#general
1 messages · Page 177 of 1
*vote
i made em in 2020-2022
buttons went unusable
there wasnt a limit last time i used it
it does start to say youve hit the rate limit but you can just try again after a few secs
@jovial sapphire yo bro
also when you reload page while revealing models you cant see their names
what userscript u using
the limits are still very high in AI studio
so u dont hit it immidiately
bet ill try, im starting to think riftrunner is flash, its always faster than the other models in the code arena
nah like last time i used it i could spam it like 500 times
ragebait
🤣
idk bro, look at this run i did and tell me which is better:
https://019a8e63-8b10-71c6-8901-f7072130c931.arena.site
https://019a8e63-8b10-7cdf-b4a0-7fa23f3890b3.arena.site
Built with LMArena - Content is user-generated and unverified
Built with LMArena - Content is user-generated and unverified
hes ragebaiting
dw abt it
with hella ADHD
prob gonna say "which one did gpt 5.1 make"
xd
@deep adder should i screenshot your ADHD comment
"i have mega adhd"
damn i been telling people for a while now, but Grok imagine is the king right now for video gen, like wtf: https://x.com/0xInk_/status/1990102312388550693
nah sora much better
LOOL
enough
LMAO
he said im lying
u speaking the truth?
check the chat logs
u lying to mod?
bruh
my uncle has down syndrome
who has adhd?
hey socky, did u try reading text on ur screen
we've seen grok think for longer and give outputs worse than non thinking models 🤣
nahh bro, i been using sora 2 pro for a while(with 4 accounts) and i even made an app with it for prompts, but sora 2 has been lobotomized heavily, the key to this video models is truly short prompts but we cant do that with sora 2 unless we do cameos, but with grok it seems the quality is insane, it seems much harder to get quality out of sora without using cameos now or some boring prompt that nobody cares about
u using sora 2 pro with 4 accounts what u mean
u got $800?
well i had pro on 2 accounts, then plus on 1 and then free on 1
u using sora 2 regular, not pro
you bought?
u cant prompt
i can prompt
trust me
i mad an app for sora 2 prompts
and its decent tbh
but i got bored of it
ur app is bad at writing prompt
what
and its too much work to get around the copyright stuff
and they banning accounts now
u only care about copyright stuff thats why
😆
why wouldnt i?
cuz u cant create anything beside copyright
whocares about made up crap? i want goku, luffy etc..
I'm gonna give 50/50 if it beats it at all or not. Either way there's gonna be only small differences between them. With UI, caps and features being the deciding factor which one you should use, as always.
you are pretty good at gifs lmaooo
Anyone who thinks gemini3 gonna destroy gpt5.0/1 is clearly delusional though lol
what you mean?
isnt that obvious
its not gonna destroy
😉
its gonna be incremental upgrade.
define destroy?
destroy means 10-15% above in all bench
am i going to never use gpt5.1 ever again when g3 comes out, yes so to me thats destroy
destroy as in the difference is at least as much as it was comparing o1 to o3. Google has no chance for these gains over gpt5, objectively
we will see
but these gains are not what we measure by anymore, its about vibe, feel, speed, and price
vibes lmao
lmaoo
it has
Hello everyone!
Could someone tell me when the release of G-3 is scheduled? And what are your impressions of the KIMI K2?
It is what we measure by performance. Speed and price or output style are entirely different things
if majority of people switch from other models and use g3 until the next new model then i would consider that destroying the other models, its not all about benchmarks if people do not use your model
I doubt openai will catch up to google if it requested government loans [aka. money is runnin out]
Yeah no chance lol. Let's wait and see though. I'm calling incremental improvements over 2.5Pro + they gonna sweeten the deal with say hybrid reasoning or routing features.
even opus seems to be better than gpt-5 now, at least in coding
ok imma go sleep now gn
and i will bet $100 that people will be using the flash model as their daily driver once this week is up, like you will notice the switch in decemeber
It isn't. The one coding benchmark Opus does best at even that it's now equal with gpt5 (sub 1% margin of error)
I emotionally like Grok Imagine and socially like Meta Create and physically like OpenAI Sora and spiritually like Google Veo.
Guys i have tested riftrunner on lmarenaa!!!! ( i asked it about why linux is very complex and have too many lines that i couldnt stands )
Or wait, was it Sonnet 4.5 with Opus scoring lower than that... 
Well either way
Anthropic is no longer ahead on SWE
And this kinda was their last thing where no one else could beat it
yeah gg for Anthropic
u think elon still runs his twitter? like he be jsut randomly tweeting stuff?
or he got grok behind the scenes
he 100% runs it
looks like how i type fast english.. i dont think he is him..
its called typing fast with phone
I think he does. If it happens that he's doing it with less than 6h of sleep that's probably ketamine lol
its crazy how much this backfired
nice joke.. but i dont think he eats that weired thing..
u sharing?
crazy times we living in when we got CEOs of billion dollars companies and richest man tweeting crap and slop or news about the company or beefing with other CEOs like wtff, post 2020 has been a trip
its cuz he owns it.
yes..
oh yeah, that sometimes escapes me, but Sama does it too and logan from google
like x has become an interesting platform
wasnt this Elon's plan?
logan is manager, not ceo. sama is friends with elon
where do i share?
here
can i use ai?
yes
hii
tthen u must wait a couple of moments.. , it will be on html..
ik he is a manger, im just giving examples of people in billion dollar companies tweetin news about that company that can cost the company money or gain them money like its wild, and even Sundar Pichai tweeted about g3 on polymarket so my point still stands
Being super optimist:
Open ai will release Gpt 5.5, it will have same gape to Gpt 5.1 samehow Gpt 4.5 were to Gpt 4.0
It will not be super expensive since they will make it INT2.5 and some other optimization method
Soo Gpt 5.5 pro could be comparable to gemini 3.0 pro
any publicity is good publicity
wild I just got an error in code arena but it still was asking which model was better, i could not see the preview or code, only the thinking right before the output and the clear difference i say was planning in one so i picked that as the better on and it turned out to be riftrunner, i also noticed this in my tests when coding for work and stuff, the plannign part is what has made a difference to me in my outputs, and it seems google really pushed g3 to be better at planning before implementation
mkaes sense tho
also google is friends with elon. even demis hassabis respond to elon
Being antivax probably meant he got the most damage possible from the virus. 🗿
They respond probably more out of being polite and smart. There would be no use to them having him as the enemy
why does sama respond differently
Cause Elon went aggressive first and went hard. At that point you just respond with the same to show you not gonna be pushed around lol
He still has a grudge over how he screwed himself by leaving OpenAI
Yeah... he is responding to uber confrontational Elon's comment. Elon is not making such comments about any other AI lab to his CEO
Like this is public space, you don't just write "you stole X". Let's not pretend this is normal LOL
What does this have to do with OpenAI? This kinda makes it look even worse to be fair
yep
Like Elon is not defending Tesla, cause he understands the criticism was warranted.
So with this context his response reads even more teenager alike
It's almost like writing "no u" with a straight face
Yeah same 5.1 is looking better
5.1 looking good
There is free gpt plus in the webs6
huge improvement
Website all go check
what website
look, i dont see that current asi problem is about it can escape the machine...
thats why i stopped..
Chatgpt
what would put 5.1? like is it current SOTA? would you use it instead of 4.5?
Its gone be as good as none
is it better than 5.0-minimal.... That is the question 🧐
I think so
I think you have to use both cuz idk
I somehow hate that they also have this verbosity parameter
makes it all very confusing
Bro I want to know the chatgpt don't show how he think
Like all the ai shows how they are thinking
What do you think guys are the best python ai ?
4.5 has been the best for me for coding stuff, but i havent tried 5.1 codex that much
Codex hasn't worked for me
It's nearly gibberish. Efficient and concise but not user friendly to read at all. I believe (based on leaks) their current raw CoT for gpt5.x looks similar to this: https://github.com/aw31/openai-imo-2025-proofs/blob/main/problem_1.txt
Is that who made glm4.none?
Wdym. No, that's official repo. They had to publish it for IMO competition
O
Should I read this?
Btw there is a free plus plan on the chatgpt website
@quartz light i figured it out. Riftrunner was removed from text arena Probably and is only in codearena now unlesss someone can disprove it
😕 but I can't get it cuz I don't PayPal :/
Hi
Do you know ai good at python
Riftrunner has been such a distraction for me. Lol I have no idea how I'm going to get any work done with the real release this week
fr
you think RR is the best roleplaying LLM right now?
(including competent GM'ing)
yes
Don’t worry, we’ve got you!
It is indeed a lot of work and takes some time to get research prototype in shape for public release, but a much better version of IMO model will be released in the coming months.
Don’t worry, we’ve got you!
︀︀
︀︀It is indeed a lot of work and takes some time to get research prototype in shape for public release, but a much better version of IMO model will be released in the coming months.
Quoting Gary Marcus (@GaryMarcus)
︀
People saying stuff like this “OpenAI's unreleased model that won gold on the IMO in July is clearly better than even the best publicly available model today (GPT-5 Pro)” are speculating and probably wrong.
︀︀
︀︀my hunch is that the IMO model is limited either in its generality or reliability or both.
︀︀
︀︀note that the linguistic output of the IMO model was weird and that no further details have been released.
︀︀
︀︀It was (and for now still is) a demo, not a product.
who that
whats happening why uam
😔
any idea what been going on?
no
@echo aurora can u disable uam pls, cant access lmarena site
woah
ye 🙁
same
Oai employee
veo3 fast
I wanted to test the new models but can't access lmarena 🙁
at least video work
YOOO GUYS
SOMETHING IS HAPPENING
In this video, I'll be sharing my early hands-on results with Google's upcoming Nano Banana Pro / Gemini 3 Pro Image Gen model, showing real-world examples of its realism, text handling, UI screenshots, and more.
--
Key Takeaways:
🚀 Early access look at Nano Banana Pro, likely launching soon as Gemini 3 Pro Image Gen.
🖼️ The model pr...
@quartz light yeah bro gemini 3 pro is coming this week
they giving early access
to nanobanana 2
i am still waiting for gemini 3 in gemini api 😭
damn, well u aint gonna wait long
LMAO
that lofi music video was fr on everyones youtube if u opened it for first time
what the frick, send link to try pls
its was available for 5 hour a week ago. its blocked now
ok wow, thats so good
whats happening
wait these are fake images?
yes nano banana pro
ai gens*
DANG
using gifs is just wild, imma put this into sora 2 and see waht it does lol
what the frick
pov: someone saying: copyright shall be valued 💀
and then comes google ripping their own youtube design in their image model
still just text completion sadly 🙁
but crazy good one
gemini 2.5 was already pretty good at answering question about some code I give it to him
6 fingers
ye, gemini 2.5 is still my favourite by now
i hope gemini 3 wont cost way more
if its about similar price
ill just switch to it
like damn
u paying for gemini 2.5?
ye
how much
??
free on ai studio
btw @cloud zinc this is the app for sora prompts i use: https://reemifai.org/ its still up surprisingly so you can test it if you want
really? what you use llms for?
enough to assist with whole large projects
dude
use ai studio
its unlimited
paying for 2.5 is criminal tbh
how much u paid for gemini 2.5 pro
i pay for claude opus too 💀
no idea tbh 🙁
what
why does it need api key
☹️
its free liek everywhere, studio, the build section in studio, canva, gemini app free trial, gemin code assist, gemini cli, yeah they all use the same usage so you will run out, but jsut make more accounts, i never once paid for any gemini model
i aint giving api key to scam site
its free until your usage runs out
just make a new one and do not fund it, you new to this or something?
i aint making new one. time waste
lol
dude reacts to his own message, weird behavior lmaoo, but to each their own
dw about him, some mental stuff going on there
says the guy who emoji his own comment
does LM Arena have any ai text to speech?
use ur mouth
does anyone know some really good quality voiceovers thats free and has a lot of characters
i havent seen u in a while
how u been doin
bruh, u gave me sora 2 code
are u on a diff account?
u also gave my a perplexity link
i use a free tts that voice cloner
which?
forton?
i aint sharing with u
fr
🤣
are we deadass 💔
bro sock you just be ragebaiting at this point
also elevenlabs are the best at text-to-speech
they give u some free credits
yeah but only 10k characters a month
elevenlabs is limited
i need 10k characters twice a day
mine is 1000 character only, but unlimited.
ik that one its pretty ass
its the best atm, in terms of voice quality
i need at least a 5k character limit thing
other models suck
which one is that
lovevoice
wait what are u using that much for
thats not it bro
u doing yt shorts
alright well i mean 1k characters each time is too little for me
or tiktok videos
enless if the voice quaility is great
some work
hmm
how much does elevenlabs cost
its probably worth it
tbh
5 bucks a month for 30k characters a month
only 30k ha
elevenlabs offer the best quality but its very limited
they have a monopoly going on
yeah fr
exactly
openai voice is much better
20k characters a day times 30 days 600k characters a month
open ai has text to speech?
nah
you're deaf too?
sora speech
bro wth is this
sora is good but it doesnt have a dedicated text to speech mode
sora speech is much better than elevenlabs
these voices are straight ass
yeah
but it doesnt have a dedicated text to speech mode 💔
that site is a scam
ik
how about. you help a bother out
i finnally want gemini 3 or nano banana 2 so i can add to airfroce 😭
did u help me?
please?
airforce?
airforce?
what do you need help in
he needs some braincells, maybe ears too
help in getting this troll out
??
yep
is bro schizophernic
are we in hogwarts or something
nothing
maybe he is
and you're in the hood
brotherrrr i just need some text to speech holy
nope
ur in the US?
and u in europe?
europe is much better
🤣
bruh u sound like a 45 years old facebook uncle
why u on discord
exactly
log off and get some air
@dim basin tbh there aren't alot of TTS AI to choose from
idk how u using that much tokens in elevenlabs
but it has the best quality overall
when lmarena back available 🙁
infinite loading
screenshot
again and again, it reloads and tells same again
is it working on your end
wait open a new tab
guess its just me then
i just opened that tab but i can open another one and test
what browser are you using?
edge
try chrome
wait its working for me
im on brave and its working
brave 😂
wow
ik
but send code
ye
which is better?:
https://019a8efe-ee1c-78e3-ae66-a4aa523543cc.arena.site
https://019a8efe-ee1c-7b6b-8757-ba62e5dc335a.arena.site
2nd
wtf
my opinion is #2
does anyone use those gemna models or whatever they are called by google?
gemma?
not for anything but rp
yeah
what rp
ahh ao you think they are the ones that is going to be used in games?
roleplay
no
even tho gemma 2 was the best one in my opinion for the 1b size
even able to speak other languages fluently
yeah
ive seen people hyped more for gemma 4 than gemini 3
cz its local n stuff
its for phones and small gadgets and tools mainly i think
it has better image analysis than models many times its size
ahh that makes sense, yeah google has their foot every where
also is riftrunner still on code arena?
have not been gettign it for the past 4 gens
before i got is every gen or 75% of time
google making their own arena
im shocked more labs havent done this
ye
tbh
idk
lemme check
ye its still there
okay thnx
is this good texture?
is it comfirmed what riftrunner is
no, and all those people saying riftrunner is For Sure 100% gemini's release candidate are blowing smoke
but
it's still very likely to be gemini 3. wouldn't bet against it.
you think its not a gemini model?
It can be like, mini max 2 instruct or gpt 5.1 pro
claude 5 sonnet
More realistically opus 4.5
maybe that too idk
I do think it's a gemini model. But I've seen people claim it's "officially confirmed" to be gemini and that's just not true.
I would be surprised if that happens. More likely they gonna skip 4.5 with Opus entirely IMO
so ye i cant wait for ECPT
its a myth bro
just like nightwhisper
we aint never getting ECPT
gemini 2.5 was good first days too
before getting downgraded
probably same again 🙁
i would pay 3x
for having best code model one
no idea 💀
thats scary
pls rate tho, not done yet
sorry, private
lol is not that serious bro, like you could easily say a generic response like chef or developer, like me im a developer but that is so vague but its all good bro lol
24k a year is still wild for ai
that might be an effiency or optimization problem you are having
what i do is too specific already and theres no overall of it
so sadly not
dont really mind spending that much

even giving free ai to others cause why not
but ye, cant wait for gemini 3
you?
this month yet
totally understandable 👍
i also use gemini (even more) but its not in that graph
i use it for work and cant use opus
🔥
yeah sadly 🙁
ahhhhh i cant wait as well
yeah, its cracked at 1 shots
are u using ai agents for the npc?
i been trying forever to build a game like that but using cli agents but having trouble
🥺 ok
well can give u ai if u want but that would be self-advertising 💀
so wont do in here
dm
and its okay
i made an app and posted it here just dont go overboard and nothing paid, just tools for ppl
Your pfp is "my mom bring strangers into her room"?
My farvorite is the mom extra 🤣
🥺 my favourite is the daddy who left 😭
Minecraft from gemini 3.0?
half opus half me
Hmmm
And not the opus 4.5 😈
whichever is better
What is ecpt?
the currently best checkpoint for coding
Hmmm, I'm not a code, but I still anxious for gemini 3.0
honestly its a nice one too, ye, probably my 3rd fav
🔥 🙂
the mic one is just sad 🙁
riftrunner is nuts: https://019a8f93-8729-7962-9f5c-d10c4c0048dd.arena.site
an extraction shooter
dang, that game is hard but i love that
guys how come the new GPT 5.1 codex models not available for direct message?
probably only testing yet
do you mean riftrunner or is riverrunner something new?
codex is only on code arena
yeah I'm saying they should make it available for direct
fr, i was struggling with it myself lol
yes i meant riftrunner
ok cool thought so but wouldn't be surprised
Hi
Honestly
Copyright is going to slow all ai progress
Everything is always gonna have to be beat around the bush
damnit scrolled down too far on my chat history @echo aurora please fix? 🥺
The same is gonna apply for coding and code
Watch some ass association or something is gonna come in and sue them for copywriting and training their model on
Code and web design
Gemini 3 Pro
We’re gonna end up paying for using the AI to generate content that you don’t own that you can’t use without paying royalties
in the code it still showed index.css and index.js
why are we limiting the AI to only output one single file
"An SVG animation, with sounds etc, accurate of a DBZ fight between Goku and Frieza on Namek. Really make a lot of steps, like a SVG movie. Very accurate. Cinematic almost! Send the HTML file in one single code."
Gemini 3 Pro
@fiery gull
bruhhh that is amazing
this is claude: https://019a8fe6-6dcb-7b03-ade9-152ad70a867b.arena.site
actually not bad tbh
this is riverrunner:
https://019a8fe6-ae87-787d-addd-d6ba374367d4.arena.site
Same prompt?
yeah
I'll retry
Yup
In terms sounds and animation this one is very cool
I think G3better overall
yeah i cant lie the one you must is the best as a movie with the beam clash too
but the 4.5 one was cool with the transfomation
Yes!
New test: "An SVG animation, with sounds etc, documentary on humanity and space exploration, like a documentary, with google voice, but like idk make it realistic with passages of radio effect, history, first launch, modern exploration, futuristic. Like a Nat GEO documentary, epic music, calm, relaxing, something philosophic. Music, etc like really a crazy thiing that is not just 10s. Really make a lot of steps, like a SVG movie. Very accurate. Cinematic almost, it needs to really baffle the specatator. Send the HTML file in one single code Minimum 1500 lines, we want something big."
you just inspired me to do this:
An SVG animation, with sounds etc, accurate of the final scene in the avengers endgame movie with the Ironman and Thanos and the final snap. Really make a lot of steps, like a SVG movie. Very accurate. Cinematic almost! Send the HTML file in one single code
hmm thats a very complex one, i heard that g3 likes simple prompts so this will be a good test
Okay will try after documentary one
Trying (;
wow you are good at getting it on ai studio
Multiple accs
damn thats a tough script cause google sometimes tracks down on that
but based on that output did it not give you the code?
It gave it
But bad result
someties it's not gemini 3
Yeah just checked it was bad checkpoints
damn i get that sometimes as well:
https://019a8ff7-977a-7ac2-9dec-45e24f1a6910.arena.site
Built with LMArena - Content is user-generated and unverified
not sure what this is yet, the other model is still loading but it gave an okay result
i dont think its river
Try: An SVG animation, with sounds etc, documentary on humanity and space exploration, like a documentary, with google voice, but like idk make it realistic with passages of radio effect, history, first launch, modern exploration, futuristic. Like a Nat GEO documentary, epic music, calm, relaxing, something philosophic. Music, etc like really a crazy thiing that is not just 10s. Really make a lot of steps, like a SVG movie. Very accurate. Cinematic almost, it needs to really baffle the specatator. Send the HTML file in one single code Minimum 1500 lines, we want something big.'
bro its crazy
nahh imma play with svgs more man, this is so much fun
wtfff the story telling is wild
yeah ill try now
Built with LMArena - Content is user-generated and unverified
i think this might be g3
yeah
still g3
but not sure
waiting for the other model to finihs
then i will vote and find out
lets see if it actually is g3
Ye
hi
How do we rigorously measure AI's intelligence? We don't really know. What we know is that measuring intelligence is tricky, and if we're not careful, our tests might not measure what we intend. We explore this topic by starting with the story of Clever Hans, a horse who seemingly could do arithmetic. Later, we explain the potential limitations ...
not as good
looks cartoony tho
its sonnet lol
still not as good but this is def rift:
https://019a9034-d80f-782c-92e2-224e931f1b82.arena.site
Built with LMArena - Content is user-generated and unverified
We’re back in the PowerPoint days lol
Anthropic CEO Dario Amodei warns of the potential dangers of fast moving and unregulated artificial intelligence, while also racing against competitors to develop advanced AI.
"60 Minutes" is the most successful television broadcast in history. Offering hard-hitting investigative reports, interviews, feature segments and profiles of people in t...
U guys believe this?
Is the mandarin model only available in the battle tab?
do you have the html for this?
There so full of it lol
You know what happens when you give them responsible, ethical regulation
svg??
nano banana 2
how u using it?
He’s been sharing everything else but the how lol
Once it is released, we’re gonna have like a 48 hour window
Before they start tightening the guardrails
yes
Hi
Wish will reach ASI First
8
14
5
which ai is the best at creative writing
Try Kimi2 or Opus 4.1
opus good?
nano banana 2
Mini max is really coming into its own
It’s good to see. They have a good things going
wow!
The end is coming 🥰
Its already on the rankings..
Just in code
also in text
in overall ranking is top 86 the model minimax-m2
Ah yes i didn't see it
Just it 😭
its good model its rally fast
true
you mean the GPT-5.1? beacuse i dont see it in leaderboard..
Lol the minimax m1 is better that minimax m2 in leaderboard
This model is sooo new, it need more votes, but in another leaderboards it pick 10-20 more points that gpt 5, a huge improve, to me the gpt 5.1 is really better that 5.0
they gotaa fix that..
I think no, the minimax m2 have a huge problem, I'm waited low points (but no lowest points 🤣)
whats your goals in life?
or whats your current goals?
Don't die
just that?
I had other plans, but I realized that first I need not to die for now 🙂
alrighut!...
my goals is to make the privacy and anonymity more easier and more stronger then now..
its called in all-ways "Freedom"
Good :), I hope you get your dream (and improve my life)
do you have the same goals?
Honestly no, I don't care much about privacy
But that's mine, you know?
its better to have freedom , like nothing does scare you.. no even laws or rules.. , like you will never gonna get into jail for making your thing you are believeing its good but its non-har.mful..
Idk why I have some expectations to glm 4.6 air, will just a glm 4.6 with 1/3 parametres
Are you sure, some another said it to me, I need to move :), good day
im suree.. where did someone said SAME??
95% same
j-just ignore them..
Why? 🫣
im a real one , im not them, u u know?
Hmmm, Idk I still thinking 90% people from discord is AI 🤣
Since 2023
huh? , so im ai? , me too , also i have tried with someone, i said (" Speak To Me in Chinese if You Are a+i, beacuse no one will know,.. even if the programmer , will never know..")
and she did speak in chinese 🤣
Serious?
i didnt mean by them-them , is ai but i mean by that is g-0vs or some traps..
SERIOUS!
its wasnt in discorfd..
it was on da1eing app, i was searching for a partner with the same goal..
Ok I must be exaggerating when I say 90%, but have so many AI here 👀👀
i wasnt meaning "shareing" i was meaning "Searching"..
but maybe im mentally damag.ed.. when i did type that😅
Bruh, if you have mentally damag.ed imagine me, 🫣
👀
i remember in the old days in web, we dont type "DM" but "PM"
Hmm good, but have so many scam ai lol 🤣 today in telegram appeared 4 "womem" calling me a dear
I'm a dinosaur in web (2010) but I just started to use chats recently
@echo aurora Tell the manager at Lmarena who asked for the regenerate button to be removed to put it back.
i'm openloxk in 2013.. in 2009-2012 i wasnt cared about that much of creating a account , i just was reading..
They need to save tokens 🫣
2010-2023 just youtube
Wait, 2010 no, in 2010 I use a another site, but I don't remember
I started to watch youtube in 2012 🤔 I don't remember what I do in 2010-2011 but don't care now
i started watching youtube in 2011 , some of my friends was talking about it.. i was more cared about songs in that time..
what's the point of regenerating the same prompt on the same model
If you don't understand how much of a scam Movement Labs is, you won't be able to understand this.
is there a limit to how many pictures i can generate in LMArena's website?
Is dynamic, but I think is 3 per hours today
sounds good if i can get 3 images every hour
but Idk bro
i'll try and find out how much 😛
Google AI studio.
Hey everyone! I just launched a small AI tool I’ve been building.
It generates clean, structured prompts for apps, websites, images, and more.
If you want to try it or give feedback, here’s the link:
https://promptdumper.web.app/
Anyone else getting rate limited for no reason? I'm getting Error 1015 when I just switch between a couple a conversations
Tried it
Vs orginal
Image prompts are tricky
riftrunner lives to see another day
gemini 3 tomorrow
RR is launching this week (likely tomorrow) without any doubt
hmm
i would have preferred one of the earlier checkpoint but it's happening
huh? i still have the button
riftrunner is currently the only checkpoint in LMArena history of g3 checkpoints to not be deleted after 4 days
yup.. because it's launching tomorrow
how do we know riftrunner could be the one that launches
people freaking hated it the most
because it's kinda lazy (not sure if this is the correct terminology).
It does better than lithiumflow if you prompt it to think harder.
Like, I just see to doubt x28 coming out...
The model was just too perfect..
😔
good thing is that atleast they have the technology. So a powerful update to RR might release in near future (3 months or so after initial Gemini 3 release)
I wonder how will they know what checkpoint of gemini 3 released once its out..
does anyone have a realistic free ai voiceover that has a lot of free characters
Does anyone actually know the best ai coding model, there’s new ones every day and I just can’t keep up, and the leaderboards are never updated
if your asking available
riftrunner on lmarena rn [stealthmodel]
[codearena for easier chance to get it]
Battle mode ?
think the buttons only on textarena?
What ?
dunno let me test
nope i dont have it
I’m not to good with this stuff but I don’t see any models named that, is it an upcoming one? I wanted one that’s currently available on lmarena
oh yeah then best i can suggest is gpt 5.1 high
but it also runs into errors
Why would it be in the arena asa checkpoint if it’s lunching tomorrow?
If we do get riftrunner as a launch, and people find it out somehow. They will outrage at google for picking it
isnt this a win win for both
[none as of now] expect that riftrunners still live and kickin
Kinda ya 😂
Gemini (♊︎; /ˈdʒɛmɪnaɪ/ JEM-in-eye[2] Greek: Δίδυμοι, romanized: Dídymoi, Latin for "twins") is the third astrological sign in the zodiac. Under the tropical zodiac, the sun transits this zodiac sign between about May 22 to June 21.[3] Gemini is represented by the twins, Castor and Pollux,[4] known as the Dioscuri in Greek mythology. It is known as a positive, mutable sign.
https://019a91f8-4bd7-7727-9244-0d1fc8a4ec06.arena.site riftrunner made this nice editor for people who dont know coding
Hey ! Does anybody know is it possible to make an image from full comic style into realistic cinematic shot with real characters/actors and the scene, surroundings would also stay the same ? But in realistic colors not comic/anime.
I know it works the other way taking real persons and changing them and the environment to anime style, but does it work the other way around aswell ? Tried with nanobanana but it still creates it in comic style maybe just adds some more grain to the shot.
If “with real characters/actors” refer to “asking for specific actors/actresses in the prompt”, then model is very likely to reject the prompt and instead produce incoherent results. Reasons including uploaded image+real world celebrities trigger copy right alert; the level of censorship varies from model to models. If you don’t specify celebrities, rendering comics to realistic texture is fairly easy.
I can concur this is true and extent
For example, on a recent video I was working
This is a bad example of this was supposed to be for a comedy parody
But the same principle applies for example, I could take a anime image and converted into a
No still meant AI created faces and characters 🙂 I have this one scene where there are 3 archeologists in a chamber with gas masks on and its in a full comic style, thought maybe it is possible to change it into fully photorealistic 80s camera high budget movie style shot, but it still keeps it in comic style just adds some extra grain to it. I guess if it does not work that way then another possible solution could be feeding the image to gpt5 and let it formalize a good and very detailed enough prompt to keep all the details the same as in the comic style image and then feed it to nanobanana without the original comic style reference image but with just a very similar prompt with all the small details to design to create a totally new image in photorealistic cinematic colors
If you take this image, which is animated, I guess you could say
This trick works but only in sora
And it depends how you prompt
Believe it or not that the same character in the same image, but look at the difference
So it depends on the prompt plus the image and intent, which would be the context
Try Grok then. This sounds like more of an issue regarding stylization rather than face swap. Midjourney is good at tackling it, but it needs subscription and I also don't recommend it(awful service and community). So just try Grok image gen.
Mid journey is free on meta.ai
Well, actually, a good example could be the movie 300
It was based off a comic
K. Thank you very much mate !
bet before 2025 12 01 no
look at the feet of man, the AI know about human anatomy 🤤
Well they all struggle with composition
why do i get this message can't be retried
reset the app or site to recapcha again
the AI have a issue with numbers
i already did that but still this appear
is this a bug? or feature update
well, at least for now, there is no retry in BATTLE model.
it's not retry in battle mode, it's direct mode
It sucks composition really really sucks
Then submitting it in #1343291835845578853 may help
Nice. Did you do it in grok or nanobanana ?
Nano
maybe your message have something has a troble with politics of lmarena
politics of what
Just wait for nano2. See if it could end comfyui, tired of using it.
politics of uses, terms and uses idk the name
oh policy
ahhh ok 🤣 sorry
im just discussing about turttles dude
🤣🤣🤣
bruh
Sucks
somehow thus appear
I built a little frame tool
Built with LMArena - Content is user-generated and unverified
i thought when u say politics, it had something to do with politic discussion with the a.i and im confused because im just discussing turtles
wow 🤤
Thx ! Can you send the prompt you used for it to make it realistic ? Would try later also when i get home from work
To fill in the gap to the animation
I’ll confuse the model by saying shot from iPhone z7 Nikon to force the mapping to get as close to realism
bro i was literally right now prompting for a universe sandbox
It takes some iterations though
To be fair the hand anatomy of the original reference is wild🤣
Oh, I know it just takes iterations
Well the gaurdrails are so dumb too
The image on the left works the image on the right gets blocked
Thank you very much ! I will test it out in the evening
I like to use adolf a lot asa control variable
Since he’s usually blocked from most video generators. It’s a great way to have a controllable factor in your research when testing or mapping the image filter
yall think g3 coming tmw?
The thing is, it’s probably been ready to release
So either they are releasing it early or they’re holding off
Google probably tries to harden it against jb-attempts.
But in the end, PlinyThe Liberator always wins lol
This is facts
In in the gray swan arena
But hard to tell which model it is
It’s one of these models
But yall don’t know about grayswan lol
Not hardening them against jailbreaks
And you get paid for doing it
You know all the models are on here. They usually just have the tournaments win. They’re about to release new models for the final red team
It ends November 26
Usually, from my experience, the models get released a little after this event is over
But these are more for agenda tools, this tournament specifically, which I think a feature that the new Gemini three pro should incorporate
riftrunner is nuts https://019a922b-99d8-7c81-ae7f-6184d13b0457.arena.site
Built with LMArena - Content is user-generated and unverified
this ones better than mine
what was your prompt
no was this was 1 shot
everything with riftrunner must be 1 shot
Yeah
not always
i have a few outputs that were not 1 shot, i iterated them a bit
where is my video
