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

111–120 of 476 posts

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

#111

Post by wongarsu »

CubsFan1060 wrote: post #107
wongarsu wrote: post #88

Earlier quoted context omitted.

I know of multiple businesses in Europe that have been doing that for a while with 70B models, and are upgrading hardware to run the new crop of 700B-1T models (really started around Kimi K2, but buying and hosting that kind of hardware takes time) Not everyone is willing (or even legally able) to send their trade secrets to OpenAI or Anthropic

What kind of hardware/price does it take to run those?

For an 8-bit quant (what people call "near lossless") you are looking at something like 4xMI350X, which comes out to about $150k after adding the rest of the server. More if you go with Nvidia instead of AMD. More if you want more than maybe 8x concurrency

But prices are changing rapidly, and not for the better

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

#112

Post by giancarlostoro »

vorticalbox wrote: post #29
Tiberium wrote: post #2

It seems to really be a nice step-up and is getting quite close to the frontier. I wish they'd start focusing on the reasoning efficiency now, though. I have a simple (relatively) test task to evaluate LLMs: writing a simple math evaluator library in Nim (it's about 400-600 lines total max), and GLM 5.2 (xhigh which maps to max effort) spent over 15 minutes (!) reasoning, spending about 45k tokens, before it finally…

This is a problem I find with opus is will spend so long thinking then going “but wait what if” To point where I stop it and simple tell it to “start writing code you can work it out as you go along” Seems writers block also effects LLM

I usually have Claude build a plan first, then I put it into an XML file it updates with phases, usually we talk about some of those tasks, and then once its good and I like it, I have Claude implement the plan.

Another thing I tell Claude to do is to not guess, but look at documentation, it messes up a lot less, might use some tokens reading docs, but at least it has a higher success rate code wise.

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

#113

Post by alecco »

kristopolous wrote: post #104

I have a script that ranks these based on codingindex from Artificial Analysis. All it does is pull a json from their main table page and parses it with the fields I care about (coding). There used to be a mailing list associated with it but eh ... there wasn't much interest. I use the script every day though. Current partial output score age size name 47.1 58 large Kimi K2.6 47.5 54 large DeepSeek V4 Pro (Reasoning,…

Consider using decrementing score order (best on top)

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

#114

Post by alecco »

cedws wrote: post #28
CuriouslyC wrote: post #11

Earlier quoted context omitted.

Be careful about unofficial providers, a lot of them misconfigure models or stealth quantize them. For a while the difference between Kimi on the official API and most third party providers was 20-40%.

OpenRouter should be penalising or banning for this.

Would that align with their VC-backed incentives?

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

#115

Post by slig »

kristopolous wrote: post #104

I have a script that ranks these based on codingindex from Artificial Analysis. All it does is pull a json from their main table page and parses it with the fields I care about (coding). There used to be a mailing list associated with it but eh ... there wasn't much interest. I use the script every day though. Current partial output score age size name 47.1 58 large Kimi K2.6 47.5 54 large DeepSeek V4 Pro (Reasoning,…

Thanks for sharing. I'm curious: why didn't you sort with the score descending?

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

#116

Post by _pdp_ »

simonw wrote: post #110

I was surprised that GLM 5.1/5.2 are not vision models - they are text input only. That's actually pretty uncommon these days. All of the OpenAI/Anthropic/Gemini models accept images, and so do the other leading open weight families - Gemma 4, Qwen 3.6, Kimi 2.x. In GLM's case image input would be useful because it's a model that scores very highly for tasks like web design, but without image input it can't take a sc…

I don't see this being such a big gap. There are some use-cases for sure but apart from UX/UI work it is not really needed. Besides, none of the frontier models can replicate actual images - the can approximate at least in my own experience.

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

#117

Post by granra »

eckelhesten wrote: post #91

Sure, but whatever you do, don't buy their (Z.ai) lite plan. I feel like i threw 15 dollars in the sea. I'm getting rate limited after 3-4 prompts. You get way less value than just paying 25 dollars for Claude or OpenAI models.

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, wouldn't it make sense to instead do z? What would that look like?

-

- nice, implement this

- starts writing files, run tests, etc.

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

#119

Post by stymaar »

simianwords wrote: post #38
xiaoyu2006 wrote: post #36

Earlier quoted context omitted.

Some models are heavily subsidized. Total params & active params are better measurement of inference cost.

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 for the same token), but lose money on actual compute done.

> there are lots of third party hosting services that will still run at breakeven/profit.

How can you be sure that they are making profit directly from token price, and are not billing at marginal cost (i.e. electricity price, without counting the cost of the GPUs) and aiming to make a profit later on from the valuable training data that they are collecting in the process?

Post reply on HN