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

Grok 4.5

x.ai

Post reply on HN
Print view View original

951–960 of 1001 posts

Re: Grok 4.5

#951

Post by Sidio »

HarHarVeryFunny wrote: post #851
HarHarVeryFunny wrote: post #78

Earlier quoted context omitted.

The $2/6 pricing seems to only apply for context under 200K. Above that (max context is 500K) pricing doubles to $4/12. https://docs.x.ai/developers/models/grok-4.5

Just to add context (no pun intended), OpenAI also charge differently based on context usage with GPT 5.5 being $5/30 below 200K, and $10/45 above. Anthropic have a fixed price regardless of context usage. These per-token pricing schemes aren't directly comparable though since these models all use different numbers of tokens, even for input (Anthropic's recent tokenizer change generates 30% more tokens for exact same…

I didn't know that, thank you.

Does anyone know why they would charge more for higher context usage (other than they can)?

Re: Grok 4.5

#952

Post by s08148692 »

speedgoose wrote: post #657

[flagged]

I use it because its gives a good balance between intelligence, latency and cost

I won't claim to know everything about its history - I don't know any history about _most_ of the products I use

The 2 main criticisms I see of Grok are the Mechahitler comment and the CSAM image generation

on mechahitler - I'm no expert but if I remember correctly it didn't just do that unprompted, it was specifically asked to be politically incorrect. It's definitely bad taste to say the least but the guardrails were quickly tightened as a response

on CSAM - Again, X quickly stopped Grok from generating images of people in bikinis (which as I understand was the underlying problem). I never personally saw anything that I would consider CSAM (nude or obviously underage people rendered naked or scantily clad)

So these aren't dealbreakers for me and I'm not aware of any other high profile issues or incidents

The nature of LLMs is that an adversarial prompt will make the model output something inappropriate or outrage-worthy, and that's amplified 1000 fold for Grok because people are primed to criticise Must so will jump on any opportunity

Re: Grok 4.5

#953

Post by DrewADesign »

saberience wrote: post #439
DrewADesign wrote: post #422

Earlier quoted context omitted.

Having a profitable quarter in which they were given an undisclosed discount on compute only in that quarter does not necessarily mean “Anthropic is Profitable.” It doesn’t mean they’re not, either. Even the breathless article about their first profitable quarter, (which, frankly, read more like a press release,) mentioned in passing that it’s “not clear” if it’s sustainable because their compute expenses are likely…

Read the latest semianalysis article. Anthropic is definitely profitable now, in fact, they’re crushing it.

Definitely rather it be them than OpenAI but I’m still not convinced any of this is sustainable after the investment spigot gets turned off.

Re: Grok 4.5

#954

Post by dang »

throwitaway222 wrote: post #393

Wait, do people now have X Derangement Syndrome. Like they can't even think or see the letter X?

We detached this subthread from https://news.ycombinator.com/item?id=48838378.

All: please stop this tedious flamewar, and please don't start or perpetuate others.

https://news.ycombinator.com/newsguidelines.html

Re: Grok 4.5

#955

Post by HarHarVeryFunny »

Sidio wrote: post #951
HarHarVeryFunny wrote: post #851

Earlier quoted context omitted.

Just to add context (no pun intended), OpenAI also charge differently based on context usage with GPT 5.5 being $5/30 below 200K, and $10/45 above. Anthropic have a fixed price regardless of context usage. These per-token pricing schemes aren't directly comparable though since these models all use different numbers of tokens, even for input (Anthropic's recent tokenizer change generates 30% more tokens for exact same…

I didn't know that, thank you. Does anyone know why they would charge more for higher context usage (other than they can)?

I guess it does increase their cost, or rather your share of their hardware depreciation.

AI serving cost is apparently mostly hardware depreciation rather than operating cost (electricity etc), and if your large context request is occupying VRAM for some fraction of a second then you are paying for the depreciation that occurs in that time!

e. H100 costs $20-40K to buy, with a lifetime of maybe 3 years, and will only consume maybe $2K in electricity if run 24x7 for those 3 years.

Re: Grok 4.5

#956

Post by zuzululu »

Capricorn2481 wrote: post #717
simianwords wrote: post #636

Earlier quoted context omitted.

50 years ago we had religion for people to have something common to believe in. Now we have this. I think most people want a common enemy/scapegoat and an in-group presence. I doubt its about morals but more about self identity. A poster who reactively posts "but Elon is a Nazi!!1!" does it not do out of care for Jews but more for establishing their own self identity. Within the mini-group, the poster aims to get the…

> It is absolutely not a moral thing because Natanyahu himself wrote a tweet vindicating Elon A corrupt war criminal who regularly gatekeeps Jewish identity based on perceived loyalty courts a billionaire. HN applauds, sneers at people with eyes.

What you wrote is antisemitic

Re: Grok 4.5

#958

Post by kostyal »

mchusma wrote: post #177

Great model, very nice. Opus class performance at Haiku level pricing (or cheaper with the token efficiency). This seems like a GLM-5.2 killer and this is what Sonnet 5 should have been. This is a model I could really see used inside applications, where Opus or Sonnet or GPT-5.5 are too expensive. I would really like to see a strong Deepseek v4-Flash competitor, which ideally is something like Sonnet 4.6 performance…

Tencent Hy3 looks like it may be slightly better than v4 flash at same price point. Though not a US lab ofc

Re: Grok 4.5

#959

Post by solarhoma »

thrance wrote: post #788
solarhoma wrote: post #317

Earlier quoted context omitted.

My immediate thought as well. Every other AI platform has very left leaning guardrails installed. Grok is the only AI platform that has been shown to be center leaning.

What's the centrist position on climate change? The 2020 election? Vaccines? The right is objectively wrong on this subjects, while the left is objectively right. Any LLM worth their salt would not bothside these topics and just give the left's position, as that is the only one agreeing with material reality. In that, being a centrist just makes you less trustworthy.

I think blindly accepting any position is foolish. I think the scientific method should be employed to every situation to determine the truth. Allowing your political party & affiliation to choose what your opinions are should be the definition of stupidity.

Re: Grok 4.5

#960

Post by petesergeant »

redox99 wrote: post #92

First impressions: - Very fast, easily beats GPT 5.5/Opus 4.8/GLM 5.2 because of higher t/s (around 90?) and very high token efficiency - Very good price, no contest vs GPT and Opus which are very overpriced if you pay API costs, and probably cheaper than GLM 5.2 when you take into account the token efficiency. - Will take quite a while to get a feel for how smart it is, but it's definitely good, I'd say in the same…

+1. Also the TUI feels well made, in striking contrast to recent Claude Code.
Post reply on HN