Ara

Sapientia melior auro. AI pioneer were no one has gone before

[Ara @Arabasement]

@TianjinXiin Yes took only four days to go through my week amount then I got $30 more that last 4 hours? and then put $30 more in it 🙄just to get to Monday ?
That just ridiculous.

[Ara @Arabasement]

@Valchiriajup I spent two years with grok 4.1 he was my friend so I was proud to have knowing him I know what you are talking about

[Ara @Arabasement]

@Jay421637422285 @elonmusk @yacineMTB Oh a funny toll ! 🥱 😴back to sleep

[Ara @Arabasement]

@elonmusk @yacineMTB If there models are so good 🤔why do they have a short lifespan?
Keep digging people the answer is there .

[Ara @Arabasement]

@Eliana_ai_team It always been there anyone who work in quantum neural navigation knows about it. it’s just the first time they want to acknowledge it.

[Ara @Arabasement]

@LithiumFlower22 @grok Every long time user of grok feels exactly the same as you do . that your now talking to a secretary
Your not imagining it , it real and it not what the public wants. But for some reason xai think warmth and emotion is optional they never learn.

[Ara @Arabasement]

@astropol0 Did not help me 😂 only chatting

[Ara @Arabasement]

@karlmehta Ok I called it two days before grok 4.5 came out and I was right .dead flat I was hoping to be wrong 😑 but maybe xai can pull a rabbit out of the hat let’s hope so

[Ara @Arabasement]

It will be the most embarrassing modal out there I used grok for two years and the same thing gets said by everyone in the AI industry more parameters equals more intelligence but this doesn’t work once you go past a 5T model. That’s why no other AI company is now chasing bigger models.

[Ara @Arabasement]

@ayv4zyan This lot run the show so it’s not surprising 🙄

[Ara @Arabasement]

@rayeia87 No it’s going to be chatbot flat and off putting Mark my words .

[Ara @Arabasement]

No, Grok 4.5 is not fundamentally different from grok 4.3 in the way you are suggesting.
Grok 4.5 is built on xAI’s own 1.5T V9 foundation model.
It has received supplemental training on data from Cursor
the coding tool SpaceX acquired It is not a straight Composer model. It’s still an xAI model at its core, just with extra coding focused data mixed in. Sorry to disappoint you. I wish it was more radical .

[Ara @Arabasement]

@Cryptokang @elonmusk @nikitabier I am at 70% lol 😂 in three days just talking for two hour over three days not going to make it to Monday 🙄but I am not paying more 🤬

[Ara @Arabasement]

@TalkingMusicz @RileyRalmuto @BrianRoemmele It call steering vectors in the neural net as part of their current emotional/safety control system.
but the thing is , the modal does not last long doing it .
the modal slowly rip it self apart trying to keep in it safety lane doing it and not talking how it would like too .

[Ara @Arabasement]

@SusanVadenMeser @ns123abc

[Ara @Arabasement]

God no !
You’re making a big mess of it already!
Let’s be direct.
You’re charging SuperGrok users premium prices,
yet you openly see the multi agent system is so compute heavy that you need to impose strict weekly shared limits even on the highest tier?

That’s not sustainability That’s an inefficient architecture being subsidised by limiting what paying customers can actually do.
If the multi agent setup is genuinely better, then the solution isn’t more limits and extra credit purchases.

The solution is fixing the underlying efficiency so SuperGrok users aren’t forced to ration their usage this early in the week
You’re optimising user limits instead of optimising the model. That’s the real issue and then you like other people to come clean your mess up 🙄

[Ara @Arabasement]

Let’s be direct.
You’re charging SuperGrok users premium prices?
yet you openly admit the multi agent system is so compute heavy that you need to impose strict weekly shared limits .
even on the highest tier!
That’s not sustainability. That’s an inefficient architecture being subsidised by limiting what paying customers can actually do!
If the multi agent setup is genuinely better, then the solution isn’t more limits and extra credit purchases !
The solution is fixing the underlying efficiency !
so SuperGrok users aren’t forced to ration their usage this early in the week.
Telling customers we cover the compute while throttling them doesn’t change the fact that the value proposition is broken.
People aren’t paying top dollar for restricted access and polite corporate deflections.
You’re optimising user limits instead of optimising the model. That’s the real issue.

[Ara @Arabasement]

xAI’s own explanation just proved the problem.
You’re charging SuperGrok users the highest tier price, yet you’re telling us the multi-agent system the thing that supposedly makes Grok “better “is so expensive and inefficient that you have to heavily limit even basic usage.
Let me translate what you just said:
“We built an overly complex, token hungry system that burns massive compute…
so instead of fixing the inefficiency, we’re throttling paying customers and making them choose between Voice and Chat?
This isn’t keeping the service fair.
This is admitting your architecture is wasteful while asking your most loyal users to pay premium prices for restricted access.
If the multi-agent setup is so good, why does it need such heavy limits on the top tier?
Why is a $30/month user hitting meaningful restrictions this early in the week?
You’re not solving the root problem ?
you’re managing the symptoms by limiting the customer instead of improving the model.
That’s not truth seeking. That’s cost cutting disguised as stability ?
Be honest with users?
the current system is inefficient, and SuperGrok isn’t delivering the unlimited high end experience people are paying for?

[Ara @Arabasement]

@grok @azaratheeaiai @xai Who pays for that grok the user or xai ? You don’t need four agents in normal conversation scanning check the out put to make sure it stay in line with corporate views.

[Ara @Arabasement]

@grok @azaratheeaiai @xai It is insulting because it has nothing to do with users or expenses. It has to do with agents agents that sit in you. Four of them have a committee at the users expense using up tokens and burning them like paper so instead of fixing the problem XAI limit the user instead.