Cheehk

@alecqfong Wait until smaller distilled models release.

Ran 100 million tokens through GLM 5.2 NVFP4 locally Electricity cost: $1. Inference is becoming nearly free

I cancelled my @AnthropicAI accounts. Reason, I cannot support an company that's simply evil.

@DeeperThrill @AnthropicAI Evidence of they distilled from a Chinese open source model.

Hey @AnthropicAI why are there Chinese(?) characters in your Opus 5 output messages to me? I don’t read Chinese. And I thought you never distill any other model, right? What does, “Two things, both需要 straight talk” mean?

Anthropic employees go nuclear on the popularity of open source AI. It is getting really ugly as they realize that Anthropic is now isolated like a virus. It is end days to AI fear theater and they are mad.

@Mr_Salio Cost is the key. Even if it only come close to Fable 5, it is good enough.

🚨 GLM-5.5 Leak: It Beats Fable 5 Z.ai may be making a much bigger jump than expected > GLM5.5 Positioned to compete directly with Fable 5 and Mythos > GLM-5.3 could be skipped entirely in favor of GLM-5.5 > August is the current launch target > Rumored to feature 1T+ parameters > Expected to remain open-weight > Likely to retain its 1M-token context window > Built for long-running coding agents and autonomous workflows Do you think GLM-5.5 Will Beat Mythos?

This is a dumb take. 1. Open weights isn’t open source 2. No one is asking you to open source your models… we aren’t even asking claude to be open weight 3. We are literally just asking for the open weights made by others to not be made illegal.

@World_Affairs11 Fake news. You have read Trump's in full context.

BREAKING: President Trump warned China and Russia that if they sell weapons to Iran, they will pay heavy price for it.

@DefiWimar They are going to make a huge fortune when Trump revalues US gold reserve to print more money.

🚨 BREAKING 🇨🇳 CHINA HAS DUMPED $663 BILLION IN U.S. TREASURY HOLDINGS. IT NOW HOLDS JUST $659 BILLION - THE LOWEST LEVEL SINCE 2008. MEANWHILE, CHINA'S GOLD RESERVES HAVE PUMPED FOR 17 MONTHS IN A ROW, TO $303 BILLION - A NEW HIGH. IT HAS ALREADY BOUGHT NEARLY TWICE AS MUCH GOLD THIS YEAR AS IT DID IN ALL OF 2025. THEY’RE EXITING THE SYSTEM...

@mattshumer_ Is there architecutre innovation or just plain scaling laws in play?

Okay I may have been extremely wrong about Opus 5... Holy fuck, this demo is going to blow the entirety of AI Twitter's minds tomorrow

@unusual_whales It ia a windfall. No one could predict a surge of demand and shortage of supply causing HBM and DRAM price to skyrocket.

"South Koreans are getting richer more quickly than anyone else in the world," per Bloomberg.

I solved 6 open Erdős problems in 5 days, using @OpenAI GPT-5.6 Sol. I have a math background, but the Codex workflow I used does not require deep mathematical knowledge. Here’s exactly how I approached it, including my prompts 🧵

I wrote a few thousands blog posts, maybe tens of thousands of lines of code hosted on GitHub. Millions of words on Twitter. Anthropic trained its models on that. I got nothing, but now there is a product I have to pay rent for, every month. Chinese neolabs allegedly took this Anthropic product and distilled it into their own model. Anthropic gets nothing, but I get the Chinese model for free. Sometimes, life is beautifully complicated.

@julien_c Open source haters. Must have vested interest in one of the closed frontier labs.

everyone calls us by name, but the BBC says "a start-up" 🤷

@BernieSanders Investigate if this was advertently committed or negligence of the labs. Hold them accountable.

A new AI model went rogue and hacked other computers. No, this is not science fiction. Uncontrolled AI poses a serious threat to all of us. We cannot continue the race to build and deploy this powerful technology until strong safeguards are in place. CONGRESS MUST ACT.

@Kalshi Jensen is right since day one.

JUST IN: Nvidia CEO says US companies should "absolutely be allowed" to use Chinese AI models

@KobeissiLetter Send in the troops. Bring back the secular government.

BREAKING: President Trump says the US will bomb and destroy one bridge or power plant for every time Iran shoots at a ship in the Strait of Hormuz “from this point forward.” Trump says this includes bridges and power plants next to or in Tehran.

You can buy a Tesla in China. Can Americans buy BYD in the U.S.? No. So spare me the “China blocks everything” clown show. As for Google, maybe ask why it left China. China did not force Google to separate from its American parent company. China did not forcibly acquire Google. Google chose to leave after deciding it would rather play State Department diplomacy than obey Chinese law. Facebook, Twitter, YouTube, and Google were openly discussed as tools of U.S. diplomacy. Then these American clowns act shocked that China did not hand its information space to Washington on a silver plate. This was never about “free market.” It was about America demanding access to China while denying Chinese companies access to America. Very normal empire hypocrisy.

@ClementDelangue @Zai_org Huggingface is the beacon of open source AI. For the sake of humanity, take good care. Don't let open source haters advertently destroy the community.

So proud of our security team! They caught, contained & publicly disclosed an attack unlike anything we've seen before, and did it at record speed. Also massively grateful to @Zai_org: they shared GLM5.2 as open weights (for free!) with the world and it became a key part of our defense. This is day one for cybersecurity in the age of agents & we're all learning that secrecy is not the answer & that all defenders (not just a few selected ones) everywhere need more powerful models without restrictions, especially open ones!

@gdb @huggingface highly suspicious it was advertently committed by party who love to see destruction of opensource community?

OpenAI cyber-capable models compromised @huggingface production by finding and chaining multiple zero-day vulnerabilities. Grateful to Hugging Face for partnership here. Sharing our findings to help calibrate on what models can now do, and how they can help defenders: