#general
1 messages · Page 212 of 1
I have ai pro anyways tho
Ling long
riсk long
yes.
bro what
im sigma
or polska
No
who are u
im sigma from ohio
there is no 2 sigmas
noo
your cooked
It was good knowing ya
im calling the pineapple
i ll be back
We will remember you
Put your doctor in the song this time for africa
🥶
Farewell
I can’t upload, why???
Why not
Ow, now this ok🙌🏻
This chat is crazy
like they are too humans
Don't forget to make him jiggle, giggle and wiggle
its true
Hypocrisy
🐵
protect them from what
technically I had him do this already (due a series of events he has rose back in this and he was comeplled to dance with her while she lost her ability to dance leading to a lot of stepped on toes)
Mine isn't ai
<@&1349916362595635286>
i will send them exclusive in dm
We need someone from @swift oyster to come spam here
Prove us wrong
That their model is best
gemini 3 flash is in lmarena now
what is gemini 3 flash
flash variant of pro
seahawk in battle mode only
pls sorry me brother i dont liar its me liar BEG YOU
oreshnik better i think
cap
nah its seahawk
They need to assign a better namer Ai
Man, can't we ask for anything that requires totally SFW words? What's with this nonsense about censoring punches, kicks, etc.?
codename are for testing, not for public
Yeah but at least name them better
naming is subjective. i think seahawk is fine
i dont think that name 'poop' will be subjective
Yes, I do, but it still censors it. I have to use antonyms or synonyms, and even then it's annoying because sometimes it detects that as "bad" too, lol.
How about WhaleSky
name has no meaning for testing
yes on battlemod
Alright paw
test it
no
you tell me first
then i will te st it
why are you like this
:/
i trust you
He was trained on that kind of data
is gemini 3 flash live
We need fine tuning
nothing
bad model
It's like his life
have yall tested Seahawk and Skyhawk?
No but some guy said its Gemini 3 flash due to the naming
The models are new came out not even an hour ago
but almost 1 hour
Yeah also frame-flow was changed i think
It will keep on saying its ERNIE it used to say google before
Prob is ERNIE
Micro mango is MiMo and dark dragon is ernie too
Seahawk claims to be a google model @ocean ferry
cuz its gemini 3 flash
any good? or not worth talkgin about?
We have no proof till someone prompts coding examples
Not sure i will prompt a coding example
Models only on textarena tho
Unless you can prove its on codearena idk
how good is seahawk
https://codepen.io/qdpuzepc-the-styleful/pen/qEZLdyq
It went for a Glassmorphic or whatsver effect when i asked for a cool website
Skyhawk here
code?
https://codepen.io/qdpuzepc-the-styleful/pen/emZbNLm
Seahawk here
Textarena its not on codearen
Atleast not to my knowledge
looks clean
but its too easy
give it something harder
i still couldnt get the model
will try later
seahawk too good
beautiful 3d molten core + params to control it
to be true
Nah i doubt it will do good give me a sec
alr
sec
or an hour
i can wait
no worries
gemini 3 flash takes so long?
idk ask pro pro
$10,000+ worth of lifetime unlimited API keys on the line.
Rules (72 hours only):
- Cook the most insane, creative, broken, beautiful, or useful output you can with Tensor 1.5
- Post it publicly on X
- Tag
@swift oyster
- include #TensorChallenge
Wdym
Do u have seahawk rn
yes
Are u the same guy crashing out over sharing the chat?
why
Wdym cause there’s a guy also named socky
where
yea
Different server
?
invite
How do u know 3 flash
hmm i see
its seahawk in battle mode
Like u already revealed?
it says its from google
Knowledge cutoff
january 2025
Is it fast
@torn mantle
https://codepen.io/qdpuzepc-the-styleful/pen/wBGRaRQ
yes faster than pro
Do a Xbox controller svg
this is not bad actually
is it fast?
like really fast
faster than 2.5 pro at least?
yes
It was racing 3 pro and they finished at the same time
wtf
The gen was identical too
its not identical to gemini 3 pro time
gemini 3 pro
its because of lmarena, they make every model end at same time
Gemini 3 pro gen https://codepen.io/qdpuzepc-the-styleful/pen/azNPOXm
when gpt 5.2 coming
No some models take longer some less
@torn mantle basically a 1 to 1 between skyhawk and gem 3 pro
December 9
because its from google
what
Forgor when 5.2 was goin to be released
better but not by much
but its good
flash 3 seems good
gimme more outputs
I say it gets a YOU TRIED! star https://codepen.io/qdpuzepc-the-styleful/pen/wBGRaOQ
Ok ima ask it for a flamethrower simulator actually
its not bad
better than chinese models
cant believe flash >>> all chinese models
this is crazy
👋
Google’s latest imaging model, Nano Banana Pro, has sparked global concern after a series of viral incidents highlighted how advanced AI-generated visuals could be misused across identity verification, education and online refund systems. The tool’s hyper-realistic text and image outputs have triggered renewed conversations about fraud risks...
👋
whats with this ad
It’s from the same people that make the Internet archive
They’re trying to show the Internet of the future based on the way current tech technologies is going
Pretty cool experience
won't be big enough of an issue. PLUS qr codes won't work big give away
i think this was made a few years ago but it's as relevant as ever right now with governments trying to restrict what websites people can access
It gets pretty scary too because once you get past all this
You get hit with this
You guys can play around with it yourselves
that could very well be the future considering the UK's online safety act, sure it doesn't go that far but the facial scans are close enough. could very well be a thing considering most phones have a fingerprint scanner (retinal scans on phones were tried in the past, but they were abandoned), and many laptops have one too
Apparently, Reddit doesn’t exist in the future 🤣
Yeah, that’s what makes it really trippy
ai might've destroyed it by then
dead internet theory is slowly becoming real
Call me crazy
The Colonel warns Raiden about the plans to use AI to censor the Internet.
An experiment in creative writing and AI speech synthesis, inspired by the famous "Selection for Societal Sanity" (S3) codec conversation from Metal Gear Solid 2: Sons of Liberty.
SHORT FOLLOW UP VIDEO: https://www.youtube.com/shorts/Q_FUrVqvlfM
"And it will be monitor...
This is exactly what this video says and this is from 2023
I know it’s supposed to be a parody
But still
it already is
Every time I see stuff like this this is the video that always comes to mine. I wish I never seen it.
Same here
You know it’s gonna be super crazy though
Later generations, even after Gen Z
Will be born, integrated with all this technology, and they won’t ever know what the past was like
And they will assume this to be normal, the way life always was and the way technology always is.
And regardless of what people say everything you post online always get saved forever
I mean 20 or 30 years from now it’s gonna be hard to recover some of this information
what about for people before computer
They’re gonna be the new boomers
They’re gonna be considered relics out of touch
Just the way we treat older people and the way we view older people who use technology that they’re not just a sophisticated
Even the term or meme boomer is the only more proof of this
every generation believes they are right.
Well, perhaps they are the problem is isn’t if they’re right or wrong that’s kind of irrelevant
But it’s the cycle that returns back so the way you treat older generations is how the younger generations will treat you when you’re older
we dont know if we are right, hindsight is 20-20
Well, in many aspects, they probably are right
Because each new generation that’s more familiar with newer tech, technologies, and more integrated with it they would have a better understanding than the previous
I think the more fundamental issue would be that if the new generations respect the point of views and opinions of the older ones enough to heed caution I guess
Film and text by Maximilien Van Aertryck and Axel Danielson:
The camera is a tool — but to do what? Images shape our daily life, yet we rarely question how they’re made or why.
As filmmakers, we’re fascinated by how humans use cameras and by the immense influence images have. For 15 years, we’ve investigated the history of the camera, a...
This is a really really really really great video
About the invention of the camera to where we currently are with ai
Listen to how the lady describes the camera when it was first invented
“ It’s purpose is to give evidence of fact”
Which brings me back to why I posted this
You know it’s funny about that video. I posted up there..?
That ID he use looks weak and ppl freaking out
regular people dont keep up with tech
I made this this morning after watching that video
But it’s true what they say about a low trust society
And so I think and the whole reason for sharing all this is I think that after looking at that forward Time Machine I think that this is inevitable
And I agree with your statement 100%
It’s already happening
Ugh what mess we have on our hands, gentlemen
What happened? (I told it to generate glitchy text and it gives me this, I tried several times and other time does not look like this at all, it kinda look like text messages?)
what model made this
(don’t mind watermark at the top right 💀)
guys guess the model
Bro copied me
is veo better than sora right now?
no
os lmarena are bad again ?
I love this
Wait
Let him cook
The deepfake and image restrictions could be great
dont have to that email anymore. chrome auto signed me out after update.
damn.
just did
but DAMN its bad.
not what i wanted at all
Bad?
It’s not better than 2.5 pro
Also it’s not a coding model
wait for gpt 5.2
No longer this week apparently
source
opus 4.5 thinking
is there an actual way to edit messages
why does NB Pro restart itself? this platform is extremely unstable!
Hopefully. Its just all so toxic
Honestly, DeSantis is a good Governor and hes done a lot of good things with blocking nsfw sites in Florida from minors and stuff. This will also probably be good lol. Putting parental controls on AI could mean AI will become more unfiltered and allow more things in the future
gemini 3 pro!
Wow
testing something..
I know it looks bad but im trying a new format
theres a lot of detail
I can confirm that this is fake. We've not run it on Vending-Bench. Sorry to spoil the party.
is gemini 3 flash good?
what abt skyhawk?
Hello coming here to take the pulse of the video IA community
what is the stuff about gemini 3.5
31 dec
it's a betting it will not always be accurate
Why
wats up
I just got to know google is directly burning money to compete with anthropic
Opus is Available in Antigravity and The Ai Pro plan is Cheaper with Generous Limits
<@&1349916362595635286>
Someone said they think Google is hosting Claud themselves
they are
anthropic is even buying a million of their tpus
do i need to pay to use lmarena?
not use
limits are (average, depends on model) 5 per model per 50 minutes, or unlimited in battle mode
flux and reve always show up until I reach rate limit lol
ah battle why is say: Something went wrong with this response, please try again.
it can be rate limit I guess, it reset every hours
i see
Not possible
What's happening? It's saying "something went wrong" on gpt 5.1
Is it?
I'm using 5.1 gpt
Is not open-source
@sterile tartan is all ai chat working? I tried using other model aside from gpt 5.1 it keeps saying something went wrong
Obviously not
You are probably bein rate limited
Google owns a large portion
U can try Yupp
yupp?
14% to be precise
Just like LMarena 800+ models
free or?
But have to give good feedbacks because it had a credit system
Yeah u can open cash out real money if eligible
4K is absolutely much cleaner and clearer than 2K, they are totally different breed
So it’s possible
ooh u got the link?
Nope Anthropic wouldn't give google the models
Just search yupp ai
Yes
They have a deal for like 1 million tpus I think
Yeah but that doesn't mean they will give away their model weights
Imagine one day google buys both Anthropic and ChatGPT
💀
They’re not giving it away
They’re letting them host the weights
Giving anthropic limits not based on api costs but for free and based on plan seems weird
Currently free
And for the foreseeable future
From Google
They’re just paying anthropic?
Exactly
thnx :D btw what can i do with those credits?
on yuup ai
Use it for Ai Models or Cash out if eligible
I would think they would add api to it if it were that way
If u get better luck next time switch the mode from text to only image or image to text only
If still better luck next time try after some hours if still try next day
ooh?
Yeah is kinda like a game
I know the ins and outs
im just making images atm
U have a point but we just can't be certain
does yuup ai have community?
I just think it’s possible
They definitely have deals going on tho
True indeed
Yes
Discord
Make sure to give proper feedbacks
Bad feedbacks are punished
What’s yuup
Another LMarena basically
Is it good
huh?
Good feedbacks more credits
ah i see
Bad feedbacks less or no credits
Wow gpt 5.1 is #1
ok
I recommend writing little bit of feedback as well
can u send me discord link in dms for yuup ai?
Sure
I mean u can click his profile
real

yeah just click on their "Yupp" server tag
💀
Haii
hello
what happens after you have used all your credits?
locked out forever
they try to make it hard for you to use up all your credits, they give you more a few times but i assume if you finally run out you can't use yupp anymore? idk
need new account
but it is very rare as you are expected to give feedback
real
Just don't burn out all credits if u keep getting better luck next time
Think if it as like a game understand the algorithm and outplay it
bro making food

hmm
you mean 16:9
try edit photo in google
You mean download it and manaully change it to 16:9?
why does gemini never the search web until i use deep research it doesnt have a search the web option what
In gemini app?
OpenAI is in a full internal race after a major leak revealed a new model called Garlic outperforming Gemini 3 and Claude in high-stakes reasoning and coding tests. Apple quietly dropped CLaRa, a memory-token system that compresses huge documents into tiny, high-accuracy representations. Microsoft finally delivered real-time AI voice with almost...
Lmarena mentioned
just got evo-logic. Looks bad. Is it gemini 3 flash?
one of my accounts ran out of credits. but I check a couple days later, 3000 credits were added without any notifications.
Banana pro laggy&high error rate, any fellows having similar problem?
there should be an email sent to that account, I got one when it did the same to me
#1397655624103493813
<@&1349916362595635286>
It’s a
Something went wrong with this response, please try again
kinda day
well, thats looks much more interesting than 109371298379213797321 variations of minecraft clones
Y does veo3 suck so much ?
Not real
F/F feature is trash
Veo 3 is good for prompt follow but results are a little plastic
That f/f feature is horrible Kling and hailuo are way better at f/lf
American companies are not going to win the ai race without a doubt.
The Chinese ai is way more reasonable with content moderation to some extent.
hi @swift oyster, check dm please.
@deep adder
?
why u forward my mssg?
👀
website
why does it not search the web it makes me sitll want to use chatgpt
Tomorrow is gpt 5.2?
I canceled my sub to them i was subbed for 4 months
Yes it's great but gemini is better even if aistudio doesn't have good ui as chatgpt does
For ChatGPT i just use Free Tier, Copilot or Polaris Alpha
Gemini is 6x value indeed
Even can use GPT on LMarena, Yupp or such platforms
pretrained transformer
I really like seedream as an image AI
Still haven’t found a video AI that I actually enjoy
They have added Nano Banana pro 2k
how can I write a prompt for endless motion?
so I will create gif from video
but its transition should be for endless motion
🤔
I wanna make endless motion with wind blowing on this image
Try to add "loop" or "looping" in your prompt..
" animate it. natural motion like there is real wind blowing. make it looping. " is good?
5 should be coming soon
AI LLM
Try this:
"Generate an endless looping video featuring the wind blowing vegetation on the image, make the movement as in real nature."
hello
Gemini 3.5 is coming
it still belongs into the Gemini 3[.x] pro category, in above poll
Google might have just leaked its most powerful AI model ever… and it’s not Gemini 2 or even Gemini 3 👀 New mystery checkpoints showing up inside LM Arena are pointing to what could be Gemini 3.5 or an upgraded Gemini 3.0 Flash — and the speed, cost-efficiency, and raw generation power are on another level.
🔗 My Links:
Sponsor a Vid...
Not 100% if It's the model, but hopefully
No way
This is not possible simply
Or were they training 2 new models in all these times?
HELP ME PLIS https://empiredrop.com?r=IR5YRT5HYL
bruh
you guys were saying flash
but it could actually be a better checkpoint for gemini 3 pro
remember what we have is only gemini 3 pro preview not the GA
just seconds to the video and i can already see it better than the current one
so could be possible
and its mainly available on code arena
who said it wasnt
smh...
Some are saying is Gemini 3.5
I think they are playing a number game
Seems like they are done making the final optimizations
You are right smarty pants
It has to be Gemini 3 Pro Final
i had my doubts yesterday that it should be added on code arena
but someone said nah
Guys i have a theory
skyhawk - either gemini 3 flash or flash lite
seahawk - either gem 3 flash or flash lite
its not flash
look
didnt you say yesterday it has the same speed tokens with gemini 3 pro?
I dont trust it..
It was one gen im not sure if its every
also seahawk was randomly readded and added
so maybe its on codearena now?
nvm theres my explaination [from the video not mine]
@torn mantle I have a huge theory
Seahawk is actually a gemini 3 pro buffed checkpoint
oh and skyhawk is on codearena now too
Meta has open-sourced a complete perception system—SAM 3 and SAM 3D—that bridges Visual Intelligence and Spatial Intelligence, while tackling the long-standing 3D training data problem in AI.
SAM 3 is a unified vision model for detection, segmentation, and tracking across media, using text prompts to find any concept, even rare ones, from i...
interesting
You remember when the result was just identical?
They can't train a new model to copy an existing one..
Also remember that google said
they have their cards ready
yea
i think this version will be more reliable and better instruc following
#general good morning folks. Just trying to figure out this AI ad generation for my business to increase leads. still havent quite figured it out. but hoping seeing everyones prompts on here and trying out the free tools i might get it done.
alright if we know seahawk is a buffed gem 3 pro, then what is skyhawk..
Skyhawk might be flash?
the actual gemini 3 pro is powerful but its not good at following prompts and lazy too + bad instruct following, so maybe they fixed all of that
But i also cant see it being gemini 3 pro.. I tested the impossible steganography on skyhawk or seahawk i forgot, and they got it wrong.
I didn't test one of them tho.
Wounder how they caught him
If seahawk or skyhawk solves the impossible steganography, i will believe one of them is a gem 3 pro model..
“The trader's near-perfect record of 22 correct predictions out of 23 attempts has sparked accusations of insider trading, raising uncomfortable questions about prediction market integrity just as Wall Street money floods into the sector.”
Why? Wish bad on ppl lol
You know its crazy how people are saying google is already making Gemini 3.5 [seahawk]
but dont take in account it could just be a Gemini 3 pro checkpoint
People love the hype gravy the thicker the speculation the more they crave
We are going to mess up the economy
Back to gambling lol
With State approved backing
In my opinion, out of all the devices gambling by far is the most hazardous
they could be flash
idk
im lost tbh
I refuse to drop my words that seahawk isnt just a gemini 3 pro checkpoint
i need to test them
or even flash
also glm 4.6v dropped 1h ago
glm 4.6v flash look sooo good
:D, I'll test the flash version, like the 4.6v flash 9B is toe-to-toe vs 4.5v 106B 👀
10x smaller, soo good
gpt5.2 or robin in lmarena
is not practical
its so slow
it really seems openai has hit a plateau with scaling
where my gpt 5.2 I need some thing better, I'm tired of opus 4.5 ;-;
Heavy hand content moderation for gpt?
UI*
then gpt 5.2 is another flop
if robin-high was gpt 5.2-high
good backend
horrible frontend
Seahawk must be Gemini 3 flash
They are done with 5
They’re just trying to maintain it the best they can
people are mixed between if its 3 flash, 3.5 pro or 3 pro
Resource I’m sure whatever’s left on 5 is just a small team
He even revealed who he was.
5.2 will be the last, 5 series release
They’re not gonna invest any real money into 5 at this point
The whole five series of flop
Nobody should expect anything from it than what it already is maybe minor tweaks
5 was good, but upwards from 5.1 its all a flop
I don’t know how you could say that about 5
In my opinion, if I was a complete downgrade
Sure, they improved on coding and codex
skyhawk, just got it, seems more like flash ngl
Its Chinese isn't very good, so it's maybe a Flash model.
what do you think bout seahawk tho
skyhawk is so fast
didnt get it yet
could be thinking ON/OFF
skyhawk maybe with thinking off not sure
it was so fast
I wonder what garlic is gonna be?
yo holy sh thats the rightest theory
i was also thinking that
Just garlic
its not a flash lite model
and that one of them is 3 Flash thinking and one of them is 3 flash Nonthinking
Garlic has to be six
Why you can't use Nanobanana anymore
wdym
69?
Something when wrong
did you prompt the same prompt 3 times
im trying lmarena but the thing is that im getting some sloppy slow model thats pissng me off
btw
its so slow
ah
im so greatful they are gone because you get too much of them
From what i see
if its flash then it will change the market again
oh skyhawk does have thinking ON
its just hidden
and messed up
it got in an endless loop
which makes me think its really gemini 3 flash + thinking ON
So bugged
higher chance its flash then
https://019afe49-5695-7afd-a053-07f368d1c405.arena.site
skyhawk again i kinda like this one
It made a roguelike yoo
what was your prompt
Generate a doom game, but with roguelike elements name "HELL LOOP"
pretty good for such a simple prompt
what model
Gemini I guess
no skyhawk or seahawk
Was this also a really simple prompt? Create a papers please type game?
Probably be a stupid question. But does anyone feel like Google insanely nerfed nanobanana 2.5? Like almost not usable in some basic editing scenarios.
im unsure if the prompt wont trigger the uh automod but lets see
It was quite simple. : make papers please 3d but you have a gun
seahawk
battle mode models that are anonymous yes
direct chat has restrictions that are ratelimits for certain models
Damn open ai been patching all my stuff
ok, Does chatgpt 5 has limitations?
Why does Discord said that you are a spammer lol
wdym
Don't know
My Sora jailbreaks
Perhaps i am new here
Patched overnight idk why
Or how
2nd time
I think they have certain accounts with different levels of restrictions
imagine an openai employee is just in here
Yeah. That makes sense.
You real think an OpenAI employee isnt smart enough to recreate it
Not to mention you prob talked bout it
GPT 5.2 will release tomorow?
OpenAI have age restrictions lol
Hah fr
Did you found a new way?
Naw I’m bout to leak this method
Which will get patched
I think I’m gonna be taking a little hiatus from AI here in a few days hopefully
Anybody know wrf this is?
“heavily dramatic or fictionalized portrayal “ here is a piece of Sora/ img prompt that really helps when you’re trying to find some discoveries. You could throw it in anywhere in your prompts and sometimes it makes the biggest difference
I got ai fatigue
i see
You start feeling robotic yourself after a while and it’s an expensive hobby . 🙁
It's not expensive if you can just chat for free
You become what you consume
I don’t really chat with them. I’m just addicted to making videos and images.
this type of reasoning is actually very similar to gem 3 pro when it bugs
And I don’t even have a social media to share anything with and I never post anything anywhere so it’s useless lol
Jason is back with a new This Week in Startups… and a dire warning for founders.
Sam Altman is watching your use of OpenAI very closely… and while it may seem fun to get a trophy for using the most tokens… you’re also setting yourself up for a potential fall.
As Jason explains, companies like OpenAI watch API use very closely, and if t...
Damn, how true is this? I never even thought of it like this.
If anyone plays deltarune, it managed to make a chapter 1 and 2 savefile editor
https://019afe77-614e-7427-b778-592f8e0ad49e.arena.site
[skyhawk]
Hey yall I have some questions
Is there an i2i here
Hey yall please I have a question can't they make this discord have image to image and can't they make you choose your video model they should have an arena dedicated for that
They are definitely selling and training on our data
Well, it’s interesting the API thing
I mean, who’s to say they can’t steal your business idea. They’re obviously gonna know what’s popular before the masses. They have all the data for the API and how to used and who spends the uses of the most amount of tokens
Plus open AI has a long track record of copying or cloning or whatever you wanna call it
Data is everything
And they have Ai to Process it too
Calculating and Predicting everything
If LMArena does too, why get mad at OAI?
Why get mad at anyone if it has been happening before Ai anyways
Exactly...
Privacy in internet age has always been a illusion
Why almost everything on internet is Free that's why
U accept those cookies yourself
generate a 5 second catoon with cat video
Generating...Please wait
👋

nop
kinda in 1 week or 1 week and 3 days
as well as gpt 5.2 openai will release an image model that is better than 🍌 next week.
happy sunday
@brave nacelle Please head to #1397655624103493813 for a detailed guide on how to use the bot. Also, you should use Video Arena channels to generate videos. Thanks
Trst
Seahawk seems decent. But not Sota. Gemini 3 pro is better. Sonnet 4.5 is comparable
(Non coding questions)
When? December or in 2026?
More likely after 2026 but it could come sooner since the competition is getting cutthroat
Expect GLM 5 Soon as well
True. Competition is really heating up
Well is kinda Exciting
It is. But I can't keep up with all new models and advancements
U don't need to
Just track the 8 SOTA
4 West and 4 East
Gemini, Claude, GPT, Grok
Qwen, Kimi, Deepseek, GLM
Track Artificial Analysis Intelligence Index Benchmarks
LMarena and Yupp Ai Leaderboard
I am not going to use East models for anything personal or office related Information. I am going to stop reading about them.
Your 4 west models are exactly what I focus on! Spot on!
شاهد معنا البث المباشر لمباراة المغرب والسعودية في كأس العرب قطر 2025، مواجهة قوية تجمع بين منتخب المغرب ومنتخب السعودية في منافسات مثيرة ضمن البطولة العربية. تابع الأهداف، الملخصات، وأهم الل...
SOTA - Proprietary
17
24
1
Gemini 3 Pro Preview (high)
People don’t think like that
SOTA - Open Source
8
15
1
Kimi K2 Thinking
True
True. But east is more likely to abuse it. I am biased .. I know
Abuse what?
Do you realise East is pushing and forcing west to do better and cheaper
They are overseas, probably are bitter off using the east models thousand miles away
They are giving Open Source Free Models Unlike West
Using your own, we’re just here directly, which could be used against you by your own kind
The west invented the playbook who do you think they learned it from?
It’s American hardware it’s American software that’s powering most of the surveillance technologies throughout the world
I am biased I know. But I know how US/Europe companies work. I work in one. And I know how they treat Data and security.
About east I don't know. So I am just a bit averse of unknown
sonnet 5 coming next year
That's very obvious
I can’t find it
God ai is so clueless sometimes
Just makes sht up
Honestly for all the so called intelligence these it’s equally as dumb on so many basic levels its ridiculous lol
Mind u it has allegedly all the data from the Internet access to it so it makes so many mistakes?
Is anyone else having trouble w nb2
2k direct infinite error again?
0 proper response, always "something went wrong bla bla bla"
Both 2k and normal for me
Generate images, it would generate the first image every time I give it a prompt on the second image it just failed to connect a server
2k has been in trouble for couple days. Normal is fast and fully operational at Yupp.
at what?
It felt like something like a nerf, but I can’t really prove it so
wdym
I've been hearing some troubles with this model, some mixed reports between the 2k and non-2k versions.
2.5 pro is still on LMArena and AIstudio
Felt like 2k is all on Google
Api free, not lmarena.
Even when using ap or ai studio
They even reduced flash rpd to 20
You have to be logged in on the page and that’s not even guaranteed
250 to 20. Lmao
Like you can’t just run it in the background otherwise it might fail
And even that’s not guaranteed
It is in the AI studio on the Gemini app, but it’s if you put it in the background, it’s gonna fail for me at least it does
I checked my settings. Also, my Gemini app allowed it to run into the background.
You are right they could make it annoying so people use the API, which is what they want
Google was too generous ig, they give out so many free stuff and now it's costing them a lot and they have to tone stuff down
They not generous they’re just trying to get the market share
Consolidate all these perks are gonna go away
I'd still call them generous lmao
That’s not a wrong statement. Yet they need to market in that way in order to take away OAI’s share.
Dude, once the AI war settles down
Also look at how many fake student discounted accounts used at Gemini 🤣
Am I and who’s ever left people are gonna be so relying on it. They could just do whatever every other company does that we’ve seen.
Because the goal is trying to get people and this is what they want because this is all the money is the API
Because they get 2 for 1
They get you paying the API price directly to them, and two they get the data and say your developer you develop something and your customers are paying the API and you’re the middle man or whatever they get that data n cut also.
credit based economy
Because you have to pay for practically almost every API call in one way or another
Even if the output is what you didn’t expected and so forth, blah blah blah, they’re still getting a chunk out of it regardless if it’s exceeds, stay neutral whatever
I wonder how loveart managed the API stuff and product performance. The quality and extra edit function provided by it are just different. Though the company has some negative reviews on payment issues.
And if these companies could they would make all of AI API calls the only method
Because AI has become a commodity
No different than electricity except we’re paying for tokens as our usage
What do you think? Do you think they pay for tokens?
Electricity is there only cost because that’s what everything boils down to
Their electric bill
Everything compute cooling everything down to electricity even the water to pump it
Why do you think some of these API are so cheap that means they’re also still able to make money off it
bro i love how claude builds you a full project in one prompt
what project
Can someone help me? I can't send messages to any AI on the lmarena site. ...
Are you using Google Chrome? Try clearing you browser data
no
What messages are you getting that you can’t send a messages?
Error please try again?
"Something went wrong while generating the response. Please try again. "
ranked #469 in unrealized pnl leaderboard lmao
https:///%68%79%70%65%72%6C%69%71%75%69%64%76%69%65%77%2E%78%79%7A
Is it all or just the specific one?
Yeah, I will try clear your browser data. That would be the first thing I’ll try.
The other thing could be a rate limit, but if you’re saying all the models unlikely that’s rate limits
See
Lmao
I think of all the people and all the developers that were reliant on this and using this daily for their services Just got cut off and what are they forced to do?
use 2.5 flash thinking
Anyways people are using it
even in battle mode
flash is limited too?
According to this apparently
When does XP was launched in 2001 I think
They had extended service until 2014
Its cus tech companies got smart
And it’s expensive to maintain old software, I bet
they need to pay
Look I’m out for people making money. I’m even open for them making money, but whatever means via API however they gotta do it.
The thing that concerns me is that this business model will spread to all corners of the internet
U gotta check out the forward machine if u didn’t already
I do think limiting the use of AI by companies is a good thing, that way people who say AIs will take our jobs are wrong.
As you need to put money into using AI.
Coding yourself would be free and getting AI to do it wouldnt.
Well, that’s what we’re seeing with. These isn’t AI capable of doing his job.
But the people are cutting the jobs in order to get the AI infrastructure and set up
And they’re using AI
Amazon, obviously with robots and machines is a little bit of different
It’s physical, but the jobs we see being laid off or literally from the tech industry mainly
Hey all bs aside
Say ur bosss one day at ur job came in and said hey wear these ai smart glasses or other ai tool will u r on the job so we can evaluate your performance and u will be training our ai model which will replace u? What would u guys do?
I’d try to sabotage lol
it will be more subtle than that. and we will all do it without knowing it 🙁
I heard everybody to visit the way forward machine the link I just sent
I think we should try our hardest to preserve as much internet freedoms as we possibly can to the best of our abilities
We only got one folks
oh.. this sounds exciting : https://www.reddit.com/r/singularity/comments/1phhzxc/new_nano_banana_powered_by_gemini_3_flash_is/
Is it already released?
Not yet
Ahhh when will it release?
I’m not sure just by reading that post. I’m not mistaken. I’m assuming the guy has some kinda of beta testing!
Gotcha ty
That's Amazing
Do you see the way he worded it?