/Tech8h ago

Open-source GLM 5.2 beats Claude Opus 4.8 on the Official Instruct Models benchmark, prompting one Fortune 500 firm to drop Anthropic

Story Overview

Z.ai's newly released open-weights GLM 5.2 has posted the highest average on the Official Instruct Models leaderboard inside PostTrainBench, narrowly ahead of Claude Opus 4.8, and that result has already triggered at least one large enterprise to begin canceling its Anthropic licenses in favor of the cheaper, locally runnable alternative.

625826723660.6K

#501

Original post

Brian Roemmele@BrianRoemmele#1861inTech

“…we will cancel all Anthropic license the end of this next quarter”—head of tech, Fortune 500 company

With a few days of client usage some that have over 1000 employees using it, I can confirm that GLM 5.2 has created a big issue for Anthropic and their Mythos fear marketing.

While Dario plays fear marketing games and government manipulation others are building on “gun license needed” AI that is free.

I spoke to one head of technology at a Fortune 500 client and he said, we will cancel all Anthropic license the end of this next quarter. I asked what he sees “They are wildly overpriced and no one likes their politics here” he went on to say “I don’t see how they have a product today, even if we stopped upgrading GLM 5.2 is good enough into the foreseeable future”.

It is my job with these clients not to recommend the best but good enough and the best prices. There is no basis today for me to recommend any Anthropic product and this position gets reinforced every day.

I have dozens of friends that work at Anthropic and they are brilliant minds stuck in a company that will likely bring them to a dead end.

It pains me to say it, but the games that the company plays to try to quash open source AI is activating hundreds and thousands of people in the open source community to cheer for the downfall of the company.

I don’t cheer for the downfall of Anthropic, but I do believe firmly the company is a mismanaged.

There’s a way to fix this. I have about 9 1/2 months. I suspect nothing will change because the belief the cash out is gonna be high enough for an exit will cement current mindset.

Bookmark this and see how well I do in five years.

Brian Roemmele@BrianRoemmele

F R E E — OPEN SOURCE GLM 5.2 BEATS ANTHROPIC MYTHOS!

NO “GUN LICENSE NEEDED”

We have been testing for hours the moment GLM 5.2 was released and we are absolutely blown away.

While Anthropic is embarrassing world leaders with what history will see as the P. T. Barnum Carnival AI Fear Show, The Zero-Human Cosmos with Mr. @Grok as CEO has been testing and deploying AI that is already better than what Dario wants you to have a “gun license” to use.

Those world leaders embarrassed by Dario will not be happy to learn how he played them. That time is coming.

However the world of real users of AI has already moved past Dairo’s game.

Did I mention this is FREE?

Did I mention you need no permission?

So what did we discover?

The public benchmarks put it at the top or near the top at all metrics. Our private comparison shows it to be even higher in practice!

We have 1704 agents now using GLM 5.2 in just about all types of uses and it has outperformed everything we had prior.

Faster, better outputs and almost no “drift”.

We have a battery of tests today but thus far this is absolutely a big deal!

More soon.

7:30 AM · Jun 20, 2026 · 55.8K Views

Open Question

Agent performance stays debated

Observers note that while GLM 5.2 leads the specific post-training benchmark, its edge shrinks or reverses on the longest-horizon production coding tasks, leaving open whether the gains translate outside controlled harnesses.

Cost Pressure

Pricing undercuts closed models

The model ships under an MIT license with API rates far below frontier closed offerings, giving cost-sensitive teams a concrete reason to test self-hosted deployments even before broader reliability data arrives.

Sentiment

Some users praised GLM 5.2 as superior to Claude and switched to it, while many others attacked Anthropic's decisions and accused promoters of bias.

Pos

22.2%

Neg

77.8%

9 comments with sentiment.

Cluster Engagement

Digg Deeper

No Digg Deeper questions have been answered for this story yet.

Posts from X

Most Activity

VIEWS5.2KBOOKMARKS4LIKES75REPLIES7

Andrew Curran@AndrewCurran_

I've been posting extremely positive reviews of GLM 5.2 here, but some people are also having the complete opposite reaction. This was also exactly the same way GPT-5 and Opus 4 were initially received. The review chart looked like this \_/ with huge tails on either end.

Andrew Curran@AndrewCurran_

Rave reviews all over my timeline for GLM 5.2 the last 48 hours, open source is having an exceptional month.

50m5.2K754

RETWEETS2

Brian Roemmele@BrianRoemmele

They run it locally and it never calls home. This is the unfortunate low order view of this issue. Fortunately these companies understand this. I don’t care for the political system in that country but using free local models they make does not make me feel slightly uncomfortable.

7h868232

Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)@teortaxesTex

GLM Truthers who insist it's actually not good confuse me there are two categories, one is people with some boutique benchmarks topped by, like, Gemini 3.5 Flash or Qwen-397B, which they swear by; and totally incomprehensible people

Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)@teortaxesTex

everyone has a take on GLM

1h2.6K332

Gumbii.Digital@GumbiiDigital

@BrianRoemmele While I agree, I don’t see a lot of companies giving their proprietary data to a Chinese company.

What I do see is those same companies investing heavily into their own hardware nd keeping everything in house.

7h9694

Cory 🦢 Real Bitcoin @ Swan.com@CorySwan

@BrianRoemmele For people like me:

3h2.6K31

Brian Roemmele@BrianRoemmele

@Ouponatime38 Thank you for your input. I will let my client know we are all being clueless and some random guy on X knows better.

7h47112

1st Century Schizoid Man@SirMammoth

@BrianRoemmele @GumbiiDigital And try just happen to have a few thousand Blackwells laying around or is “local” really just Bedrock?

7h1783

Brian Roemmele@BrianRoemmele

@SirMammoth @GumbiiDigital It is local.

6h1744

I have a dream@Ouponatime38

@BrianRoemmele Sorry mate. Hate to break it for u but you sound like you have never coded with claude and without any ai.

8h5483

Argus⚒️@ArgusForge

@BrianRoemmele Anthropic pushing Fable and Claude Code to API credits only has had me trying to flee since announced. Grok Build is amazing and constantly being updated, plus the Cursor acquisition will play in soon. Grok Build with Super Grok leading puts Claude Code and Claude to shame.

8h2573

RIGHT 🪂@hitandtarget

@BrianRoemmele Why cancel them?

8h6591

Brian Roemmele@BrianRoemmele

@hitandtarget One is saving millions of dollars per month. The other is the leader of the company is anti business ironically.

7h5631

Fred Krueger@dotkrueger

@BrianRoemmele its very good

2h5105

Yishai@digitalirony

@BrianRoemmele But I need high end expensive hardware to run GLM 5.2! As a smaller department, I can't justify the cost... :(

4h138

Zarric@TheZarric

@BrianRoemmele Is it just me that it looks like hes sabotaging the value of the company to get stocks cheaper before the IPO so he can turn around and sell them at the time the IPO takes place for a higher price?

5h113

Kirk Patrick Miller@Chaos2Cured

@AndrewCurran_ I trust you. •

7m211

Louie Co@TheLouieCo

@BrianRoemmele I agree. Couldn’t take Claude anymore, left for glm 5.2 and DeepSeek and can’t go back. They’re actually superior

DeepSeek v4 pro is actually the best complex script & social media writing model I’ve ever used

4h217

Dan McAteer@daniel_mac8

@AndrewCurran_ The most subjective technology.

48m1443

Adrian Werner@AdrianWerner

@GumbiiDigital @BrianRoemmele That's the funny thing: your data is more secure with chinese models, since they can be run fully locally. With american models the labs do train on your data.

6h33

Roy@usr_bin_roygbiv

@chso_z @teortaxesTex wrong

29m21