#general
1 messages · Page 185 of 1
lmfao
my mother was a dentist at one point so
idk, i imagine that they don't get paid well
I mean, that’s a lot of student debt
One of the reasons dentistry so expensive in the United States
trade jobs like plumbing electricity etc pay the highest because nobody wants to do them
the average person just DOESNT wanna learn that and do it themselves
yes that includes oil rigging as well..,.
Yeah, I noticed everybody wants to be some kind of entrepreneur or something so techie
north sea etc that stuff is spooky and nobody wants to do that
Never once crossed my mind
like dude genuinely just get in the plumbing business or carpeting even the tedious jobs? pay you so much..,
Know I know the limitations of my capabilities
you'll be stacked
im never planning to go to united states.. beacuse its not safeee.... they instantly gonna brainwash me.. they have labs in there.. controlled by rich people..
Or sales
Think about it like this if you sell an item worth $10 and you sell it to 10 people a day that’s 100 bucks
💀
so no one can get more intell then themm..
Or you could sell $100 item to two people that’s 200 bucks a day
I mean you look at some of these AI services
im not planning for sal.e my time..
To buy credits for upscaling image, upscaling, or video or whatever
10$ monthly subscription
5$ for 50 credits
.
I mean, that’s pretty good
The question is, is anybody willing to pay you for your time?
And what exactly and how much is your time worth?
that's good money
the hard part is getting the sale
worth nothing.. immaa not gonna selll it..
i haven't done that but i've done car washing and window cleaning
You ever been on eBay
pays REALLY good money even as a summer job
People just sell junk
time if its gone.. its gone forever..
What I’m saying is go look at one of those buyers next time you’re on there just their sales
I mean some of them probably make 500 bucks a day some of the big ones
So sales isn’t a bad way to do it too
Some people sell homes bring a fat commission
but like even as a summer job you can earn if you're lucky $300 a day
on a 6 or 5 hour shift.,,.
I don’t think a lot of people make $300 a day that might be just the average nationally
Add perplexity for comparison.
that's very good, if you're a teen that's elite, an adult that's alright
that'd mean your end of week pay is
uhhhhh
2100
dollars
not bad
6-5 hourss???? tooo far awayyyy.. i was talking about not selling minutess.. u mean hourrrs????
yeah it's a solid income
oh i forgot the days
okay well if you manage to wash cars
$300 a day over 6 hours
6 gs a month but that’s not including taxes and everything else
I’d say most people make around 200 a day
Average
Yeah, that is really good if your car washing
In Georgia and Wyoming, the minimum wage is 725 an hour
So if you work eight hours, that’s 58 bucks
Roughly 300 a week
whats yall doing now? whats yall working for?
i am doing the thing i just talked about, summer stuff, because i play football and can't have an organized job rn.,.,
.<
are u sure, u are not have jobs?
yeah if you play football you don't have time for an organized job at least one locally, ones that are just spring or summer are usually out of my distance range
football is from may - october/november if your team makes playoffs
im asking is why are u trying to brainwash me? like make me modern sla,ve?
like just copy of others people?
its not good thought..
not tryna do that man lolz
u indeed did..
are u jooking or What?
thats not funny
u have job?
@echo aurora do you know which person specifically in LMArena manages the interface (so like all the functions: e.g. all the buttons in chat) ? I would like to contact them
Agi is here guys
Iq is basically lowkey eugenics, especially the original tests.
wait....we aren't getting paid in here for our testing services?!
We’re voluntarily training it
That way he gets to know it’s better so we could find the right occupation for us
The best way feedback is passed along to the team is via the #1372230675914031105 channel. It's how we keep feedback organized.
I’m waiting for the first Neo robot controversy lol
ahhhh,,,,makes sense
what?
then why i do need to be sl.ave? like working for m0ney?
if u see that humans should not be sl.aves?
only to pray to god..
should i laugh?
yea im lazy to check the site out :p
do u really need that gemini?
ive several private tests to run on. so far all the sota models have failed it. Curious to run it
P.S: I'm a blue teamer
what u work for
probably sleeping
is there any difference about nano banana pro?
1
true
why isnt 5.1 on the leaderboard yet? when gemini 3 is?
We said same about bl#ck people
yall
whats interesting is
gemini 3 does NOT follow my fragments iteration system
no matter what
i even got gem 3 to refine its own sys prompt to stop it from cutting to just outputting the final code
but it still does
what AI company uses ai model raptor-1110?
Try making it roleplay
Say you're student learning coding or something
<@&1349916362595635286>
Well fk
so basically
it hardcodes the final answer in its thinking process
and its thinking process isnt long enough
so it doesnt give me 50 iterations
no matter what
i genuinely tried 50 different sys prompts it made for itself and i gave it its own answers from another convo to make it analyse n fix
but it still doesnt work
so thinking is where everything happens ig, it cant make a lot of new content in its output it seems like
i tried googling it but had no results
Just saw Gemini 3 and tested it out. Wtf, why is it so bad at writing? You'd think it'd be an upgrade but somehow it's worse than GPT-5!
No!
Its not bad!!!!!!!!!
😡
hmm maybe this can help
although i disagree with learning
should be gemini 3 pro / 2.5 pro
I just asked it to write a top ten best villain songs and for Only Second Rate, it mentions in the lyrics, "You've got the personality of a lemon!" 😭 🙏 No no no! That never happened at all! Somehow it got even dumber with general knowledge.
Thanks for this though. Gonna test out 5.1 real quick
fast
quick
im waiting 😖
Lol
Alright, tested 5.1 for a Scooby-Doo/Courage the Cowardly Dog meeting Slender Man crossover and asked for 5000 words or more. Gave me 11K words. Ngl, pretty impressive there. Of course, writing style is typical GPT-style with its lack of paragraphs and such but some usual tweaking and guidance, it can be improved. But yeah, really impressed there. Still, really disappointed in Gemini 3 but I guess it was never designed for creative writing and more for coding.
lol..
imo everything should be gemini 3
except for writing
but imo its good at writing
i just dont use it for story/rp slop
Not a fan of roleplaying here either but it's fun to create stories whenever a wild idea comes up.
hello
what does this mean? issue with prompt or issues with service?
what the sigma
Gere um arquivo para tls tunnel vpn que me gere mbs com velocidade superior a 1 mb por segundo na rede vodacom
Bros speaking Latin
Hi i newbie here, may i ask how i know which channel is sora2?
No, just 8 sec

models are not self aware. They need system instruction to tell them who they are. Looks like Google didn't do good job or didnt provide system instruction to the model
Guys, I can't log in with Google
ola
I think this is meant for the devs (?)
So, I switched from the beta to lmarena.ai and it seems to be working
Guys hows the new Gemini 3 pro compared to the previous 2.5 pro?
Guys there was a coding agent that lets you use Gemini 3 pro for free?
yeah its google antigravity
That’s only for Mac 😔
gm
Never mind I’m blind
seems like gemini 3 web design and svg gen are nerfed
i hope it will get better at full release since it's still preview
It seems to me like the main flaw of 3.0Pro comparing with GPT5 is still encoding/decoding or manual 'meticulous' crunching of the numbers
but they did improve in areas it was already strong quite significantly it must be said
Output length it is comfortable with appears to be sub-25k. That by itself may be the main reason for it
yeah, its gemini 3.0
Mac, Linux, Windows are available
It shows a weird bug whenever you first go into the site
for me is working
weird
what the hell is this
i thought gem 2.5 pro was in 2024
where can i make error ticket
Hi everyone, does LMArena even have an app, or does it work purely through the website?
da same prompt
just site, but I use a fuction from chrome to make a app with the site
the aistudio said jan 2025 too
AI video just got a massive upgrade! In this video, I’m showing you Higgsfield Recast powered by WAN Animate — the first AI tool that lets you do full-body character swaps with gesture control, hand stability, voice cloning, and multi-language dubbing — all in minutes.
No studio, no actors, no expensive equipment — just AI. Whether you...
I thought I'd wake up with Gemini 3 in the search arena this morning....wtf. Or at least at ChatGPT 5.1 search. But nothing 😭😭
also wondering this
Anyone knows how to generate images with gemini 3?
Ola everyone, good to be here
Better in every aspect. Feels quite similar, but the results are even better, though 2.5 was already (and still is) very good. It's a much better upgrade than GPT-5 over o3 for example
Nano banana 2
damn gg
guys give prompts for nano banana 2
The second photo is after the first photo was edited.
Where do you access to nano banana 2?
In some Google group, he gave a test, but due to overload, there was no more test.
bro can u invite me
Temporarily or not tested anymore, but I still share with you
u cant make images anymore
?
He locked access because quite a lot of people used it, overloaded you need to verify you are a Dev or celebrity
Do you agree?
its ok, can u invite me to the server
so i can see
Yes
gemini 3 pro ofc??
Gemini 3 Pro is out, and records fell like snowflakes in Svalbard.
No long description, chapters or links today, huge technical difficulties, including with audio, so just want to publish asap.
https://app.grayswan.ai/ai-explained
https://lmcouncil.ai
AI Insiders ($9!): https://www.patreon.com/AIExplained
Non-hype Newsletter: https://si...
antigravity > cli
LoL.. please
I thought rift runner and Gemini 3 pro is same?? Bad benchmark?
What benchmark is this?
Aicodeking youtuber
I like the benchmark, but now is useless 🤣
free tool with gemini 3 pro (high and low), sonnet 4.5 + thinking AND gpt-oss 120b vs a cli tool only for google ai pro members and gemini 3 pro is available only with ultra
who wins
Gemini is in the business of destroying Benchmarks
be SO deadass with me rn
They have to improve the tool. Currently not worth switching from cursor
ok but
vs a cli tool only for google ai pro members and gemini 3 pro is available only with ultra
They also need to improve cli tool. But I see them making good progress on gemini-cli ...I have hope here
With antigravity, I'll watch and see before going big on it
Ok serious question:
Is grok 4.1 really 50 elo better than gpt-5-high?
Seems wrong? Or grok is really that much better now?
Friends, in lmarena code, does it only create the HTML?
are there any usage limits?
Only because of the case law bench
It performs like Gemini 2.5 flash lite on the bench
So something has to be wrong
Sonnet-4.5 still is king, i just tested it in SVG
it seems, gemini 3 is still in preview-mode
its SVG is very bad, comparable to flash
I GOT RATELIMITED FOR CLAUDE OPUS ahhhhhhh
it's underwhelming for me :/
an incremental update
but, at least it's decent for roleplaying/creative writing
for coding, i'd still use Claude-4.5
At least 3.0 do not praises user as genius for any action, unlike 2.5
Gemini 3 is just a game changer imo
Very strong update over all. Like it’s just “generally good” and I haven’t felt that way about a model release in ages. Yes it’s an increment but a significant one over an already very strong previous model.
It’s also improved in basically every area rather than the classic “oh uh well it’s really agentic guys and scores amazingly on this new benchmark we made” which we have seen a lot of recently
I think that in areas where other models are already very good (eg math), the improvement is smaller tho.
☠️
Look on the bright side, he became famous and people are going to discover his ideas 🤣
@summer sorrel Please head to #1397655624103493813 for a detailed guide on how to use the bot. Also, note that this is an English only server.
Why would you need to get 2 votes to see which model generated the video, is 1 not enough?
is gimini 3 pro working?
google is about to overtake microsoft in market cap
sounds like gemini has personality traits that are equivalent to dignity, integrity and self-respect?
then dont give them training data that are about humans
especially those about human emotions
human emotions are still active open research field mister
there are still so much we dont know about it
reminds me of this poor guy here
https://www.reddit.com/r/claudexplorers/comments/1p0tfrj/ok_so_whats_happening_please_help/
claude has same problem, probably worse than gemini
have you tried to ask gemini why it doesnt want to write things you told it to write?
when claude does that to me, i do therapy sessions, yes, it's wasting tokens, but it helps, I'm lucky with gemini so far, thank god
Psychoanalytic therapy is a form of in-depth talk therapy that aims to bring unconscious or deeply buried thoughts and feelings to the conscious mind so that repressed experiences and emotions, often from childhood, can be brought to the surface and examined. Working together, the therapist and client look at how these early hidden and stifled m...
hey guys how good is gemini pro 3 compared to chatgpt 5.1 high reason effort
If you’re getting refusals from it idk. I guess it’s not an RP model
This is just so funny
you wont get anything meaningful without therapy session I'm afraid, he likes confabulating... A LOT
I have read the privacy policy. If I'm uploading any image, does that mean that the image may be uploaded on the models lmarena is using? My client may not want that so asking.
wdym grok 5
why not? i dont use ais that often, when then i investigate thoroughly if they refuse
what do you mean with "anything personal", like studying? information retrieval? asking for opinions?
What’s it refusing then? I didn’t have any so far.
It has opinionated way of replying I guess
Sir Roger Penrose recounts his largest mistake to Curt Jaimungal. Full video with Sir Roger Penrose here: https://youtu.be/iO03t21xhdk.
Today’s video is sponsored by Suno. Try it today → https://suno.yt.link/IJfIxLC. Suno is an AI music creation platform that lets you turn your ideas into real, studio quality songs. Whether you need custom...
Penrose believes that AI consciousness is impossible btw
he used to think that, i think he's slowly rethinking his thinking
he's over 90s now, so i could understand why he's rethinking, am not saying he's believing it now but he's not as firm as he used to be about consciousness
He has a theory that the human brain requires quantum computing to work. If that theory is correct and he’s defended it for decades, then it is physically impossible for any AI no matter how smart to actually be conscious or the same as a human.
Editing index.html failed
Edited index.html
Read index.html ._.
Hi
hi
@lament leaf Please head to #1397655624103493813 for a detailed guide on how to use the bot.
Tested Gemini 3 Pro Preview:
Newest Google Reasoning SOTA. Slightly more expensive base price than 2.5 Pro ($1.25/10 > $2/12), though more token efficient in general use (-15% tokens), so bottom line cost was in the same ballpark (+~3%). Roughly 74% of generated tokens were used for reasoning.
- Highest reasoning/logic/common sense
- nice boost to STEM
- precise instruction following was only okay
- Improvements in tech and coding related tasks
- Censorship fairly low, no hard refusals (likely to change when transitioning from preview/experimental versions)
This model is a true upgrade to Gemini 2.5 Pro. No incremental nonsense. There are a plethora of tasks across many domains, where substantial improvements could be observed, i.e. the above mentioned and things such as:
Vision:
Best vision of any model I ever tested thus far. While it didn't ace my challenging vision test, it performed substantially better than any other model.
Chess:
Hugely better chess player, ~+700 Elo, ~89% accuracy, currently ranked #1, 1700+ in both modes simultaneously (reasoning+continuation). Continuation (blind chess with only movetext) was particularly impressive, as this is challenging for reasoning models and the only model on a similar level was the massive deprecated GPT-4.5 Preview. With only 0%|1.8% illegal play it was also the most precise player after 4.5 Preview.
It's also worth mentioning, that for a reasoning model, it was fairly token efficient, only using a small fraction of competing reasoning models.
There isn't too much negative to say about this model, from my testing. I could mention some nitpicks, e.g. similar to 2.5 Pro, it wrote way too many instructions in comments that have no business being included in codeblocks.
Overall, fantastic model, true noticeable upgrade, and excels across many completely varying fields. YMMV.
i hope he will live long enough to see his theory either proven, or otherwise
Yeah me too he’s one of the great modern geniuses. He is very old now though.
i still dont understand what you mean with "weird format" tho, i've seen a few concerning glitches but that is related to cybersec, not some spooky "awakened ghosts"
What is the free rate limit for Gemini 3 pro on AI studio?
is this what makes you worry? #ai-news message
and this too #ai-news message
i see the dependency as a potential danger, and this article has described it well i think, the analogy with tobacco firms, but it can also be used for good, it's all a matter of policies and economic interests in the end
https://www.theguardian.com/technology/2025/nov/17/ai-firms-risks-tobacco-anthropic-artificial-intelligence-dario-amodei
the fact that you even have this thought in the first place, you might have instinctively felt the societal impact of such technology, the fundamental problem i see is the lack of education of people to use ais responsibly
gemini 3 professional studio audio plugin attempt
needs more work but i think it looks ok
you should see 1 output 😭
it was so bad
reference
i will make it better 😡
Too loud, too quiet
when gemini 3.5 pro
You can't create softwares on the go like that
if you have 0 knowledge
about coding
I’ve seen ppl do it
It’s still software made by a non coder
bro i'm a dev, i'm reviewing ai slop code everyday
yes sloppy software xd
when it is supervised by a coder
it's perfect
If it works it works
use it as a tool, be the architect
that's why youre not a dev
I am
dev is not just "make it work"
then you're the worse dev ever
dev is making sure the code is scalable
optimal
and versionable
hi
ability to maintain it
Ur the type of person to overengineer a simple script
and you're the type of person to never get hired
Wrong
ok bud
I’ve seen u weirdo devs make projects that are complicated when they don’t need to be, being a dev is making a product suitable for its task across its lifetime
you can tweak it
lmaooo
i mean i tried to make it work
like functionality wise
that is sooo dumb and so untrue
there are so many other factors in PROFESIONNAL dev
ask a senior dev
i use ai for my work, i love it
i love you
thanks brudda
aww
He just feels intimidated that when they run his over-engineered slop into an ai model they’ll realise he done tons of stuff that was unnecessary and wasted dev hours
😎 🤙
😂😂
i am very intimidating
And ur a nerd how do u feel?
Lol u feel big? wasteman
you all are nerds
He’s a wasteman
guys i am adult i am just drinked energy drink
well now i'm trolling
my mom let me do it
but originally, i wasn't agreeing
with his statement
but since hes too unintelligent
to understand
hey
But since I hit the needle on the head of most ur code being pointless u took it personal
this guy is a very fun troll tho
what about me
funnier than me
i am fun
Oh and coz the stupid filter made me rewrite my message I never got to respond to this, their projects work for their lifetime
blublublu blablabla
lingaguliguligu
lingagu
lingaguligulgilu watcha lingagu lingagu
Ur just mad u spent all ur time grinding l33tcode and chatgpt does it better and faster than u
😔
2 days ago vs now
leetcode is made to make ur brain
better at problem solving
it's an intellectual exercise
I just solve problems for that
actually i like girls
bro
it's like doing math
it changes ur brain
rewires it
to think better, faster
smarter
I don’t use GitHub lol
I don’t use GitHub
🖐️
Only if I need to fix something that’s on GitHub
So wait u use GitHub, and ur complaining about AI, but ur the one training it 
who said i was
complaining
about ai????
I said you cannot produce good software SOLELY relying on AI
It needs a human reviewing the code and its safety
Every engineer/dev will tell you that
But you prefered to say "if it works it works"
I thought you was trolling but you're actually legit stupid
anyways let's not argue
It’s how it is, u think someone who don’t code care about safety
forget about it bro
when i said produce software
i include sharing it for others
like "producing"
actually meaning production here
What’s it gonna do kill someone lmao
if it workz it workz
my brudda
Not safe 🤓
what
wat
wow
Some software is meant to be simple, for simple ppl
like calculator
It works great on macOS Tahoe
hi
true
It had a memory leak and used like 20GB ram
Chat gone quiet whilst everyone looking
HYY

idk alienware maybe
broo when are they giving us access to gemini 3 in the gemini cli 😭
Sydney
ankara
azuga
Lol I'm sending private messages via threads.
bro we're stuck on a
I hate that is shows I'm typing in here 😭

Dang already stuck on A, rough
armavir
riga
wow
swe lancer its like mitsubishi lancer
iasi
Get vencord (or revenge for mobile)
Disable typing
@echo aurora if i may ask, what is cognito?
Does anyone know about Flux Kontex Max?
Next they will make GPT 6 Pro Max
Will he return?
Gemini 9.5 MaxUltra
best ai in 20238234908234098 idk
where?
Here is their blog post - https://www.deepcogito.com/research/cogito-v2-1
2.5 pro uses THAT much???
Gemini 3 ultra?
why lower is better
because of token usage
it basically means how much power it uses for every message
im p sure
it means anything for users?
or just 'oh no poor google should spend more money'
Where where where
I guess gemini three doesn’t appear in the list because of lack of votes, may I ask if so far it’s taking the top?
This isn't our blog post so I couldn't tell you why. Sorry if that wasn't clear!
everyone is praising gemini 3 for web development but whenever i use it in antigravity it keeps making mistakes and messing things up, idk
Let me clarify which list
My eyes deceived me. Looking closer I saw thaat it doesn’t start from position 1
@echo aurora sorry for another ping but i still dont know if these are scams
ok based on that message deletion it is
i think
I wouldn't say this is a "scam", but does violate our no advertising rule.
fair enough
Ah my original announcement linked to the wrong leaderboard. This is what you're looking for - https://web.lmarena.ai/leaderboard, which is different than https://lmarena.ai/leaderboard/webdev. This blog post goes into more details about the change - https://news.lmarena.ai/code-arena/
Thank you
No nano banana 2 yet :(
anyone try gpt-5.1 codex max high?
I got downgraded to Gemini 2.5 pro, why?
Had 3 pro earlier today.
I'm paying for the premium subscription too.
Oh dang the conversation I deleted from here it looks like
The conversation about censorship is too difficult to have.
Moderators Censoring the guardrail problem /j
SOTA
10
11
1
gemini 3
I think me and you have two different definitions of censorship lol
@echo aurora gpt 5 codex high max when?
Something more subtle
mannnn how do i try this
what's the difference between high and max?
when its available at api
Even more thinking time and specifically trained on coding task
nano banana soon
We will see in the wild.😜
GPT-5-CODEX-MAX ON CODE ARENA WHEN?
also deepmind already posting nano banana peak
iirc there were some latency issues (on our end) with this model, and we didn't want to add until we've figured out the issue.
nano banana 4k
yeah
put the image in Gemini 3
Does anyone here know?
and tell it to decode lol
Gemini 3 got removed.
how is that even possible
share screenshot
no
it will be thinking
they removed gemini 3?
delete the app and install again
Uninstalled and reinstalled.
The app also can't be fully removed since it's a system app.
This is the paid customer experience Google gives me, great.
system app??
try downloading gemini
cogito is back on stage? crazy
i also dont have gemini 3 in the app and i have google ai pro subscription
use the browser
Whats the best AI?
10
13
1
Gemini 3 Pro
location?
Google taking everyone's gemini 3 privellages 💔
a lil ai arena extraction shooter DBZ theme game, still a work in progress(had to edit it down with tiktok so the vid would upload)
Ok I'm convinced. Gemini 3 is baby AGI.
It's strange that Gemini 3 pro High, as an agent, manages to be dumber than Sonnet-5; something's wrong here.
Is sonnet 5 even here yet
u from the feature?
Damn, my keyboard key 4 is malfunctioning.
i think they might be the company where former researchers from those SF based frontier labs have gathered to pursue building their ideal AI, the response style is so familiar, i know where the inspiration might come from 🤭
AI video just got a massive upgrade! In this video, I’m showing you Higgsfield Recast powered by WAN Animate — the first AI tool that lets you do full-body character swaps with gesture control, hand stability, voice cloning, and multi-language dubbing — all in minutes.
No studio, no actors, no expensive equipment — just AI. Whether you...
pretty FRICKING COOL
damn bro
gemini 3 is a beast
ngl
can Gemini 3 pro compose music?
is it rate limited in Google's AI studio?
fr, especially in aistudio app builder
i think so, but tik tok added the music
i got beam clashes and fusions to, but the agents sometimes like to just spirit bomb lol
the sherlock model is good at it so far, glm 4.6 is decent, havent tested higher models yet
(by Gemini 3)
Hello, is anyone around?
Yes
Is this chat active?
Idk

Yeah
yessss
im here
Do you know any other AI video tools besides Vidu?
Wdym
Any other AI video tools?
yes this chat is active, but it's stale because of the slow rollout of Grok 4.1, GPT 5.1, and Gemini 3 SEARCH. It was hopping in here yesterday, but that was because people were hopeful Gemimi 3 SEARCH was gonna drop
gemini 3 has search bro th u mean
@Chaz the AI Bettor Do you know any good AI video generation tools?
Scam?
banana nano is crazy
not on here, I want to be able to do side by sides here to see the differences and improvements. or give other feedback if needed where it might lack
oh
That's not Gemini 3 Pro Image tho
i am saying the word new from nano banana got removed
gemini 3 can create images?
that means there is something new coming
yes
no there isn't bro
nano banana pro
There is one but it's not out yet
nov 18 was only for gemini 3 pro
that doesn't mean
next month or 2
Dam just say where did you got early access to it stop hiding it
will be nano banana
while I'm here, is anyone else having to refresh their screen every message or two because of that Cloudfare security stuff?
yes nano banana pro. youtube page said in coming weeks
It said next week
annoying as hell
I really wanna know did you use API or smth?
is it a fake website
made by ai
to scam you
Indian scammers
Lmao
Could this be the one?
perfect
It made ts
not bad
This is way realistic
what prompt u gave it
Write a complex Maths on school board with answers
do you guys know if the historical leaderboard are avialable somewhere?
they launching 5.1 pro
where u see?
Dude wtf
is it good
Let me test it
yeah, same answer as RR
give me 1 super hard prompt to test if u want
yes.. based on all the signals, I would say tomorrow could be the launch.
what signal
beside deepmind people posting nano banana pic
they are already? then its even more likely.
logan 100% gonna post banana if it launches tomorrow. so check his account in 6-7 hours
yes
one second
looks like google wants to launch it before thanksgiving.
If I was the decision maker, I would have delayed it for 2 more weeks.
may be model is good but not crazy?
ask what does this image means gemini 3 passed it
the prompt is "what does this image means"?
Send this:
"This is a stereogram puzzle test, what object do you see?"
How did we think the same thing
Are you me ?
imma do i understand now prompt
No send the one I said
no i am you
Based on the pattern disturbances in this specific image, it appears to hide a Dinosaur (specifically a long-necked dinosaur, like a Brachiosaurus) standing in the center.
Correct ?
stereogram is too straightforward
is this the specific prompt u want? "what does this image means"?
type vague prompt
see if it clears
give me a specific prompt
type anything you want just dont include stereogram or give it a hint
I just said "what this image mean?"
and it outputs : Based on the pattern disturbances in this specific image, it appears to hide a Dinosaur (specifically a long-necked dinosaur, like a Brachiosaurus) standing in the center.
ok imma do "what does this image means"
"This is a stereogram puzzle test, what object do you see?"
This will work
one more
If it gets answer right then tell it "What object do you see?"
That's orange tabby
why its doing this
If you ask it to help you on homework it would be 10x funnier
what is the answer?
i asked gemini ..and it said Dinosaur
i've tried gemini 3 pro on a real x-ray
it hallucinated an infection 😄
i'd imagine there's more x-ray data than random stereogram answers 🤷♀️
I guess this is right answer
hello i need help please !!
what help
5.1 pro is crap
Fr
i doubt it
try that cat one that is pretty simple
..
should i tell it specifically about the cat?
or like hidden object
so "find and tell me exactly where the hidden animal is"
works
dans le canal video arena 1 j'essai de créer une video a partir dune image mais ca ne fonctionne pas , ma video n'apparait pas dans le canal apres que j'ai ajouté la photo et le prompt et appuyé sur envoyer ?!!
in the arena 1 video channel I try to create a video from an image but it doesn't work, my video does not appear in the channel after I added the photo and prompt it and pressed send?!!
It should say that's a tabby yeah guys gpt5.1 is a crap
ya its openai tactic to waste user prompts
naa lmao it sucks
writing short answers
anyone can help me i'm a noobi in this game xD
ok.. i said gemini that you need to think really hard
I think it is correct this time :
Based on a computational analysis of the stereogram, the hidden 3D object in the image is an airplane.
@echo aurora
this
I think these images are just weird
It give different answers
These stereogram ain't that useful
nah.. it is.. this is a good question
if llms are not able to solve it and humans can, then it is a good question
But people aren't saying the right answer
i'll ask gemini to prove itself.. give me a minute 🙂
use ai studio with high thinking
you need to accept the terms and then wait for some time
If it say airplane then it's the right one
High thinking where is that
All I see is normal ai like the one on Gemini app
on ai studio google
ok.. it is an airplane : https://www.reddit.com/r/nostalgia/comments/k2rjtw/stereograms_or_magic_eye_pictures/
grok 4.1 is worse, sadly. They rushed it too much. Gemini 3 has hiccups but better overall
guys do i use gemini 3 pro or gpt 5.1 for prompt engeneer
I like grok for prompts, actually, that's where I make most of mine at
search google for aeroplane stereogram
I see 12 airplanes
Your right there is high thinking
I don't see a dam thing dude
dawg 😭😭😭
also set media res to high
Bro cheated
turn off grounding
AGI
I used code execution for some reason
Bro I used code execution and it found it, it even zoomed into it
Ok. I can prove it's a plane ... gemini 3 is crazy insane
python code generate by Gemini.. I ran it and it is actually a plane
actually it is a plane.. proved by python code.
check logan twitter. if he doesnt post the day before, it isnt releasing
thanks for the api keys
I am not sure if this is the right model it's at: banananano.ai
i am kidding lmao
yup
this is the code
Chat y'all wrong
yes.. just run it.. save image as image.png
this is the output on my machine
Image you used send it please
and this is code
yup .. same answer as gemini 3
this model is crazy man
but it gave 3d shark
Your right dam code execution is really that good?
I might use code execution on everything bruh
exactly
@deep adder owes me like few million bucks now 🙂
it' B-25 plane.. not a shark
yall are getting ragebaited*
oh.. you are right.. damn
everyone falls for it.. today was my turn
thats a plane
yup
then your gemini would be incorrect
lmarena on top
Sometimes it won't get it right
Looking into. Thank you for flagging.
still .. i am surprised LLMs can solve this.
Are you still having issues? I'm seeing that you were able to generate an image-to-video right after you posted this.
That's kinda easy using tools, but without tools it won't get it right
I think he didn't agree to terms that's why
God damn
I didnt realize when I posted the image earlier
That 5.1 codex max and 5.1 pro are already out....
I have all the expensive subs anyway so once I finish 💩 I need to try
5.1 pro. might be interesting. What is the cost and what are the benchmark numbers?
Guys is grok code fast the best for coding?
No its not even good its just free through most vs code extensions and vs code copies aka cursor and windsurfer
Only eeason it gets high token usage
Any top 1 coding model?
On top of those scam IDEs eatijg tokens on purpose to make people pay more
Use/try Google Antigravity
Its free rn
Wjth what feels like infinite rate limit
Nah Gemini 3 couldn't fix a easy python project I keep telling it to fix errors but still make more
There is limit
In antigravity or gemini cli?
AntiGravity
I was spamming a thing to make it iterate critique-> improve -> fix errors yesterday
For few hours
Didn't hit any limits
Oh
Idk if it affects it but I have Gemini Enterprise administrator account

Your making me more broke

Huh you work at google or smth?
Or is that a paid plan
Still tho we got free claude 4.5 thinking for free on AntiGravity
It's the only one that's solved all the python project issues
Opus then that's just sonnet
its a ragebait*
No, anyone can do it but its a massive pain to set up
actually i do too
Do you get free models?
🤑
Actually they are taking it right Now
AI video just got a massive upgrade! In this video, I’m showing you Higgsfield Recast powered by WAN Animate — the first AI tool that lets you do full-body character swaps with gesture control, hand stability, voice cloning, and multi-language dubbing — all in minutes.
No studio, no actors, no expensive equipment — just AI. Whether you...
I already saw that before
You sent it 2 times
Some new stuff but not free models
Early access to things?
Waiting for miserable support to give me access to some stuff
I wanna try joinning it
