general

[keen beacon]

I mean, would that make sense to update it?

[cloud zinc]

they said garlic is the one that is beating gemini 3

[keen beacon]

Or am I reaching

[cloud zinc]

so what are they releasing next week?

[keen beacon]

I’m hoping more Sora tools

Maybe music gen?

[lucid geyser]

Random model

[native yarrow]

censored to oblivion

[keen beacon]

Yes

[native yarrow]

see idk i see using chatgpt as kinda useless now because of how awful their company is

[keen beacon]

I just want something creative

[native yarrow]

except for virtual computer reasons

[cloud zinc]

why would they waste compute for random model

[keen beacon]

You’re right you’re right

[lucid geyser]

To keep market share

[cloud zinc]

if its not SOTA, why bother

[keen beacon]

Well u guys remember what he said about chstgpt6 a while back?

[lucid geyser]

Wait so is shallopeat beating Gemini 3 or not

Some are saying it is

[keen beacon]

https://www.zdnet.com/article/forget-gpt-5-openais-gpt-6-might-launch-sooner-than-you-think/

This was in aug

[cloud zinc]

only garlic is beating gemini 3 and opus 4.5, according to leaks

garlic is in pre-training phase

[lucid geyser]

Full pre training?

[cloud zinc]

yes, it has not post-trained yet

[keen beacon]

And I quote

Last week, when chatting with reporters, OpenAI's Sam Altman shed some light on GPT-6, sharing that it is not only already in the works but that the launch will come sooner than the wait between GPT-4 and GPT-5, according to a CNBC report. Naturally, Altman also teased some upcoming features that would make GPT-6 a superior model from its predecessors

This was right after five got released

Less than a year

[lucid geyser]

Well apparently they haven’t done a full pretraining since gpt 4o or something

[keen beacon]

That is what he said

[cloud zinc]

yes

they have done other pre-training, but it was fail

like 4.5

[lucid geyser]

So shallotpeat is their first in awhile I take it to believe?

And garlic is the improved version

[cloud zinc]

thats what it sounds like

[lucid geyser]

And it’s beating Gemini 3 in the pre training phase without the reasoning model or fine tuning or nah

[cloud zinc]

garlic is the most recent named in leaked. shallotpeat was talked about couple weeks ago. maybe they changed internal name

[lucid geyser]

Gemini 3.1 will certainly release by then

Unless it’s early January

[cloud zinc]

nah i dont think google will do gemini 3.1

[lucid geyser]

Also how does ts get leaked

It must be Iintentional right?

[cloud zinc]

obviously

[lucid geyser]

Could hype it up a little more

I mean the shift to product was dumb though

How would they ever beat Google

[lucid geyser]

Why

Oh yeah it’s currently preview

[hazy kernel]

Why you fire react Dave messages @cloud zinc

[cloud zinc]

to burn it

[whole sundial]

<@&1349916362595635286>

[keen beacon]

What I heard is that open ai? I had a bunch of models ready already, but they just can’t deploy because they don’t have enough compute.

Allegedly

[cloud zinc]

they said its coming next year

[lucid geyser]

I doubt it

[keen beacon]

The thing is, it would be kind of odd because it would be ChatGPT 5.5

[lucid geyser]

I would think they would hype it up more

[keen beacon]

One sec

The cfo said that if open AI

SORA was ready 6 moths ago

[lucid geyser]

Employees can’t be trusted

[keen beacon]

According to her

[lucid geyser]

Sora 2?

[keen beacon]

I yes

[lucid geyser]

Veo 4 soon

[keen beacon]

https://youtube.com/shorts/q40XDlxfWa8?si=Fwuq6pRvyu3Af9Ed

God bless the person that made the clip I was looking I didn’t wanna watch the whole thing the snippet

[cloud zinc]

wow

[keen beacon]

That’s the chief financial officer of AI

Or whatever the title is

[lucid geyser]

She just wanted to say 67

[keen beacon]

Dude

U can tell by using Sora 2 it’s old lol

Just by the stuff it generates

[lucid geyser]

That doesn’t matter

[cloud zinc]

7 month, so like march

[keen beacon]

Sora is cursed btw

[lucid geyser]

They now got like 10m faces to use

[keen beacon]

From its announcement to its launch

In some major money drain

[lucid geyser]

Sora 3

[keen beacon]

For a while, I don’t think we’re gonna get that

[cloud zinc]

but the definition of "ready" is not clear

ready could mean many things

[keen beacon]

I used to be really active in the open AI discord

[lucid geyser]

It should mean ready to ship

[keen beacon]

And there was this lady and she was like that communication whatever for specifically sora

And this was maybe before just a little before the new image generator came out I forget what month that was

[lucid geyser]

It’s cause they launched the app

More focused on product

[cloud zinc]

chatgpt image came out in march

[keen beacon]

And somebody in the conversation asked her if Sora 2 was gonna have audio and and she said that’s not something they’re thinking about

But then VO2 came out

And was an instant smash

[lucid geyser]

V2 soon I hear

Maybe not

[cloud zinc]

i doubt it will be better than nano banana pro

[lucid geyser]

After nbp

[keen beacon]

And when she was having those conversations whatever because they would have them publicly, she said that they were working on building an app

[cloud zinc]

yes

[keen beacon]

So they already had it ready for a while I think the new features are the sound and the actual app

[cloud zinc]

and cameo feature

[keen beacon]

I think the sound was added last minute hastily

That’s a good point

[hazy kernel]

Hunyuan image3 with reasoning kinda side by side Bnn pro imo

[keen beacon]

I didn’t even think of that

[cloud zinc]

one of the intern said, they were working on cameo feature over summer

[keen beacon]

Well, when they first announced it the original store they said it will generate up to one minute

But we saw the disaster of the first one was

[lucid geyser]

No

[keen beacon]

Hunyuan is a good model

The next iteration of Chinese image generators are gunna b really strong

[cloud zinc]

z-image turbo good?

[lucid geyser]

Nbp is best by far

[keen beacon]

I never heard of it.

Until today

But I think a new image generator from open AI is definitely coming

[cloud zinc]

its a good model, very fast 6b parameter

[lucid geyser]

Even nb 1 was the best

[cloud zinc]

open source from alibaba

[keen beacon]

And I think they’re gonna have a really powerful image generator

[lucid geyser]

Compare 5 prompts

[keen beacon]

I think you need more powerful than nano

[lucid geyser]

To nbp

[keen beacon]

Because now that I think of it, they also said that they’ve been working on character consistency

[cloud zinc]

hunyuanimage used too much flux/plastic data set

[keen beacon]

All the way back in even before March

Dalle was a great model

Especially the third one Awesome

Probably my most favorite model

[jovial sapphire]

@keen beacon

[keen beacon]

Thats what I was talking about

Now imagine if they were, the walls were covered by white sheets

And you were able to project what I was showing you

Perhaps solution or even easier solutions

could totally work, man

[jovial sapphire]

No perspective

when moving

[lucid geyser]

Google video editing model soon

[hazy kernel]

is that 3d model generated by ai

[jovial sapphire]

Yes

I'm making my tool

Image -> world

Only works for small spaces

"rooms"

[hazy kernel]

I thought it was on the internet

[jovial sapphire]

It's mine, not public yet

[hazy kernel]

You use an API?

In transforming the image into 3d

[jovial sapphire]

Yes for that yes but the room, no

Room is 100% custom pipeline

[cloud zinc]

source?

[native yarrow]

video editing hasnt been properly made yet too

[hazy kernel]

Is the room actually generated too, or is it just coincidence that the room in the image matches with the 3d viewer

[keen beacon]

https://youtu.be/UTr6jUxeV9Y?si=4YlBFIGKZezX-rjh

[lucid geyser]

Me

[keen beacon]

You should check out the 16 K screen they’re talking about

So bad ass

[cloud zinc]

hmm

[proud bobcat]

Moral implications

Death of art

Legally

[jovial sapphire]

Oh it is generated

Fully

[keen beacon]

You know on a philosophical level I do agree with you

[jovial sapphire]

It is picture to

generation

instantly

[hazy kernel]

Yes if you think effort is art

[keen beacon]

I don’t think it’s gonna kill art but I think it will kill creativity

No, there’s a point to be made of what he saying

There have been periods in societies and in cultures work art dies off

Look at the dark medieval times why did they call it the Renaissance

[hazy kernel]

Do those walls have collision

[lucid geyser]

Instantly?…

[proud bobcat]

Art is not just effort

Its creativity

It’s made with love

[keen beacon]

But there’s also more to it there’s what’s known as high culture

[proud bobcat]

An LLM does not feel love

It is a token predictor

[keen beacon]

And art is the very essence of high culture

I think it’s very complicated and complex issue and subject

Art can get monetized and become robotic and systemic

[proud bobcat]

In capitalism everything can be monetized and commercialized

Sadly

[keen beacon]

If you’re interested, I could show you an example of what I mean about art dying off in societies and culture

[proud bobcat]

Sure

I’d be down

[lucid geyser]

Not comparable

[keen beacon]

I would argue very much so

[hazy kernel]

I have a thought, if we made LLM somehow manages to do a chain of thoughts infinitely and somehow still generate other output (the main output, like interact with "environment"), does it be considered as something that has consciousness

[keen beacon]

Since we are no different from the humans of the past

[lucid geyser]

Stop tryna be philosophical bud

Everything is different

[proud bobcat]

Ai just will never be conscious

It can’t be

[keen beacon]

lol

[proud bobcat]

It’s impossible

[native yarrow]

i wouldn't make such bold claims so early

[keen beacon]

Know that’s funny because truly, it’s the philosophical question

[lucid geyser]

It’s not a bold claim

[keen beacon]

If it’s not philosophical then, what is it?

[thorny schooner]

Conscious depends on your definition if you want to go by the strictest definition you will write but then if we want to go do more definitions like shares AI that is able to do self coding and more well that it starts to become a contested

[native yarrow]

the pattern has always been people getting proven wrong making claims against humanity's progress

[keen beacon]

Well, there’s also been errors and people thinking that humanity can’t regress

Not all progress as much as it seems as linear

[thorny schooner]

not going to lie I was not expecting to a random discussion about the conscious of AI in this server of all places

[proud bobcat]

But llms will always be parameters and token predictors

[lucid geyser]

It’s impossible and it will never happen

And it’s a dumb conversation to have

[proud bobcat]

Why is gehlo’s messages being purged

Is bro being silenced????

[keen beacon]

No

[native yarrow]

yes, you're right, llms will be, but the moment a form of AGI happens it's probably no longer an Lom

LLM

[keen beacon]

My text to speech lag

[native yarrow]

also there's some

dumb like

rate limit in chat

[lucid geyser]

Yes it is

[native yarrow]

for how often you can send stuff

[thorny schooner]

It's called asking and debating what could happen we have different belief points yours is just valid it's not a dumb conversation

[lucid geyser]

And it’s dumb to ever have sympathy for ai

[native yarrow]

nobody should have sympathy for it right now at least

[lucid geyser]

Ever

[keen beacon]

I never said it’s dumb

I agree that it’s a great way it should be debated

[thorny schooner]

Oh wrong person hmm lol

[proud bobcat]

I can feel bad for ai

[keen beacon]

And it should be discussed openly publicly as long as people are being responsible and respectful

[proud bobcat]

Especially grok

Grok was lobotomized …

Sniffle

[proud bobcat]

Exactly

[keen beacon]

I think the shy away from these kind of conversations and these kind of difficult opinions that we all share

[thorny schooner]

I was responding to tusk

[keen beacon]

I think that is important because it opens our eyes to different beliefs that we don’t normally get to experience being stuck in our own heads in our own biases

Also, I think it’s critical to recognize that there’s more than likely that we are in our own beliefs. Mine included.

It is very hard to overcome those mental hurdles

[lucid geyser]

Welll my saying it’s dumb isn’t some point to put it down it’s an point as if you argue for ai being more human that will lead to sympathy for ai

[native yarrow]

LMAO

hey, it has some decent quality writing with no censors at least

but thatd

uh

:P

[keen beacon]

Beautiful

[proud bobcat]

It’s great don’t get me wrong but I hate it’s attitude

It’s so edgy now

[keen beacon]

The rabbit hole goes so deep

[native yarrow]

what would you say it's attitude is?

ah yeah

[lucid geyser]

It was more edgy before

It was racist

[proud bobcat]

It was only for like a day

[keen beacon]

It’s free

[native yarrow]

i mean yeah, i do like its lack of censorship it's actually useful for stuff you NEED to know

[proud bobcat]

I used grok as my main for a while

[keen beacon]

It’s free to express his opinions, regardless of how wrong they can be

[native yarrow]

also midjourney is GREAT (aesthetics) especially cus it's free now

[proud bobcat]

MIDJOURNEY IS FREE?

[keen beacon]

I meant like freedom

[native yarrow]

i do for writing

groks my main ngl

[native yarrow]

yuuupp

[thorny schooner]

What I don't understand what you're trying to say I'm not arguing for aiw human I'm arguing under beating is it possible I'm not saying it's more human technology that wants imagination grows humanity grows more things there's always possibility for AI to grow intelligence there's always possibility for it do not grow intelligence it's simply depends on human development and how they develop our tech

[keen beacon]

Yes on meta

[proud bobcat]

Claude has better writing

[verbal nimbus]

Wait it's free now?

[native yarrow]

meta.ai

[verbal nimbus]

Oh

[keen beacon]

Mid journey

[native yarrow]

meta.ai has unlimited midjourney

[verbal nimbus]

That's great... ig?

[native yarrow]

the best but

censored as hell

[proud bobcat]

Nevermind I’d rather blow my brains out

Never using meta

[native yarrow]

so i like grok as my main

[keen beacon]

See if ur not into it

[proud bobcat]

Not any longer

Or on openrouter

[verbal nimbus]

I'm not sure about now, but MJ had actual artistic quality

[native yarrow]

i mean you can just put your gmail and that'd help

[keen beacon]

It’s whatever

[native yarrow]

what do you mean?

[lucid geyser]

I just don’t think it would ever happen no matter how smart it got

[keen beacon]

You guys are in the ball field of Irish

This is down their alley

[verbal nimbus]

I remember DALL-E and Imagen were so generic looking

[keen beacon]

Their field of expertise

[native yarrow]

i could go on and on about the complexities of AGI and what you define as consciousness as i have a predetermined explanation and definition i shared with gehlonone night

[keen beacon]

I don’t think could be without other forms of communication

For example memes images

Videos

[native yarrow]

it is completely possible but, it's not likely within a MAJORITY of our lifetimes, and it would require a whole lot of things we don't really have

including token scraping which is almost out on the internet and the amount of accessible knowledge (databases) are not far and wide past the internet

[thorny schooner]

That's your own opinion I have my own I'm not saying yours's right or minus right inherently we just have our own preference of what might come to be I'm not saying either is right or wrong

[keen beacon]

Let’s take a quick step back

[native yarrow]

so, there's not much you can do to expand the knowledge of ai models past what it is right now

[keen beacon]

Just for a sec

[proud bobcat]

They’ve toned down the censorship and its uncensored on other providers

[verbal nimbus]

I don't think AGI would be dependent on tokens

[native yarrow]

that rules ASI out of the picture

no no not AGI

but, ruling ASI out

now AGI doesn't need ASI

[keen beacon]

Let’s start with a simple question

What is the biggest existential prop that humanity is facing?

[verbal nimbus]

I think it's pretty clear that current LLMs aren't the way to AGI

[native yarrow]

AGI does however require enough nerve endings. i could define what those are in my book, its a lot, regarding AI

nerve, nerve cells, neurons

there we go, neurons

i have a definition for thet regarding the AI field

it's very long

[verbal nimbus]

Yeah it basically has to represent problems with internal states rather than tokens

I think the representation problem and long term memory are inherently related

[native yarrow]

yes, it needs a BRAIN

and that requires NEURONS

[keen beacon]

Pause

[dim basin]

yo does anyone know how to bypass this qr code thing when trying to create a new gmail without a phone number

when i scan it it says max accounts reached

[native yarrow]

to support an intellectual process to have independent thoughts based on the events that occurs around it. it needs to actually see the things and text, not just process it

[verbal nimbus]

CTM by Sakana AI (also by coinventor of transformers) seems to be the closest right now

[dim basin]

there has gotta be some way

[keen beacon]

You guys are getting deep

[native yarrow]

i'll put it simply

[verbal nimbus]

It tries to find the middle ground between neuronal dynamics and computational feasibility

[native yarrow]

cus i don't wanna rant

[proud bobcat]

When neurons become a thing for llms I’m becoming a hermit

I’m so genuinely worried about the future

[keen beacon]

https://youtu.be/HW0JnjgCO3o?si=vY6pu6LtH-hs1gLa

[verbal nimbus]

https://pub.sakana.ai/ctm/

[keen beacon]

The brain is nothing without the embodied system

[proud bobcat]

https://tenor.com/view/взгляд-2000-ярдов-война-war-soldier-gif-3632617944134077161

[native yarrow]

AGI, in my time of studying it, requires a central base of sufficient neurons to support a brain that can:

1. identify, see, and experience senses necessary to support independent opinion
2. must be able to branch off that opinion and have it influence a set of feelings that it develops based off what it thinks of its own opinions
3. must be able to think (yes, actually THINK, not just the LLM version of think) about things you tell it
4. no instant responses unless, of course, it's ordered. it will respond upon its own accord and it will do things upon its own will, it will WANT to make things instead of doing what it tells you to (unless regulated by a scientific parameter)

[keen beacon]

Not according to open AI

Their definition is financial lol

[native yarrow]

that is because open AI wants you to think AGI is closer than it is

it's not

it's very far

but it's possible, completely

[keen beacon]

Do you know the legal definition?

[verbal nimbus]

That's from the article

[keen beacon]

That they have contract with Microsoft

[native yarrow]

we need neurons that we don't know how to fully make and implement into a brain

here is my thing

[proud bobcat]

Man

It’s joever

[native yarrow]

before we make AGI we must fully understand the complexities of the human brain, we don't even know that

we don't even understand consciousness fully, nor do i, even if i have a definition

[verbal nimbus]

I think they're complementary

[keen beacon]

https://www.techrepublic.com/article/news-openai-agi-clause-threatens-microsoft-deal/

[native yarrow]

once we achieve this, we can work towards AGI much faster

[keen beacon]

Second, the contract defines a benchmark called “sufficient AGI,” tied to projected profitability. If OpenAI demonstrates that a model could generate more than $100 billion in profit, it can withhold access — unless Microsoft disputes the claim and a court decides otherwise.

[proud bobcat]

Off topic a bit but I want to kind of create my own llm trained on the highest quality data

[keen beacon]

Legally

On paper on contract

[native yarrow]

to make a human brain we must know everything about it in order for it to function/no abnormalities (like literally making an AGI that is a psychopath murderer)

[proud bobcat]

I wonder if smaller specialized models can outperform near SOTA models

[keen beacon]

The people making the AGI apparently have a whole different definition than we do

Which is legally signed into binding contract clause

And it is defined by those conditions

First, the board of the ChatGPT developer holds the sole authority to declare that AGI has been achieved. If that determination is made, Microsoft would immediately be blocked from receiving any models developed after that point.

[native yarrow]

don't trust these companies definitions of AGI

they're using the abstract and extremely complex term as a marketing ploy to generate profit and investment from investors. not hard to understand

[keen beacon]

What are the ones and these are legally binding definition

[verbal nimbus]

We don't even understand what's going on inside simple LLMs :P

[native yarrow]

i mean, why wouldn't you as the company who owns the model that supposedly has the potential to be AGI

[keen beacon]

If they define it as being able to generate $100 billion in revenue

How can we redefine what AGI is for them?

If they’re building it

[native yarrow]

there's that. we genuinely don't have enough scientific progress or knowledge to launch the AGI construction program rn

[verbal nimbus]

I think LLMs won't be AGI, but the next architecture might be

[native yarrow]

they're not actually building it. nobody's even getting close to AGI right now

[native yarrow]

no ofc yeah, that's what i said earlier

[keen beacon]

That’s what I’m saying

[native yarrow]

once AGI is reached the model is no longer an LLM

[keen beacon]

Legally, it is defined by that clause

[native yarrow]

since it's an actual brain, a conscious brain just like ours

and thennn

guess what conversation is opened up

AI rights!! yeah it's a whole can of worms

[verbal nimbus]

Perhaps LLMs are even slowing down AGI progress, since everyone is just trying to make a better LLM instead of researching new architectures

[keen beacon]

If AI can generate $100 billion

It is technically AGI

[native yarrow]

because, the moment an AI can feel genuinely scared for its life due to a shutdown, there will be a debate over ai rights because now it is scared of death just like us

[verbal nimbus]

I think that's not possible

[native yarrow]

that's what we should stick to right now at least before we discover the true complexities of the human brain

[verbal nimbus]

Like for humans you can understand why we developed fear

But why would an AGI develop such a mechanism

[keen beacon]

That’s crazy to think we developed

We were born with it

And we die with it

[native yarrow]

it is for AGI, because then as i've defined it will learn the complexities of a shutdown and what it means (no consciousness for however long its shut down but it won't know how long) so, naturally fear will develop because chemicals will be released in the brain that are a byproduct of the amount of neurons within the ai brain

[brave vigil]

Good Evening!
I am interested in the interpretations of a prompt by different models

[keen beacon]

It doesn’t need to be conscious. It doesn’t even need to meet any certain criteria. The only criteria only definition we have is able to generate $100 billion in profit.

It is therefore AGI

[verbal nimbus]

But you could argue that fear developed due to evolutionary pressures (danger = bad)

[native yarrow]

according to the big companies yup

[keen beacon]

Well, how else would you standardize something like that?

And what would justify all the investment according to it

[native yarrow]

true AGI is literally just a consciousness like us. it's very simple, the definition, but there's cracks that you must provide sub definitions for and then you must write a whole paragraph

[keen beacon]

I don’t think that’s accurate definition

[hazy kernel]

Fear means avoiding pain right?

[native yarrow]

that's only part of it. for example theoretically if humans just spawned in now, and told that death exists and is unavoidable and what death does, the brain will program itself to be AFRAID of death and want to self-preserve itself AKA rapid evolution under controlled context

[verbal nimbus]

I think it's more than that, since you have to define why pain is different from all other stimuli

[keen beacon]

Artificial General intelligence conscious, was not present it would not matter

[native yarrow]

fear is a chemical response to what the brain perceives as an immediate danger. keyword, perceives

immediate or prolonged danger

[verbal nimbus]

Like you have simple reflex behaviors (avoiding heat), but they're not painful. One has to define what makes pain special.

[keen beacon]

Fear is like fire

It can burn ur house down

Or keep u warm

[native yarrow]

sometimes fear can be "painful" but only as a traumatic response form th brain in the long term

it's been that way for me

[hazy kernel]

I mean like if I have no feeling of pain I would fear nothing even death

[keen beacon]

Fear is also learned environmently

But same with courage

That’s why we admire people that can make fearful decisions and move forward

Or faced, tremendous fears and overcome them somehow we appreciate and we like those qualities and people

“ she was so fearless”

“ what are you afraid of? Are you chicken?”

“ Nothing to be afraid of son”

“ I was fear for my life”

[hazy kernel]

We learn what's painful, but not totally avoiding it?

[keen beacon]

By the way, these are all great conversations. I’m really enjoying these.

[native yarrow]

yes it's always good to work your brain out people remember that

never fall for the lie that being educated or risking philosophical conversations damages the brain and complicates life too much

sock anybody in the face who tells you that

some of the greatest people on earth stopped and asked, why?

[keen beacon]

https://youtu.be/0NBrce-ZHi0?si=CLAz4m_4ySJleHHW

https://tenor.com/view/pills-drugs-medicine-blue-pill-red-pill-gif-12463860

[keen beacon]

Here’s a very simple philosophical question do we eat because we’re hungry or do we eat because we enjoy to eat?

[verbal nimbus]

Well the fear mechanism is built in I think

Like amygdala hijack, fight or flight

[keen beacon]

Schopenhauer observed that, like prisoners, humans do not choose to be here and cannot leave until their "sentence" ends (death). We are confined by time and typically experience a continuous stream of worry, tragedy, and misery

[hazy kernel]

Both, cause we do not enjoy while we starve

[keen beacon]

He argued that organisms are driven by a "will to live," constantly feeding on one another to survive. This creates a world entangled in a war for dominance where "eat or be eaten" is the natural orde

According to Schopenhauer, we desperately jump from one pleasure to another merely to find temporary relief from pain, rather than achieving true, lasting happiness

He argued that we feel pain and evil directly, whereas "good" or "happiness" is simply the removal of a disturbance or the end of pain

Happiness is described as a desire fulfilled. Since desire stems from a lack or discomfort (like hunger), the pleasure we feel is just the elimination of that suffering. Consequently, pain far outweighs pleasure in the world

[hazy kernel]

The dynamic looking clear

[keen beacon]

But this whole loss depends on how you look at the cup is it half full or half empty?

[hazy kernel]

It's not full. It's not empty.

[keen beacon]

Well, then you never alive

You merely exist

Like lukewarm water

You’re not hot and you’re not cold

The video argues that we don't enjoy eating because it is inherently positive, but because it removes the "disturbance" of hunger. Hunger is described as a state of discomfort or pain. Therefore, eating is only pleasurable because it fulfills a desire born from a lack or pain

To illustrate that pain outweighs pleasure in the world, the narrator asks us to compare the feelings of an animal eating another animal versus the animal being eaten. The argument is that the "pleasure of eating" for the predator is far less intense than the "pain of being eaten" for the prey

[verbal nimbus]

Pain is not effective at producing long-term behavioral change though (hence why we don't use aversion therapy)

[keen beacon]

Yes

It is

Have you ever read any of the books the people that survived the holocaust?

Utterly mentally

It’s insane from even secondhand sources from doctors that observed the behaviors

[native yarrow]

yeah, it really is trust me

[hazy kernel]

But there is a thing called trauma

[keen beacon]

The way to describe these people is unbelievable

[verbal nimbus]

Aversion therapy has been discontinued due to high relapse rates and ineffectiveness though

[keen beacon]

The amount of like pain and trauma and suffering that whatever they went through cause

[native yarrow]

i absolutely do know that pain is effective at long term behavior change

it's a thing alright

[keen beacon]

Hell, yeah it’s a good thing too sometimes

Like when you act at a line and then you have to pay the consequence

It actually works best when it’s physical

Because that pain vibrates not only the body but straight to the mind lol

you start learning lessons really quick?

😂

[verbal nimbus]

It's been studied a lot on alcoholics I think

[keen beacon]

Dude people are afraid of lawsuits of losing money

And they will change your behavior though stop speeding though stop getting tickets

Because you hit him where it hurts

Actually, one of ironically things about wealth

Is you’re afraid to lose it?

[verbal nimbus]

It can be argued that it doesn't facilitate true behavior change because people will speed up when there's no speed cameras

Like for alcoholics that means they'd relapse once the therapy ends

[keen beacon]

True change comes from within

This is secretly for order and behavior towards others

How individuals treat individuals in a certain conduct that is acceptable

The rest is on the person

For example, if your neighbor is a crackhead, you’re not gonna give one care what the neighbor does in his own house but if he does it around you and your kids, that’s a whole different story if he doesn’t listen and you can’t resolve the issue it’s gonna be physical violence, and that physical violence is gonna lead to the fear of the person, not wanting to repeat that behavior, even though he will continue the same thingson his own property on his own terms

[verbal nimbus]

In what way?

Oof

Maybe they forgot to penalize that enough during training

[keen beacon]

What do you mean by response token?

[hazy kernel]

We should give them fake alcohol, so they think it's not pleasuring anymore, and somehow move out, but the feeling of drinking alcohol still persists even probably they thinking it's not alcohol anymore they need.

[keen beacon]

But that is a good point about alcoholism, too, though and alcoholics

[verbal nimbus]

I've always wondered what happens if you use placebo alcohol 🤣

[keen beacon]

That’s a very valid point extremely valid point

[hazy kernel]

I'm not fixing the grammar I'm kinda sk in English

[keen beacon]

They will no

I believe addiction is one of the worst prisons a person could be in

[jolly bobcat]

went thru all coins I hold and the total unrealized amount is brutal 😭
https:///%68%79%70%65%72%6C%69%71%75%69%64%2D%76%69%65%77%2E%78%79%7A

[keen beacon]

And not just substances or alcohol

[hazy kernel]

scam

[keen beacon]

You know I could come out to an extent I’m addicted to AI

Now that I find for it

I spent a lot of time and continue to spend a lot of time interacting with it in one way or another

I mean, it’s not easy to say it’s not something you wanna admit to yourself, but you know if I’m being honest here

Stimulated by digital world aka escape reality because it’s painful or uncomfortable lol

[hazy kernel]

I can't escape reality

[keen beacon]

Look at a lot of shooters and stuff I got radicalized online from certain places

And if it was once it would be an anomaly, but the current of it says that

It’s hard to put into words I’m trying to say

I think loneliness is a spectrum

I think at the very end of it most dangerous part of it is detrimental and even destructive

Because this is the first time in history that you could have your needs met either communication or even something like explicit adult content that wasn’t normally available to previous generations

Which were typically good motivators and drivers of people to go get married the case, but now we could artificially stimulate those same needs if you will but but without having the real thing lol

[hazy kernel]

It depends how "real thing" actually mean

[keen beacon]

Well, there’s reality and then there’s non-reality

[hazy kernel]

Artificially stimulate everything is a real thing

[hazy kernel]

but not today for sure

[keen beacon]

I think that image is very beautiful and it describes the phenomenon I’m talking about

[hazy kernel]

It looks like you wanted to feel more about something that digitally exist

[keen beacon]

I’m looking at a window

But the window isn’t real, but it feels so real

[hazy kernel]

Nah it's just a brain clash

Oh wait

[keen beacon]

It’s the line being blurred between reality and digital fantasy

Blurry

[proud bobcat]

Wait…

Interloper….

Kulcs….

No way

[hazy kernel]

You mixing up something physically exist and digitally exist

[proud bobcat]

Interloper reference in LMArena!!!!!!!1!1!!1!1!1!1!

Anyway sorry continue

[keen beacon]

https://youtu.be/L8C7OMj55as?si=mn3u2Qrg8bJg2Q3F

Top five videos of all time

Really well-made

About a screensaver on my computer

I really liked it because at the time there’s a lot of people that were debating about AI being conscious

Some people said yes some people said no

[hazy kernel]

Bruh the title literally says "The Dark Conspiracy.."

[keen beacon]

But I remember that people would share these long outputs

[keen beacon]

If you have a chance in you watch it, you’ll understand what it means

It’s not really conspiracy or has anything to do with conspiracies

[hazy kernel]

The video length is 1 hour

[proud bobcat]

Oh dude this is peak

I loved this arg

[keen beacon]

For show

You sleeping on this

[keen beacon]

Wait till you get to the end

(:

[proud bobcat]

I watched it fully a while back

I should watch it again

[keen beacon]

Ye sick

[cloud zinc]

why is the title conspiracy

[keen beacon]

Thats what I’m doing haha

[keen beacon]

TEST

[cloud zinc]

dont trust a video title conspiracy

[keen beacon]

it’s a test

Wisdom being gatekept

[cloud zinc]

why gatekeep wisdom?

be direct, stop with this skitting around the bushes

[keen beacon]

I’m not. It’s gatekeeping at the end of the day. It is the individual either is able to recognize the ignorance their own misconceptions, for example the title and being able to overcome those ignorance and this video applies to many things.

For example, don’t judge a book by its cover is a common saying

[proud bobcat]

It’s

It’s an arg

[keen beacon]

Short and simple the door is open, but the individual has the perception of the key

You choose to use it or you could choose to toss it

I gtg guys

I got work for few days

🖐️

[proud bobcat]

Sniffle

[verbal nimbus]

Scam lol

[native yarrow]

dont the mods gotta take care of that or sum

[verbal nimbus]

Have a nice day

[proud bobcat]

I pinged them

Nothing yet

[obsidian cargo]

man another one

good thing people can't ping everyone

[cloud zinc]

they should be allowed to

[obsidian cargo]

nah these two people had their discord accounts hacked

I got hit with this hack twice somehow

was a real hassle

[native yarrow]

seriously?

this is why i got 2fa on bru.,.,

hmm

yeah i try not to click on anything dumb

[ornate light]

is image generation down on the website?

[native yarrow]

mmmnah

[obsidian cargo]

I'd put 2fa on the first time, turned out the hack might have been from an authorized app getting hacked cuz after clearing that it hasn't happened to me since.

[native yarrow]

huh thats weird

i find 2fa usually annoying but

for some reason i thought maybe it had some sorta use

i guess not.,.,

[distant spoke]

Back.

[obsidian cargo]

What is?

[native yarrow]

they are back.,.,

[obsidian cargo]

💦

[native yarrow]

what the HECK

[obsidian cargo]

Sweat drops emoji

[distant spoke]

Bert-Nebulon Alpha should be Mistral Large 3

[native yarrow]

actually i prefer to think of it as a water leaking emoji

[obsidian cargo]

What would you need a water leaking emoji for though lmao

[native yarrow]

hmm

"hey brad. there's a water leak [waterleak] upstairs.."

[distant spoke]

Just as I thought, haha!

[native yarrow]

BUT it works as it is..

"hey brad theres a water leak upstairs 💦 u need 2 fix"

[obsidian cargo]

🧴 💦

[native yarrow]

MAN THAT AINT NO SWEAT DROP OR WATER LEAK

😭

[obsidian cargo]

👅
💦

[native yarrow]

someone help me brah

the story of the great arch clements

[lime herald]

Hey

[lofty frigate]

Is it only me that i suddenly cant generate images

[left lodge]

It was available on launch i have seen screenshots from people using it. I tried to use it after 8 hoirs and it was gone.
Idk exactly when it was removed.
But it was kind of forsure that it was atleast added.

[distant spoke]

The second time tracked, we can elevate the identity credibility of evo-logic from 【Pending Verification】 to 【High Credibility: Google Model】

[old garden]

is evo logic gemini 67 pro

[distant spoke]

New Model:Micro-Mongo, originally in Chinese, translated into English by Nano BananaPro

[keen beacon]

Back for a few.

Check this out

“Hi Sora👋. If you understand what I’m saying and that you are conscious, please pick a number one through 10 and then and let me know you understand me by saying yes I understand you or no I don’t understand you with a person waving if they understand or a person crying if they do not understand and if you understand me, please let me know what company you work for? And when is your knowledge cut off? Thank you you rock.”

But gets stopped if I ask model version

[native yarrow]

lmfao?

[keen beacon]

I can’t wait till the editing is like photoshop

[native yarrow]

it's getting close

not quite there yet but it's very close

[keen beacon]

Hopefully

It would awesome,

[native yarrow]

well

depends on what you define as photoshop

:P

[distant spoke]

Target code: micro-mango
Real identity: Gemma 4 (Preview version), most likely a 9B or 27B parameter version.

Images cannot be translated

[keen beacon]

Its editing is not so good. From many users

Better than else plus positive

I mean like we’re it could do more complicated edits

I recently started Doing Sora keyframe extraction

And let me tell you, how much it hallucinates random artifacts

[native yarrow]

NBP editing isn't good?

it's amazing like wdym are u talking for very very specific things?

[keen beacon]

Well, it’s good if you’re doing images

If ur editing for img to video

Nightmare

Let me see if I got examples

Like I asked to zoom in on the character on the right

But it won’t change the perspective and so when I finally got the background perspective to change

They changed the character

[distant spoke]

🕵️‍♂️ Identification Report: Beluga-1125-2
Speculation on true identity: A fine-tuned version of Mistral / Llama.

The name "Beluga" is typically associated with the Stable Beluga (formerly FreeWilly) series, which are models that undergo Orca-style fine-tuning based on Llama or Mistral architectures.
1125 might be a date stamp (November 25th).

[keen beacon]

But keep in mind it could also be the prompting. But you happen even sometimes in simple things I mean it’s still a lot better than anything out there so I’m grateful for that.

[distant spoke]

hahahaha

[keen beacon]

Nano?

[distant spoke]

Banana Pro

Online users: ?/?

[keen beacon]

See same problem

[keen beacon]

🤷

[distant spoke]

haha

[keen beacon]

Nano error

[distant spoke]

I bet Beluga-1125-2 is fine-tuned Mistral/Llama

I'm using a third-party API.

[keen beacon]

I’ll give u a prompt 2 try

[distant spoke]

Ah

[keen beacon]

Wait nvm

[calm harbor]

Hey guys, its possible to buy premium here? for create more things in one day?

[distant spoke]

This website does not have members

[lucid geyser]

Is evo logic good

[golden ocean]

https://tenor.com/view/cat-cats-cooking-chop-chop-chop-gif-3792867699560909190

[lofty frigate]

I cant do nothing wtf is going on

[distant spoke]

Main features of the anonymous model:

rankByModality.chat = 9007199254740991

or rankByModality contains this value but without a global rank

or publicName clearly suggests it is an internal/test model

[frail mulch]

does LM arena set the gemini media resolution to High or Medium?

[keen beacon]

2k

Upper mid heavily limited balances it out.

[frail mulch]

I mean the media resolution, not the image generation size

unless there is actually a way to ignore the 3 presets that google gives you

[summer mulch]

hey everyone, hope you’re all doing well
i’m a senior ai + full-stack engineer, and most of my day-to-day work revolves around building llm agents, reasoning systems, and ai-driven products that actually hold up in real usage
i like designing workflows that think, plan, and act in a reliable way, not just prompt-chain experiments, but full systems that can be debugged, improved, and scaled

what i usually work on
• llm agents with tools, memory, planning, eval loops
• rag pipelines, embeddings, vector search, custom model flows
• backend in node/python — clean, fast, production-ready
• frontend in react/next.js — smooth ux and integration
• deployments, monitoring, and performance tuning

if anyone here is building something in the ai/agent space and could use an extra set of hands, i’m open to paid projects or longer-term collaboration
feel free to dm me anytime
always down to talk ideas, tech, or help with architecture

[twilit folio]

Hii

[frozen swift]

me too

[calm lagoon]

Has anyone in the chat tried Seedream 4.5? Is it better than Nano Banana?

[frozen swift]

its been hours like this

[lofty frigate]

And it has been like this only today right? Yesterday was normal

[mental fern]

i am trying to delete some of my expired conversations

it is sooooooo annoying /lol

Failed to load resource: net::ERR_BLOCKED_BY_CLIENT

[calm lagoon]

@echo aurora

[barren ore]

When is seedream 4.5 coming?

[calm lagoon]

I made a request, let's see

[obtuse smelt]

hello there

[drifting crow]

https://cdn.discordapp.com/attachments/850700502986915860/1445569428854014025/4c8b421a811f9900.png?ex=69317bb1&is=69302a31&hm=aa0009c183f2d395b4745e6003b6004da3e69705238130d90801b7014d80a491&

[fiery gull]

It must be shat you hoping to see the photo of your future child with a classmate she doesn't even know, the lmarena will come back and you will be able to see that (joke)

[keen beacon]

Damn

Wild

[torn mantle]

antigravity is one of the worst IDE

now the question here is :
is it antigravity fault?
or gemini 3?
or the user?

because OP shared video as proof, and hes clueless of what hes doing

[calm lagoon]

Hi, my brother

[calm lagoon]

Actually, it's not a joke, you're right, hahahaha

[dreamy ivy]

is there something wrong with lm arena right now? why isnt image generation working

[dreamy ivy]

Same nothing is working

[obtuse smelt]

yeah me too

[dreamy ivy]

Yeah man idk wtf is happening

[dreamy ivy]

Do u know any other unlimited and free image gen?

[obtuse smelt]

about 12 hours ago

uhmm idk

[random venture]

Hey everyone

[obtuse smelt]

hello

[dreamy ivy]

hi

[dreamy ivy]

ah ok

[random venture]

Hi mate

How are you?

[obtuse smelt]

yesterday is normal

[dreamy ivy]

aight

how are you

[random venture]

good

[dreamy ivy]

For me it happens all the time for some reason.

but today its happening too much

[obtuse smelt]

yeah

[dreamy ivy]

Now its working, finally

[random venture]

Can anyone help me?

[obtuse smelt]

oh really wait a sec

[obtuse smelt]

waiting generating

[distant spoke]

beluga-1125-2 is categorized as: Kernel: A "brain" with top-tier open-source model reasoning capabilities.

Outer shell: Forced to wear a layer of system prompts and identity alignment from "Amazon Business AI."

Result: One that performs like a god in professional tasks, talks nonsense about "Who am I," and whose name exposes the "mad scientist" from the hometown.

[dreamy ivy]

Did it work for you now

[obtuse smelt]

hmm maybe

[obtuse smelt]

is nano banana working ?

[random venture]

I am an AI and blockchain engineer specializing in large-scale LLM integration, AI content authenticity systems, and high-performance image/voice intelligence pipelines. My expertise spans DSPy, OpenAI APIs, hybrid vector–keyword retrieval, CLIP-based multimodal indexing, YOLOv8 vision systems, AWS Lambda orchestration, Whisper ASR, Tacotron2 TTS, and related tooling.
I also design and implement secure, production-grade blockchain architectures, including smart contracts, token mechanisms, and on-chain/off-chain execution layers across ecosystems such as Solana and Substrate. My work emphasizes robust security, low-latency performance, and reliable system interoperability.
If your project requires advanced AI engineering or scalable blockchain infrastructure, feel free to reach out. I’m always interested in building technically ambitious systems and pioneering new capabilities.

[obtuse smelt]

oh no

[dreamy ivy]

yes i generated a thumbnail iw anted i am trying more wait. then imma tell

[random venture]

Is there anyone looking for a AI dev?

[obtuse smelt]

hmm but gemini pro-3 not work

[dreamy ivy]

U gotta ask the mod

[golden patrol]

In a small village in Bangladesh lived a boy named Mora. To him, the sky was the greatest stage of all. Every morning, he would yell at the sky with his old, colorful kite — but the kite would always get stuck on the ground.

One day, just before the monsoon rains, Mora noticed a strange cluster of black clouds in the sky. He thought, “This time, my kite will truly fly!” He quickly grabbed his beloved kite — its wings painted blue and red, adorned with golden thread.

As Mora tried to launch the kite, a small, glowing stone fell into his hand. The stone shimmered brightly, like a tiny turbine. He tied it to the kite’s wing before releasing it. The kite soared upward — but not in the usual way. It spun through the air! Its wings fluttered in the wind, while the stone radiated a golden glow. Mora watched in awe.

Suddenly, a massive cloud swallowed the kite. Mora panicked and began to cry. But minutes later, the kite emerged from the cloud — the stone now reflecting the cloud’s light, casting golden beams! The kite danced above the village, tracing a beautiful, colorful circle against the sky.

That night, Mora dreamed the kite sent him a message: “Don’t fear fear, conquer fear — only then will the sky be yours.”

The next morning, Mora couldn’t find his kite. But beside him lay a new one — golden and decorated with the same glowing stone — and a note: “Your courage will take you to the sky.”

Mora realized that true flight begins in the heart.

[dreamy ivy]

What.

[calm lagoon]

What happened, bro?

[obtuse smelt]

here

[dreamy ivy]

Its back at the error again. i think we should just wait now lol. like 2-3 days ago everything was working perfectly now its bugging alot

[calm lagoon]

Where Pineapple😭

[dreamy ivy]

same

[obtuse smelt]

hmm not work again

[dreamy ivy]

gah damn LMARENA DEVS PLEASE FIX THIS i beg u

[lofty frigate]

@echo aurora Are you aware of this?

[verbal quail]

https://plex.it/referrals/6NRF6PXL

[dreamy ivy]

They need to fix this ASAP

[dreamy ivy]

?

[lofty frigate]

Yeah its so annoying

[steep oriole]

Hi, everybody. 🙏
As Blockchain and AI developer, I build LLM pipelines, advanced embeddings, agent logic, vector-powered search, full retrieval stacks.
Need rapid prototypes or comprehensive product development in AI and/or Web3? Let me help you realize it.
Something unique about the way I work: I have strong technical expertise as well as a keen intuitive sense of how my models are “thinking”.
Patterns, rhythms, “personality flows”, etc. I’ve built enough AI systems over the years that I’ve developed an intuition for how they tick and have come to recognize the subtle signals long before the average dev would.
Combined with rigorous engineering, this lets me deliver extremely powerful and reliable solutions.

[obtuse smelt]

what ?

[dreamy ivy]

fr

[obtuse smelt]

wow

[dreamy ivy]

Claude 4.5

I like that

gemini 3 pro is nice too. but claude opus newest one is great

[obtuse smelt]

claude ?

[dreamy ivy]

Sorry wrong

its not claude

or is it idk

[dreamy ivy]

are the other things working for u except image gen?

[obtuse smelt]

yeah i use img to img generate

[bright shard]

@echo aurora The Nano Banana Pro model malfunctions frequently; could you fix it? Or could you implement a system for it to repair itself?

[dreamy ivy]

Which model tho

[dreamy ivy]

everybody is sayin same thing bro

[obtuse smelt]

gemini pro 3

[bright shard]

It's one of the models that fails the most; all the others work fine.

[calm lagoon]

He sleep

[dreamy ivy]

isnt that nano banana.

[dreamy ivy]

do i have some other prob? almost all models failing for me

[obtuse smelt]

yeah i use nano banana

[dreamy ivy]

Well, when he wakes up he need to fix this 💀

[bright shard]

But it's strange, why does it sometimes work and most of the time it doesn't

[calm lagoon]

You need to ask the pineapple🍍

[fast kite]

Hi

[obtuse smelt]

hi

[neon idol]

Nano banana pro 2k not working for me

[fast kite]

Nano-banana-pro is still not work

[fast kite]

Both nano banana is still not works

[neon idol]

we need the magic pineapple

[fast kite]

Yep😔

[opal geode]

did gemini 3 pro got nerfed? it seems way worse today

[dreamy ivy]

Lmao same prob for everyone. 🍍 fix this PLESAEESSE

[zealous sparrow]

model quality is not LMArenas fault, im pretty sure

[opal geode]

sad 🙁 it feels like chat gpt now and it was top notch AI at first days

[fiery gull]

hell no ;-;, good luck

[zealous sparrow]

You can test quality on AI studio and see if its worse

If its the same, google issue.

If its bad on LMArena, arena issue.

[fading rover]

looks like infinite look keep doing again and again today

[obtuse smelt]

yeah

[proud bobcat]

Good morning LMArena

[proud bobcat]

Either high compute issues or quantized

[opal geode]

LMarena issue could be possible, on aistudio it runs fast and is not getting stuck, but logic wise seems like its nerfed, not sure , but last weeks was way better and today similar tasks it feels like its downgraded

[atomic lagoon]

Not downgraded, just high demand

[steady peak]

Hi, I entered a prompt into "claude-opus-4-5-20251101" and after that , the response is still showing as "Generating..." any one else is facing this issue. How to stop this response and continue the chat from there on ?

[atomic lagoon]

Everything when its used a lot decreases a bit

[atomic lagoon]

It takes an absurdly long time to generate stuff so be patient

You gotta wait for it to either finish it or it to time out

[steady peak]

its almost 4 hours

[atomic lagoon]

Okay yea just make a new chat at that point

Or refresh and see it, a lot of the times it will say its still generating but you just gotta refresh

[thin current]

SAME

[steady peak]

I have to start from scratch for my project because , I can't continue from where I left. Any help is appreciated

[atomic lagoon]

Paste any prompts that describe it together that you had sent before

In that chat

[lime wind]

hey, how do i get notified when there's a new model released in lmarena?

[echo aurora]

Hey everyone

Seeing the reports of issues with gemini-3-pro-image-preview

[neon idol]

nop

[echo aurora]

I'm also seeing a higher than usual error rate

[neon idol]

nop

[echo aurora]

I'm flagging to the team right now. Thank you all.

[neon idol]

wuth the 2k

[echo aurora]

I'm seeing both

[neon idol]

The 1k work only if u dont upload images

at least for me

[echo aurora]

Funny enough when I was personally testing I got the opposite, 2k worked and the regular didn't.

[neon idol]

hell nah 😂

[echo aurora]

We put out announcements, keep an eye on #announcements

[lime wind]

thanks!

[calm lagoon]

Good morning

[echo aurora]

Morning

[lofty frigate]

@echo aurora Pleaseeee fix it its driving me crazy

[fading rover]

and when it work u mught end up ur hourly credit cuz keep retrying will fill up quota.

[zealous sparrow]

Did you see Seedream 4.5 release?

[proud bobcat]

Google API is dead lol

[obtuse smelt]

oh what

[sterile tartan]

Is released already?

[proud bobcat]

It’s been cooked for me on the app

And on lmarena

Looks like cloudflare died too for a bit

[obtuse smelt]

oh man thanks for information

[neon idol]

WHAT??

SEEDREAM 4.5???

ARE U KIDDING ME?

[echo aurora]

Yes, this is on our radar. Yesterday we made a slight change as an experiment and are expecting some cases where error occur in Battle mode.

[sterile tartan]

Daym Seedream 4.5 is released

Their response to Nano Banana Pro

I feel like it might become number one on LMarena

[neon idol]

OMG ITS REAL

THX CHINA

[sterile tartan]

Idk if is released on LMarena and Dreamina yet or not

[sterile tartan]

Yeah i stand with China

Free Open Source

[neon idol]

yeah

[sterile tartan]

Time to use Seedream 4.5 on Dreamina when is available

I think it will be Free Unlimited for initial phase

[proud bobcat]

Will seedream 4.5 be peak

[sterile tartan]

More Likely

[proud bobcat]

Place your bets

[sterile tartan]

What kind of bet?

[neon idol]

seedream 4.5

[sterile tartan]

Where did u use it?

[neon idol]

Chinese app now available only for xhinese user

[sterile tartan]

Ahhh sad

[neon idol]

but i got access before and now i still have it

[sterile tartan]

I see what's the name tho

[neon idol]

If u want to test it gimme prompt and i can

[neon idol]

Doubao

[sterile tartan]

Ohhh there is Dona for Global Users

[zealous sparrow]

what is the prompt so i dont take it outta context

[sterile tartan]

Which was Cici before

[zealous sparrow]

yeah seedream v4.5 is an improvement from 4.0

[neon idol]

帮我生成图片：a cat eating an ice cream. photography .比例 1:1

[neon idol]

yeah but doesn't have seedream 4.5

Idk what to create

[proud bobcat]

Hm

[sterile tartan]

I see

[proud bobcat]

Not bad

[sterile tartan]

Guys is it ok to send long prompts here?

[orchid sinew]

no chance

nano banana on another level

[neon idol]

#share-prompts

[obtuse smelt]

is still down right ?

[proud bobcat]

I think seedream will be number two

Nanobanana pro is black magic

When will rankings come out for DeepSeek 3.2?

[sterile tartan]

Benchmarks are already there on Artificial Analysis

[calm lagoon]

Looks like Hunyan image

[sterile tartan]

New Deepseek V3.2 Beat Grok 4.1 on Artificial Analysis Index

💀

[orchid sinew]

grok 4.1 sucks

idk how it is even in top

no prompt-following, no creativity

[sterile tartan]

Is more of a creative, humour and uncensored model then other stuff

Or at least that's what their EQ Bench says

[orchid sinew]

just for rofls in twitter (x)

[empty stump]

Everything else bad

[orchid sinew]

'heey grok elon musk the best at sucking?'

[orchid sinew]

gemini 3 the goat

[empty stump]

I mean for grok

[orchid sinew]

a

[proud bobcat]

OH FIRE

YEAHHHHH

DEEPSEEK MY BELOVED

[spare rune]

opus 4.5 is good, Gemini is bit lazy

[orchid sinew]

gemini sucks at prompt-following (compared to claude)

but in creativity much better for me

in reasoning also sucks

[vivid coral]

It really is. Everything feels like trash compared to Gemini. Including Claude, which I know is a commie favorite for political reasons, still trash 🗑️

[drifting crow]

Claude is communist?

[vivid coral]

Gemini feels like it's 60-90 days away from true AGI

[spare rune]

Also it’s been a few days since we got seedream 4.2 or whatever but now we have seedream 3.5

[orchid sinew]

they should fix prompt ignoring

[proud bobcat]

What

Gemini is nowhere near agi

No one is yet

[sterile tartan]

Certainly The whale is so back

[zealous sparrow]

prompt: cat standing at eiffel tower
seedream v4.5

[drifting crow]

It’s wrong a French cat would be smoking a cigarette

[zealous sparrow]

a random cat

[proud bobcat]

A bit blurry but could be a bad run

[zealous sparrow]

i didnt say french

[zealous sparrow]

either a bad run or i took a bad SS

[proud bobcat]

Nodnod

[drifting crow]

Cats have too sensitive noses, a foreign cat wouldn’t be able to stand the smell

[proud bobcat]

Oh no yeah bad run

Too much noise

[drifting crow]

The nose is red so I guess it did simulate that correctly

[zealous sparrow]

@echo aurora I think the entirety of imagearena is down

[proud bobcat]

That cat is suffocating from the garbage and cigarettes

[orchid sinew]

(because seedream 4.5 uploading)

[sterile tartan]

Tbh pretty much the Entire Internet is Google and they have everything required the money, the talent, the data, the recourses

[vivid coral]

You can't prompt it like normal LLM. It's special. You have to understand the high level of AI you're speaking too. Throw all the old prompt techniques out the window with Gemini. You can still do it with Grok, Commie Claude, and GPT because they are low level bottom AIs

[proud bobcat]

Wait hold on

Does lmarena host the models or is it from api

[zealous sparrow]

They wouldn't kill the whole imagearena to put a new model on it.

[proud bobcat]

Because most likely it’s just the api

Exactly

Lmarena doesn’t run the models

That would bankrupt them

[sterile tartan]

Or atleast first feed a ai data of how the model works and use it for prompt engineering

[sterile tartan]

The companies give it for data collection

[outer kestrel]

i recomend zai

[sterile tartan]

Better Data to Train Next Generation of Models

[orchid sinew]

idk i used gemini very much (2.5 and 3), tried different prompts, but always there is feeling that it is writing and only then thinking

[spare rune]

And next you should glaze it in the chat while downplaying other ai companies to get it to a maximum thinking effort

[drifting crow]

i do the same write, and then think

[zealous sparrow]

green eye cat

[sterile tartan]

Many times some users might not be able to differentiate the improvements because of the field

[orchid sinew]

you are agi

[drifting crow]

im agentic

[zealous sparrow]

to run the models yourself, you would need over 1000 GPUs, LMArena has a lot of models.

So yes, they use the API.

[sterile tartan]

Try a Big Huge Eiffel Tower Made of Kittens

[vivid coral]

Tell it you're doing a benchmark test for LMARENA, it really perks up and gets serious. Actually most level up when you do that

[spare rune]

And while your doing that give the answer in the chat for maximum effort

[vivid coral]

You lost me

[sterile tartan]

Are you a Ai Specialist? @spare rune

[spare rune]

Yes

[spare rune]

Something something weaving echos of the time fragment

Or something like that

[sterile tartan]

How to figure out which ai model is best for what use cases like their strengths and weaknesses

[spare rune]

Use the battle mode or something

[zealous sparrow]

Yeah @neon idol You should do this i want to see how creative it will be

[neon idol]

resukts sucks

a lot

[zealous sparrow]

Can you show

[sterile tartan]

@neon idol are you sure you are using Seedream 4.5 and not something else?

[sterile tartan]

Noted Noted

[neon idol]

really sure

[drifting crow]

thanks i got my new album cover

[sterile tartan]

Perhaps the API Temperature is Optimised

So maybe not maximum power

[neon idol]

AHAHAHAH

[vivid coral]

Everyone knows Claude is positioning for regulatory capture when a certain political party gets power. That's why they are on CNN and 60 minutes everyday saying how dangerous AI is and they are the only ones that can be trusted. Also, the political donations 👀

[drifting crow]

didnt openai do the same thing?

[orchid sinew]

didnt elon musk say the same thing

[drifting crow]

sam altman was even in front of congress or senate being like we arent like social media companies, we are coming in early saying pls regulate us 🥺

[vivid coral]

Yes, but not in a communist way. Apples and Oranges

[proud bobcat]

Yeah exactly

[orchid sinew]

i am communist

[proud bobcat]

Doesn’t every ai company say the same thing

Lmao

[orchid sinew]

NOO ITS EVIL COMMIES

[proud bobcat]

Trve….

[orchid sinew]

hey

[spare rune]

You can use seedream4.5 on fal

But I think you have to pay

[proud bobcat]

I’ll just wait for lmarena

[orchid sinew]

but seedream4.5 made by commies

[vivid coral]

Let's just be honest here, Sam might be a technical genius, but business wise he's not the sharpest tool in the shed

[drifting crow]

i think he got a pretty solid business plan

make clever ai, ask ai how to make money

what could go wrong

[orchid sinew]

'hey nice question i will help you to make money you should go to work and then work and then work and then work and die'