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

Grok 4.3

docs.x.ai

Post reply on HN
Print view View original

281–290 of 608 posts

Re: Grok 4.3

#281

Post by accrual »

tecoholic wrote: post #190
Barbing wrote: post #175

Grok 4.3 was completed ahead of its CEO’s lesson on this common safety resource: Asked if he knew anything about OpenAI's "safety card," Musk smiled and replied: "Safety card? Why would it be a card?" https://www.axios.com/2026/04/30/musk-openai-safety-grok Low relevancy in spite of cluster size and musical chair gas generators for time being: Later in his testimony, Musk was asked about a claim he made last summer t…

Seriously though, why is it a model "card", safety "card"? I had to lookup to learn that it comes from HuggingFace's vague definition of "README" in the model's repo. This is such a specific thing that I don't think anyone except a very small population would know - not the users, not the c-suites. I don't like Musk or Grok. But not knowing what's a safety card is not a signal of anything IMO.

> Seriously though, why is it a model "card", safety "card"?

My assumption is because "card" has a more formal tone than a README, which is more like a quick "how to use the software" guide.

Collin's dictionary says about "cards":

> A card is a piece of stiff paper or thin cardboard on which something is written or printed. (1)

> A card is a piece of cardboard or plastic, or a small document, which shows information about you and which you carry with you, for example to prove your identity. (2)

> A card is a piece of thin cardboard carried by someone such as a business person in order to give to other people. A card shows the name, address, phone number, and other details of the person who carries it. (6)

Since companies spend a lot of resources training the model, and the model doesn't really change after release, I feel "card" is meant to give weight or heft to the discussion about the model.

It's not meant to be updated like a README or other software documents, it's meant to be handed out to others as a firm, unchanging "this is a summary of the model and its specifications", like a business card for models.

Re: Grok 4.3

#282

Post by WarmWash »

kardianos wrote: post #265
Barbing wrote: post #175

Grok 4.3 was completed ahead of its CEO’s lesson on this common safety resource: Asked if he knew anything about OpenAI's "safety card," Musk smiled and replied: "Safety card? Why would it be a card?" https://www.axios.com/2026/04/30/musk-openai-safety-grok Low relevancy in spite of cluster size and musical chair gas generators for time being: Later in his testimony, Musk was asked about a claim he made last summer t…

Elon has publicly stated that he cares a great deal about safety. He has stated that the only safe models are those which align greatest with truth, that which is in reality. In this, xAI has lived up, as it has proved to hallucinate least (or close to least) in benchmarks. If you read that, quote again, he is saying "how can you quantify safety in a card?"

The irony that the guy who lies incessantly for years now with empty promises about his businesses is most concerned with truth...

Re: Grok 4.3

#283

Post by Aurornis »

kardianos wrote: post #265
Barbing wrote: post #175

Grok 4.3 was completed ahead of its CEO’s lesson on this common safety resource: Asked if he knew anything about OpenAI's "safety card," Musk smiled and replied: "Safety card? Why would it be a card?" https://www.axios.com/2026/04/30/musk-openai-safety-grok Low relevancy in spite of cluster size and musical chair gas generators for time being: Later in his testimony, Musk was asked about a claim he made last summer t…

Elon has publicly stated that he cares a great deal about safety. He has stated that the only safe models are those which align greatest with truth, that which is in reality. In this, xAI has lived up, as it has proved to hallucinate least (or close to least) in benchmarks. If you read that, quote again, he is saying "how can you quantify safety in a card?"

> If you read that, quote again, he is saying "how can you quantify safety in a card?"

Everyone familiar with LLM research understands what is meant by “card”.

He was being obtuse to try to dodge the question and simultaneously give performance for his fans.

Re: Grok 4.3

#284

Post by bakies »

xiphias2 wrote: post #130

It's just at the Chinese levels for coding, so right now it's just a money earing thing for investors. I hope the Cursor guys help them catch up to be closer to frontier models because they badly need help in it.

I'm rooting for the china models so I can run it at home. Qwen is getting pretty good for how big it is. Idgaf about this asshole and his mechahitler.

Re: Grok 4.3

#285

Post by aembleton »

ToucanLoucan wrote: post #188
agrounds wrote: post #147

Earlier quoted context omitted.

>if an AI is confidently telling you something wrong it's hard to work with. But they all do that. It just comes with the territory. Grok will absolutely do the same thing another time you try it.

It is really, really genuinely concerning how many people think there are profound measurable differences between these things. Like yeah tonally I guess there are. But with regard to references and information? You’re literally just using three different slot machines and claiming one is hot. I suppose though I shouldn’t be that surprised then since Vegas and every other casino on Earth has been built on duping peop…

> You’re literally just using three different slot machines and claiming one is hot.

It's a fair point. I haven't tested many queries across them all and checked their answers, but if I want to ask one of them a question - right now its Grok just because I trust its answers more.

Re: Grok 4.3

#286

Post by WarmWash »

kardianos wrote: post #269
ezoe wrote: post #206

While the tread is swapping between "OMG Claude good. OpenAI was done for" and "OMG Codex good. Anthropic was done for". I've never heard about Gemini and Grok. It works mostly similar performance, but people don't mention that much. Still, my impression is, Gemini hallucinate too much while Grok is always less capable than competitors so it's not worth using it.

Gemini 2.5 and 3 can code, but they are also dumb. They don't model the world well. It's hard to use them for programming tasks. I haven't tried grok4.2 or grok4.3 yet for coding, but it wasn't up to the challenge as an agent yet. It looks like grok4.3 shifted its training and operates always as an agent first judging on some web usage. Musk knows grok is behind and states it publically. Now with grok4.3 release I do…

Gemini weakness is coding, but it will go toe to toe with 5.5 for science, (classic) engineering, finance, basically not programming stuff. It also does it while using about 1/4 the tokens.

Re: Grok 4.3

#287

Post by Scroll_Swe »

AntiUSAbah wrote: post #164
sundarurfriend wrote: post #27

As an English-as-second-language speaker and writer, one thing Grok really shines at is capturing the tone and level of "formality" of a piece of text and the replicating it correctly. It seems to understand the little human subtleties of language in a way the other major providers don't. Chatgpt goes overly stiff and formal sounding, or ends up in a weird "aye guvnor" type informal language (Claude is sometimes bett…

[flagged]

[flagged]

Re: Grok 4.3

#288

Post by vrganj »

gadders wrote: post #272
vrganj wrote: post #258

Earlier quoted context omitted.

What's liberal identity politics have to do with leftism? Liberalism is a center-right ideology. Us leftists are concerned with class issues, not identity issues. Focusing on identity is nothing but a way to distract from class.

Lol. Gender ideology is very much a policy of left wing parties. You may go for the One True Scotsman argument and say it's not proper leftism, and you may be right, but that doesn't stop it being policy.

You think Lenin was into gender issues? You think Lenin wasn't a leftist?

Re: Grok 4.3

#289

Post by aembleton »

agrounds wrote: post #147
aembleton wrote: post #135

Earlier quoted context omitted.

I've tried Grok, Gemini and ChatGPT. There have been 2 times now where Gemini and ChatGPT confidently gave me an incorrect answer whereas Grok was correct. I'm now paying for Grok Lite or whatever it is $10 plan. The first question was around setting up timers for a Fox ESS battery in Home Assistant and disconnecting Fox ESS from the cloud. The second was around cornering speed in Sunnypilot and Frogpilot. Somewhat n…

>if an AI is confidently telling you something wrong it's hard to work with. But they all do that. It just comes with the territory. Grok will absolutely do the same thing another time you try it.

> Grok will absolutely do the same thing another time you try it.

True; it's just not happened yet. It will at some point though. With the Sunnypilot example it right out told me that it is not possible on that fork which I appreciated. The others all seem to hallucinate some setting.

Re: Grok 4.3

#290

Post by 2ndorderthought »

derangedHorse wrote: post #231
2ndorderthought wrote: post #117

Earlier quoted context omitted.

[flagged]

> so of course grok is fine talking about it. Probably even offers strategies for them does free accounting has money laundering strategies etc... The slander comes in when you assume Elon knew and was complicit with their crimes to the point he'd intentionally normalize it as a discussion topic in Grok. You even went so far as to say it's willing to assist in committing crimes.

He is aware of the csam generation. He blamed the users and the official stance from his team was not to offer any fixes. That is the last I heard.

https://arstechnica.com/tech-policy/2026/01/x-blames-users-f...

I do not see the slander. These are his viewpoints. He says him, grok, and his team aren't responsible for what users do. Other companies, countries and people feel differently about the responsibility for AI models generating csam for money.

Grok and xais depictions of it are that it isn't woke and is maximally based and is politically incorrect by design. So yes, chosing to avoid being correct about policies like laws and avoid social norms lead me to believe that the generation of hate speech(some of which was illegal in certain localities), csam, etc are an expected outcome. Like Elon musk said, it's the users fault not groks. So I would not be surprised if it offered other illegal advice or helped criminals forward criminal activities. Especially more than has already been reported.

Here are some of the crimes that grok is being implicated in as far as I know today: https://www.irishtimes.com/crime-law/2026/03/03/number-of-ga...

https://www.france24.com/en/europe/20251121-france-to-invest...

https://www.robertkinglawfirm.com/mass-torts/grok-lawsuit/

https://news.bloomberglaw.com/litigation/grok-maker-xai-face...

https://www.msn.com/en-us/news/technology/musk-testifies-xai...

Among others.

I don't see that as slanderous. I see it as factual and an expected outcome for the stated goals of the product and the responses to the outcomes of the product itself by the company and its leadership.

I legitimately do expect there to be more lawsuits and possibly criminal persecution against musk, xai, over grok and no I would not be surprised if the tool is currently being used for more crime. Especially given the response to the sexual crime allegations that have been made.

I don't think Elon personally intends to normalize this. But I think that may happen anyways because I think the response was too soft.

Yes I do think grok can be used to aid crimes and criminal activity like the many lawsuits and journalists currently suggest. I don't think grok is "willing" it's not a person. I know it currently has been implicated in generating material leading to the arrests of individuals. Which I would be very surprised if that was legal.

https://factually.co/fact-checks/technology/grok-created-ill...

Post reply on HN