial Analysis

#315

Post by stymaar »

simianwords wrote: post #309
stymaar wrote: post #296

Earlier quoted context omitted.

> Or you can simply believe the intuition Sure, you can believe you intuition as much as you want, but telling strangers over the internet that they are wrong because “I trust my intuition” is… awkward.

At some point it does come to intuition. Even if the companies IPO and share their financials, you can always argue that they might be lying.

Again, there's a difference between relying on intuition in your life (which we all do, lacking perfect information that would allow us to avoid relying on it), and telling someone they are wrong because your intuition says so.

Re: GLM-5.2 is the new leading open weights model on Artificial Analysis

#316

Post by gertlabs »

GLM 5.2 is the first model we've tested that is unambiguously on par with, or better than Opus 4.6 (although as usual, we have GLM 5.2 and most other Chinese models a bit below most other benchmarks with more vulnerable test methodologies).

Data at https://gertlabs.com/rankings

Re: GLM-5.2 is the new leading open weights model on Artificial Analysis

#317

Post by Retro_Dev »

adithyaharish wrote: post #313

why do not all open source LLM's have open weights like this model?

"open source" means that the code itself (for LLMs - this is training code) is available to the general public. "open weights" means that the weights (trained over time) are available publicly, rather than locked behind a paywalled chat. I do not know of an open source LLM that is not also open weights (unless they never bothered training it). Models like Claude and Gemini are neither open source, nor are they open weights.

Re: GLM-5.2 is the new leading open weights model on Artificial Analysis

#318

Post by drob518 »

tekchip wrote: post #194
kristopolous wrote: post #162

Earlier quoted context omitted.

I'm actually more curious about IBM. Their granite series appears to be nowhere close to competitive. They had Watson, remember, it won on jeopardy like 15 years ago? They've been at this for a long time Maybe it's good at something else?

IBM doesn't do technology they do contracts. Any "technology" is marketing stunts. They hire a bunch of "fellows" outside contractors to make a thing they can be first at or whatever, do the stunt, then get a bunch of 5-10 year contracts with customers off the stunt. They then fuck it up for that length of time but still get paid due to those contracts. After that space of time the folks theyve burned have moved on,…

Yes, but the key point is that nobody got fired for buying it from IBM.

Re: GLM-5.2 is the new leading open weights model on Artificial Analysis

#319

Post by jgalt212 »

stymaar wrote: post #119
simianwords wrote: post #38

Earlier quoted context omitted.

No models are subsidised -- there are lots of third party hosting services that will still run at breakeven/profit. (except Deepseek after discount)

> No models are subsidised We have no proof in either direction, it's not like we had access to their financial numbers in details. And the pricing itself muddies the water, as input tokens that are already in the KV cache are practically free for the provider, whereas other tokens are expensive. So they could still make money overall thanks to people having multi-turn conversation (and as such, paying multiple times…

> as input tokens that are already in the KV cache are practically free for the provider,

not at today's RAM prices.

Post reply on HN