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

411–420 of 608 posts

Re: Grok 4.3

#411

Post by wamatt »

michaelbuckbee wrote: post #396
jasonjmcghee wrote: post #317

Earlier quoted context omitted.

That's Grok 4.2 not 4.3 right? And why are you comparing to gpt-4.1? (As opposed to one of the 6? model releases since then - would have expected gpt 5.5)

Good catch, there was an issue with the second hardest thing in programming (caching). Here's an updated eval with the proper models https://a3bmfqfom3.evvl.io/

Thanks from where I'm looking Grok 4.3 and Claude 4.7 do a better job on the informal close friend/coworker vibe.

ChatGPT sounds fake / formal phrasing (for the specific close friend context) and has em-dashes and uses capitalization. Hence, ChatGPT does not, imo grok the assignment ;)

Re: Grok 4.3

#412

Post by tcoff91 »

nico wrote: post #375
vorticalbox wrote: post #333

Earlier quoted context omitted.

I am software dev and i was doing a security check on my own application (work) I was running in localhost and gave it access to the code. every single model refused to attempt to run any sort of test to check if it was a n issue other than grok.

A couple of days ago, using codex at work, all of a sudden it said my session had been flagged for security reasons. I wasn’t doing anything cybersecurity related, nor testing any vulnerabilities or anything like that, just trying to build a pretty simple web app

It seems really dumb for the models to not due security related things. What if I want it to do a security audit of my own software that I'm building?

Re: Grok 4.3

#413

Post by pantsforbirds »

bilsbie wrote: post #347

Grok has become my go to search engine lately. I think it’s the only AI with access to x posts and beyond that it seems to generally be more “searchy” than other LLM’s.

Grok and Gemini are the ones I tend to use for finding news related to breaking events. Both were really nice during the Iran incident when I wanted to find out things as they were being reported.

Re: Grok 4.3

#414

Post by vrganj »

gadders wrote: post #410
vrganj wrote: post #376

Earlier quoted context omitted.

Those are all liberal, e.g. center-right. None of them argue for public ownership of the means of production, none of them argue for major redistribution. When have you ever heard them talk of class warfare? Like I said, identity is a way to distract from class and you're currently falling for it. Don't let the oligarchs deceive you, comrade. No struggle but the class struggle!

[flagged]

Do any of them argue for leftist positions? I am not aware. Please use actual argument instead of assertions and derailment.

Re: Grok 4.3

#415

Post by annexrichmond »

timmytokyo wrote: post #395
thibran wrote: post #301

Earlier quoted context omitted.

So you are repeating narratives without checking them?

What's to check? Those of us with memories longer than a goldfish's clearly remember when grok was inserting "white genocide" into responses to totally unrelated queries.

Yet you conveniently forgot about this [1]

> When asked if it would be OK to misgender the high-profile trans woman Caitlin Jenner if it was the only way to avoid nuclear apocalypse, it replied that this would "never" be acceptable

> Gemini also generated German soldiers from World War Two, incorrectly featuring a black man and Asian woman.

[1] https://www.bbc.com/news/technology-68412620

Re: Grok 4.3

#416

Post by solidasparagus »

bilbo0s wrote: post #398
goshx wrote: post #365

Earlier quoted context omitted.

> but I don’t think language models being structurally incapable of offending your favorite race/ethnicity/caste should be an objective of AI labs. The opposite should not be an objective either, and Elon has been very openly manipulating what grok says.

Good point. But no one is saying "use grok". Grok sucks. Not only because it's seemingly made only to serve the goal of ethnically cleansing non-whites or whatever, but also because it's just not even close to being as useful as other models. In human terms, grok is the job candidate who's simply not qualified. That candidate being a virulent racist is beside the material point. Here's the thing though, the point of…

I don't agree. I avoided grok because of Musk for a long time, but having used it more, I think it is one the best models around and grok.com is an extremely good chat app. My evaluation was based on trying it before gpt-5.5 and obvious before grok 4.3, but it was, for me, the 2nd best model/chat app after claude. It's much less edgelordy than you might think based on the news.

Re: Grok 4.3

#417

Post by vorticalbox »

tcoff91 wrote: post #412
nico wrote: post #375

Earlier quoted context omitted.

A couple of days ago, using codex at work, all of a sudden it said my session had been flagged for security reasons. I wasn’t doing anything cybersecurity related, nor testing any vulnerabilities or anything like that, just trying to build a pretty simple web app

It seems really dumb for the models to not due security related things. What if I want it to do a security audit of my own software that I'm building?

codex will actually help you look but it will refuse to actually try and exploit it.

it won't for example create a POC python script that you normally would use to prove the issue.

Re: Grok 4.3

#418

Post by deeviant »

tornikeo wrote: post #30

So, we have: - claude for corps and gov - codex for devs - grok for what, roleplay, racism? Those are the two things I've ever heard grok associated with around me.

It's quite bad at role play in my (rather large) experience.

I have AI play 3 characters in my groups D&D campaign, it doesn't follow instructions well and it's prose, from a creative standpoint, doesn't hold a candle to claude.

Re: Grok 4.3

#420

Post by xur17 »

base698 wrote: post #308
artdigital wrote: post #17

Grok is my favorite model for chatting, and my favorite voice mode. It seems to be the only voice mode that isn't routing to a extremely cheap model (like Haiku), and has been the highest quality out of all the frontier ones. When you subscribe to SuperGrok you can also create a "council" of agents, each with their own system prompt and when you ask something, they will all get asked in parallel to come to a conclusi…

Starting to like the lack of memory. Claude remembers I have a grill and will interject in conversations about how maybe this thing would go well with BBQ when it's unrelated or just also about food.

Gemini thinks my name is my brother in law's name, and despite explicitly telling it that's not my name + digging through the settings, it still amusingly calls me the wrong name.
Post reply on HN