#general
1 messages Ā· Page 246 of 1
I can't use the "Smart Router" function. Can you make it work?
email verification link is broken
http:///nextjs-api/callback/email?token%3Dpkce_255483a15ecfbf8fe7e869728763bb636b00643eebb4fd419cd8511c%26type%3Demail%26signup_intent_id%3D&source=gmail&ust=1769756852297000&usg=AOvVaw0Grn64s_nRXEEoaxD_-IGB
is this project vibe coded?
algo google login does not work, not able to download source zip
what can i do to login and download generated source code zip?
please help
I want some tips from the people who made the new video on #announcements
it's really smooth, and I especially need to know how to do that typography animation
<@&1349916362595635286>
I want to request that you please enable the chat search, as if you have a large chat history on LMarena it becomes very messy to figure out the old chat.
Just use built-in browser search tool
Can this ai generate violent vidos??
we are violating the tos with this one š„
we are reading the ToS with this one 
hello
Your chat gone now š
U deleted chat?
but my prompt was "mango mango MANGO MANGO MANGO MANGO MANGO"
It will be in history
Bruh why
š„² ( if you don't know what I am reacting to it froze on me mean it registers as the previous response after I did a retry now it's stuck won't let me continue)
hello guys, how add files now?
the quality of seedream-4-high-res seems to have gone down a bit..im getting mutations with human pics that have never happened before.. š„²
morning
Hello friends, would any of you like to join me in a competition to see who can create the best website with the best image? We'll build a complete website using [method/method - context needed].
via Arena ai
Why it flags any edit or correction for a character when using a character sheet as a ref?
It used to work fine just one day ago!
What's the difference to CanaryArenaai
It has the latest experimental features.
Use canaryarena.ai to upload PDFs.
Hello friends, would any of you like to join me in a competition to see who can create the best website with the best image? We'll build a complete website using [method/method - context needed].
via Arena ai
Now we want enthusiasm from anyone who wants to play this competition with me. Tag me!
MOD WE GOT A PROBLEM
A cross-origin resource sharing (CORS) request was blocked because the response to the associated preflight request failed, had an unsuccessful HTTP status code, and/or was a redirect.
To fix this issue, ensure all CORS preflight OPTIONS requests are answered with a successful HTTP status code (2xx) and do not redirect.
2 requests
Request Status Preflight Request Problem
5629909088-terms-of-use?_rsc=nit70 blocked
5629909088-terms-of-use?_rsc=nit70 Response to preflight was a redirect
3765052346-privacy-policy?_rsc=nit70 blocked
3765052346-privacy-policy?_rsc=nit70 Response to preflight was a redirect
Learn more: Cross-Origin Resource Sharing (CORS)
Something went wrong. Please try again. has happening repeatly so i check the chrome dev tools
host-console-events.js:1 Failed to fetch RSC payload for https://arena.ai/ko/privacy-policy. Falling back to browser navigation. TypeError: Failed to fetch
at window.fetch (host-network-events.js:1:168643)
at async y (4429-f1b2cb3f81f68daā¦zNvCX2MezK:1:147456)
at async _ (4429-f1b2cb3f81f68daā¦zNvCX2MezK:1:146311)
at async Object.l [as task] (4429-f1b2cb3f81f68daā¦NzNvCX2MezK:1:73637)
(anonymous) @ host-console-events.js:1
_ @ 4429-f1b2cb3f81f68daā¦Tr8rcJNzNvCX2MezK:1
and also
host-console-events.js:1 Failed to fetch RSC payload for https://arena.ai/ko/privacy-policy. Falling back to browser navigation. TypeError: Failed to fetch
at window.fetch (host-network-events.js:1:168643)
at async y (4429-f1b2cb3f81f68daā¦zNvCX2MezK:1:147456)
at async _ (4429-f1b2cb3f81f68daā¦zNvCX2MezK:1:146311)
at async Object.l [as task] (4429-f1b2cb3f81f68daā¦NzNvCX2MezK:1:73637)
(anonymous) @ host-console-events.js:1
_ @ 4429-f1b2cb3f81f68daā¦Tr8rcJNzNvCX2MezK:1
403 and 405 happend again
sorry for ping
it just happened again.. š„²
whoops sorry
Huh
so....anyone knows what is that means?
I don't think these errors are very critical.
oh okay what about this?
this
same
Seems like gemini 3 image not working @echo aurora . just "something went wrong" error
so the reason of (Something went wrong. Please try again.) was error 403 and 405 huh
Idk
.
aw man i cant use nano banana pro
when this error happen something went wrong happens
Same
nvm it worked
well it fixed
You know what wuld be neat. For the wrods that cause "violate terms of use" to actually be highlighted so in long texts i actually now what i need to change
but we have to check it because it's sometime unstable
Yeah, it would be nice. Write it on #1372230675914031105
huh
strange i got used to the arena name so quickly
i was saying "I LIKED LMARENA MORE!" at the beginning but im meh with arena now
basically just a name change and nothing else
and new ui
They wrote that they changed the name because the service used to provide only text models, but now we have images, code, and videos.
i use lmarena to generate this idk why :/
Tbh I like new UI, but it's looks a little vibecoded š
i don 't get it. please explain š
is that a dragon ball star
hi
I am giving him the image, he is not generating the video.
I provided him an image and sent a file inside it because he takes single images and cannot take separate file images, so I put the beneficial images inside a single image.
Is gpt 1.5 image high fidelity really better than nano banana pro?
.txt file upload 
Is there a breakdown of ranks by finance/law/coding?
the captcha just doesn't budge and it just stuck at
"generating everything " is this like mass problem?
turn off adblock is solves captcha bug for me, dont know why
Kimi and DeepSeek scores surprisingly high on this one: https://github.com/fairydreaming/lineage-bench-results/blob/main/lineage-8_64_128_192/README.md
Afaik it's an independent benchmark so not really possible to benchmaxx?
yea
finan something
probably benchmaxxed
k2.5 doesnt feel like a smart model
nano is down?
Can you guys stop giving annoying and impossible recaptcha security?
Something unexpected happened. Please try reloading the page.
always appears every time it regenerates, especially when the nano banana error often occurs
At least replace it with a word or number captcha. Instead of using an image captcha. I'm not even sure you can pass it.
are we getting sonnet 4.7 today?
@echo aurora Model search in the leaderboard seems to be broken
nvm, i never see an announcement, lol
it was displayed yesterday
ooh
nevermind? i sorted by overall and it works
What does it mean?
lmao the arrow on instruction following is super tiny
It means that model has image vision
actually not just image vision, it means you can upload files to it
-# ig
Drop the "LM" just Arena. It's cleaner
Supports images
It means the model supports images.
They did the same yesterday.
Who is using the Arena.ai correctly now?
It's displaying reload button continuously
The web is down or broken? Yesterday was doing fine but now using even a dumb prompt it gives me this: anyone else?
also for some reason arena.ai doesnt work but when i put lmarena.ai it works so its better than nothin at least
Oh this search option is damn good
i have it everyday and you need to start new chat that's the only thing i also have no idea what's causing it
Nano completely nerfed. Gemini is so bad what the heck happened?
Is this normal?
Cloudflare Ray ID: 9c58932088fb3c79
Your IP: 2a09:bac1:****:148::4d0:6c
Yes.
Hmm...
You need to clear your browser history and data
u just leaked ur ip
btw
for lmarena site thats normal yes
it'd be abnormal if lmarean did work without issues
Ah, my IP address has been leaked. But whatever, it's leaked anyway.
Itās itās cloud flare thatās the problem
Im really bummed out how bad nano 2 got an Gemini also
check my ip 80.250.99.87
prolly fake but netherlands
its nvidas
nvidias*
I bet you the AI canāt do that
I used Chrome's incognito mode. It's still like that now.
Ah, speaking of it, Gemini Pro's brain is quite lazy and often stops thinking after a few rounds of conversation. From my experience in Google AI Studio, if I add "think" at the end of a message sent to the AI, it will thinkābut the "think" at the end of messages before that must be removed. I'm thinking if there were a feature that could automatically add "think" to each message in a round of conversation and then remove the "think" from the previous round, it would indeed be very convenient.
whos that lol
Gemini is nerfed hard
For the video arena can you guys make it so we can use direct chat like direct models instead of the battle mode
Nano 2 canāt even edit anything correctly anymore seems like
What happened
Maybe API works great?
hm
Turn of Internet, turn it on.
Reload website.
It works, thanks
yes
Arena has upgrade?
bye LMArena
hello Arena
Hello, I am unable to create video. I am new here. Anybody can help me with this?
Still not fixed issue with recaptcha
š
Anyone help!
I cant verify anymore!
Cloudflare was billion times better than recaptcha!!!
why remove the ability to create desktop applications
all it does now is the same format each time which is this
yesterday it used to build c++ applications just fine
@echo aurora
What does style control mean?
@echo aurora Also can you fix recaptcha issue and replace to cloudflare? I cant verify anymore and cant chat.
New ui is insanely good
which model can Alligator be, It's pretty good
Pray for me it's a good output
OpenAi needs to work on the speed factor asap
I can't take it anymore
what best ai for dungeon master ?
Human intelligence
Genie 3 today from google
3d world generator
They keep teasing with 3 genies on their twitter
Has to be today
any claude
yea claude is usually best
running on hopes and prayers
and maybe even gpt 5.3 today
when someone replied Genie he replied think bigger
another useless release lets go
honestly yeah i have no hope in openAI after bad coding frontend
they fix the frontend and my hope will go up
The AGI benchmaxx was extremely stupid for 5.2
isn't the main problem with 5.2 is that it's mega slow?
no its dumb
xhigh does take a lot of time
just to not get stuff right
well they're all kinda dumb sometimes, even opus 4.5 š
still #2 in webdev which is impressive
i used it only a couple of times and i just remember it's being painfully slow
models often fail simplebench
which is just multiple simple questions
Like
Anna had an egg for breakfast, what did Anna have for breakfast?
And you have to add more so it gets confused
at least this isnt just me, i find nano bana errors way more often with longer prompts
No bro I'm trying gemin app working and gpt app trying working
do you mean the same prompts are working on the gemini app but not lmarena?
Yas bro
yh it is quite odd how the same model works in other places, on the arena though every other image model for me seems fine.
ive been using gpt image 1.5, way higher generation success rate for me at least
Bro lmarena error ha ?
whats lmarena
Yeah "lmarena" were unsuitable after they added image and video models
Right Bro
???
It doesn't even think more than 4 mins
In chatgpt
Give me a prompt that took forever
What should I do if I want to upload a 327-page PDF file to a chat, but there is a limit of only 90 pages? Will I be able to upload multiple files?
Nope. Tried it on api.
what are you building
howre you even able to upload pdfs, teach me
just use AI studio
anyone got a site where i can just get 1 nano banana generation through for free? (if someone can generate it for me and it works for them, id appreciate that)
(higgsfield dont work either š)
Which lab's AI is best for: roleplaying, creative writing, world-building, GM (gamemaster)-ing, logically consistent stories, extremely long roleplaying games (spanning months/years), maintaining very realistic sandbox worlds, managing many NPCs which act like humans in game, immersive descriptions
2
8
hi
that actually sounds like it would be cool, how do u set that up?
higgsfield
does that have credits?
yeah, idk if their are any without them
however higgsfield tends to fail less often (it is slower though, like a lot slower)
i hate credits
if you log into lmarena though, you do get like 8 generations over 2 (i think its 8 thats a guess)
i have been using lmarena before i joined this server
im getting annoyed with being rate limited tho
is it erroring for you? on the arena?
being rate limited?
na i mean do you manage to get generations through without it saying error
its probbaly not working for me at all since im trying to edit an image over just generating one
Hello
hello

Hello im new & i use AI for my thumbnails
Hi everyone š
nice updated site guys, arena.ai rocks. Question, when i input an image and i want it edited or change its content with dall-e 3 i get the error "selected model does not support image input" what input must the image be if not png or jpg?
Png or jpg ?
man it's been overnight since I last used arena.ai and I'm still getting captcha'd. So waiting doesn't really work @echo aurora
that is what i am saying, i put jpg/png and both not working...unless this did not go for me which in this case, i am sorry
A place where you can find the best tools to build something real
Who has the security Verification error continuously?
me
nope
Oh my god
I waited overnight I'm still getting constantly captcha'd with a high failure rate
Can someone please tell me what this means technically?
Like I am in the middle of a conversation and then suddenly this strange black page with a strange text in it interrupts everything.
That's terrible
Did you use the old url ?
Nope it's canaryarena but even when I use the old URL (arena or lmarena), the same thing happens.
Oh
What happens if you use https://arena.ai/ instead of canary? When this error happens refreshing the page normally fixes it right away.
That's very strange. This has been escalated to the team. I'll share an update when I have more information. We'll be using #1451574502369656842 as the main thread for this.
Yeah I know and it doesn't cause much of a problem but I am curious to know why it happens technically.
I can't verify
This is only happening in this site
Unable to connect to arena error is happening, can you verify?
aināt nobody gonna recognize ts icon
Thats happening when clicking the recover button.
Hmm
Idk why but this is now the 2nd time i was logged-out automatically
This happens to me often; I can't access my chats.
Same
Just reload on that page
I already tried
Seems like people are experiencing this a lot more post site migration yesterday, do I have that correct?
The Something unexpected happened. Please try reloading the page error page?
Um in my opinion...that's correct
Its weird they took a literal seprate domain for canary
They could've done canary.arena.ai
Bro why is this such low resolution, its literally blurry on the site.
The last sentences are literally not visible. @echo aurora
Does a refresh normally fix it for you? Are do you get stuck here?
Did you scroll down
I believe the fade is intentional, but yeah seems a bit strong.
When I press the sqaure button of captcha for verify it's displaying immediatly
And then after a refresh, are you still stuck?
Same as beforeš«
Yes
But now it doesnt allow any scrolling and that button also doesnt do anything
Looks like you've hit today's video limit. Check back in 24 hours.
I didn't even generate a video and it said this. Why?
Okay I'll share with the team. It'd expect to see more reports of this if it's widespread
what a hell happened, cant load pictures
What's happening? Are you getting an error/
You can't scroll?
yep, error
No, whe i load a picture to change, but now seems okay
I think it because
Verifies the same token multiple times in back end or
Use wrong reCAPTCHA keys
Just my opinion
i just give it a proper prompt, nothing special (using AI-studio at google)
sometimes i even let Gemini craft that prompt
Ok, the prompt should include everything you want Gemini to do as GM.
(including all background knowledge, and also what it should not do)
Yeah
Full message is rendered
But the message is big & long
I wasnāt able to repro this. Wondering if itās browser related
If I wasn't logged in and all my chats suddenly disappeared today, is that it? Are they gone forever?
Here letās do this though, I donāt want #general being a place thatās just answering questions/bugs etc. It should be a channel used for discussing AI. Letās use our #1343291835845578853 or #1372230675914031105 forums so we can keep things more organized and easier to keep track of.
Maybe I need to spin up an ask-here channel for one off questions.
Video generation on the website is great, but when will we have a side-by-side or direct chat available for video generation?
is lm arena getting genie 3?
no
likely there wont be an API
U.S and ultra plan only
damn, so the best tech for creation is behind a paywall i wonder where this will lead
especially when the inequality gap is increasing, scary times we are living in
When I press edit image I get this error; without even uploading an image
its really weird
its a good dataset ngl
ive tried it on new chinese models k2.5 / qwen / minimax 2.1 / glm 4.7 and they all fails most of the time
gemini 3 pro on the other hand is giving me accurate answers which is surprising giving its score on this benchmark
its really confusing
lmarena is broken again
ok
prob because a lot of compute goes into one world
surprisingly deepseek 3.2 is answering lot of these questions right
how many generations do we get a day in video arena on free account?
any fixes for this?
2 or 3 i think, i forgot
im on chrome
like i said idk
ye thank you
arena.ai is new domain
it should redirect from lmarena.ai tho
thats the url im using
You signed in? Btw i think bugs and questions havr their own channel (for the future)
yeah i was signed in before this. and thanks
Id forward this off to askhere (like pineapple linked). Seems likr the domain move messed up chat histories. (One of the newest bug posts link to a fix on lmarena docs)
how do you measure compute?
How and what are you trying
Just start a battle, select one as better, sends another prompt
the new domain and website is cool and all, very happy about it but why do i get robot verification like for every single input prompt i type?
Chinese and russian api resellers
Still worth at $3/1k prompts for them with captchas
any fix tho?
Does anyone know how to delete chats? I can't find the delete option, it only shows the option to archive.
Fowarding this to #ask-here and will respond there
why wont it let me past this
bro
.... .. / -.-. .... .- -
i enabled a vpn and it worked?
i rebooted crome and it worked too lol
lul
Thanks, don't hesitate to use the @ Moderator ping
Thank you
Ask here channel? What a weird idea
anyone
We're going to try it out for now. #general was becoming too much of a question channel and would like to create more room for discussions related to AI.
I think it might be better to do it the other way around. Grneral will always be the first place newcomers visit
So let general be overwhelmed by newbies, and create a general-pro channel or something for people who actually hang around here
Its going to leave a very sour taste in new people's mouths if the first interaction here is them being told they can't ask things in #general
Starting to think OpenAI is going crazy. Like they are expecting to lose $14 billion dollars in 2026, and run out of money by 2027, yet they make another model just for them to run out of money even faster.
they cant make anything good for the life of them
That Google CAPTCHA crap is completely out of control, every action on the site requires a CAPTCHA and again it fails with every solution, instead of redoing the useless restyling of the site, can't the developers focus on the bugs?
Would encourage you to check out this message - #1451574502369656842 message
Yeah we'll see what happens. It's being tested out so may make a change. I appreciate the feedback and thoughts on this new setup.
@echo aurora Hereās my fanart of you
#1466367471895318548 @echo aurora Sorry for pinging but this seems like a significant issue for the legal side
Hey just saw that post, trying to repro asap

It's fantastic 
When did they add translator?
Thank you
lmao š¤£
almost all application react same way errors are everywhere

mr pineapple do you guys have internal testing like when AI company say "hey we need model test NOW" do you guys test the model or is that inside info]
#ask-here š š
oh sorry i forgot
is it just me or does Kimi Beat claude in some tests because Kimi is cooking
Hello fellow AI enthusiasts!
Give android app pls lmaarena
Can anyone here explain to me what ground truth is in machine learning?
I'm trying to study how AIs work so digging deep
ground truth is the a priori correct output to a given input
it's what you train against in your loss function
Yoo 210K members
Okay, I have a new question. If I get that infinite generation error again, I know I have to start a new chat. Does anyone have a recommendation on how to export an entire chat and import it into another one? Or am I just screwed and have to start from scratch?
game over tbh
When did they add "Smart Router"
Kimi K2.5 Thinking super high on the expert leaderboard š
The Gemini's brains are lazy and often just output without thinking after a few rounds of conversation. How do you handle this?
Iām here to provide helpful and respectful responses. I cannot translate or reproduce content that contains offensive or explicit racial slurs.
If you want, I can help you rephrase it in a neutral or non-offensive way while keeping the meaning. Do you want me to do that?
ā¹ļø
Hi
lawl peak
where can I find search bar?
============
Search Bar & Archive Chat Now Available 
<@&1372208635530448926> - Two new features have rolled out to everyone.
Search Bar - Your chats are now searchable, with the option to filter by modality.
Archive Chat - Archive chat sessions to keep them for later without cluttering your chat history.
With these features now live, the process for deleting a chat session has changed. Follow the steps in this article to learn how to delete chats going forward.
Thanks~
YES FINALLY SEARCH
converted the mp4 to webp
why is the frontend so inefficient?
it just loops a video instead of usin webp :p
what happened to the website why is every model having a rate limit or getting an error for sendnig a long prompt
This made my day! Thank you!
I'd encourage you to use #ask-here
Which one better for writing script guys . ??
Okay so basically the output we train the model to deliver for a certain input?
I like the old delete chat directly instead of archive search then delete
The issue is browser related.
The behavior or use patterns appear to trigger many of the problems
exactly
hello
Yeah basically Opus 4.5 for a tenth of the price
Tbh I feel no difference between Opus 4.5 and K2.5 for advanced SWE stuff
i'm working with cuda and game engines, opus is almost perfect on this, but kimi fails a lot
but for frontend stuff same level
maybe backend too, didnt test
On kernel optimization it understands sh1t almost as good as Opus
Who has been displaying the retry button continuously
opus is definetely better imo, but for being an open model, and its price, its pretty insane. better then gemini-3-flash at least (they're same price)
i really dont like rate limit but i guess i will have to deal with it
Hi guys!
how good is gpt image 1.5 high fidelity cuz it looking cool so far
Bro
can we generate images privately ? im new here
Is there no feature to change or delete prompts?
you can on arena website
It says reached limit, i keep opening up in incognito browser
once rate limit end you can use it again
in the same chat? it wont allow me
.txt file upload 
have you tried clicking retry
?
says try again in 53 minutes
wait until 53 mins ended and click retry
I have gemini pro that i pay for, but im tired of removing the logo and it isnt high quality sometime
think of the rate limit as a cooldown
np
can we also make videos?
Gemini and nano got nerfed
idk
So the entire generation is free, but all we have o really pay in is the cool down time
Yeah it did good at first, but idk what happened after 3.0
nerfed how?
Yeeah it was really good up to new year
i wouldnt know cuz i didnt know about it back then
2.5 flash was better lol then the current Gemini the state itās in
I herd 3.5 comin soon maybe that why?
what will change on 3.5?
any idea when that would release? its frustrating having to hit the try again just to get a worse image or the same
too many users not enough TPUs
so is gpt image 1.5 best image generation to use on arena?
workload isnt the best anymore
Hi
It depends on what you mean by best I like 1.5 but I donāt like the style
oh
But itās good in other areas
yeah there's no clear best
for example it took me about 20 images just to come up with this and had to photoshop some areas
it's probably one of the best, yes
was that made on gpt image?
i cant tell tbh
faded colours seems like nano
my eyes think of that one meme "they are both same image"
But hereās where it gets crazy
So this is gpt
But hereās where ChatGPT is really good at
It could preserve the quality of these other images which the other models canāt
The point being itās hard to say, which oneās best because they all have strengths and weaknesses
oh
that look cool
hiii
Guy's hey give me apps for android phone LmaArena pls
hey guys i have a question how can i generate sora 2 AI videos here selecting the sora 2 model?
how long does video generation take on arena website?
Hi
Pls join my server in which we regularly hold Giveaways Which contains Xbox Passes, Xbox codes, Any Steam game Permanent And many more. Please Join it to gain extra rewards, Thank You
join the 3rd link (viking) from my bio
gpt-image-1.5-high-fidelity how good is it really?
Imo not as good as š
I mean still great
But itās no banana
i heard social media posters designers use it alot
as it gives good effects compared to NB
hi
what?
is Gemini broken again ?
@echo aurora whats the point of the searchbar if you cant rename your chats
new GPT image is fine to me
Orginal
Letās do one nine anime
GPT
Nano
Original
You be the judge
nano banan pro 2k is not working
specifically in side by side, i dont know about other modes
Google API is typically quite finicky when you are getting rate limited. You don't need to wait long, like a minute at most and then it works. But when the API is shared another user may take the slot if you don't time it right and then you get limited once again. š¤·āāļø
btw just noticed the server logo changed. That icon is completely unintuitive to be honest
couldn't find it lol
It looks to me like this column in the middle:
wouldn't associate that with lmarena šæ
@echo aurora it seems that "manage cookies" option doesn't work on the website, it's only possible to accept all of them so that prompt window is fairly useless now huh
Oh ok that's actually intentional #announcements message
interesting choice
why all the sudden the website keep asking me for Security Verification? How to make it go away indefinitly?
you dont its to stop bots and things
this is madness! I just woke up!
im surpised it even exists how do they run it with no paid things
HYPER REALISTIC PHONE CAMERA A BOY COME IN DOOR AND HE TALK IN CREEPY WAY "TECHNOBLADE IS NEVER DIE" AND A DEVIL CAME AND GO TO HELL AND "TECHNOBLADE"
HEY AND SAY "HELLO FRIENDS I THINK YOU KNOW ME AND I AM GOING TO FIND A PENGUIN"
Make in Minecraft mode
is never die š
Sounds like someone had a fever dream
A cozy early-morning bedroom scene. A wooden study desk is placed beside a neatly made bed. On the desk, an open laptop is angled toward the bed, showing a live video call. Soft sunrise light enters through a nearby window, with the sky transitioning from dark blue night to warm orange morning tones. On the laptop screen, a young boy appears, calm and well-groomed, smiling gently and speaking warmly. He looks directly into the camera with a caring expression. He says: āHey⦠good morning. Wake up. See the sky and how it changes from the darkest night to the warmest day, just like your smile changes my life. Are you still sleeping? Wake up. Look at the window and see the sky. Hello, beautiful. Wake up. Good morning.ā Cinematic, realistic lighting, soft romantic mood, steady camera, high resolution.
guys the models is not on here
yall holy sheet. i just tested sora 2 vs grok imagine, and grok imagine won
how do i do that
guess wich one is sora and wich is grok.
luck.
the prompt was "an video of a hyperrealistic car crash inspired by beamng. the camera is in slow motion and there are two cars crashing in a T-Bone manner."
is nbpro broken rn?
Hello guys!
Yeah, I'm unable to login with the Google account.
they just switched to a new domain and now they have issues.
Both are terrible tbh š
im not russian tho
Hello chat
hi
still, grok beats sora 2.
@echo aurora
Yup
aw man
tuff shader
No way
gonbe what u doing here
Is nano banana pro working
wdym
it did lol
its uniplayz
the one on the right is grok
ik
it is but its kinda buggy rn
sometimes it works sometimes it doesnt
Ig grok done a better job
Nopecangetanyresult
Smh
new Arean, yes good! but how do I delete all my cronology and whatever? It gives to me aonly the 'archive' command
Please stop annoying recaptcha š¤¦š»
i need nbpro to work
Grok 2nd
seems like qwen has a new anonymous cloaked model on arena
i managed to get the system prompt for it
iam probably the first to do this
might get banned for prompt engineering
idk
Why?
the model in question is anonymous 1222
i thought u could get banned for leaking system prompts
I didnāt know that rule. Iām not aware of that.
That would be a ridiculous rule if that was the case
there is so much new cloaked models
ą¦ą¦ą¦ą¦¾ ą¦ą¦æą¦”ą¦æą¦ ą¦¬ą¦¾ą¦Øą¦¾ą¦¬ą§ą¦Ø ą¦Æą§ą¦®ą¦Ø ą¦¬ą¦¾ą¦ą§ą¦ą¦¾ ą¦Ŗą§ą¦²ą¦æą¦¶ ą¦ą¦¾ ą¦ ą¦Øą§ą¦ ą¦ą§ą¦²ą§ ą¦Ŗą§ą¦²ą¦æą¦¶ ą¦ą§ ą¦²ą¦¾ą¦ą¦Øą§ ą¦¦ą¦¾ą§ ą¦ą¦°ą¦æą§ą§ ą¦ą§ą¦°ą§ą¦Øą¦æą¦ ą¦¶ą¦æą¦ą¦¾ą¦ą§ą¦ą§ą¦Ø
hereās what i think is the new Gemini 3.5 model
wrong channel, buddy
it didnāt follow the prompt perfectly BUT the visuals look really good
3.5 code name snow bunny
interesting name choice š
man i can't wait for gemini 3.5 pro
seems like arena is broke for me
everything i try and do seems like it says someth ing went wrong
it broke for everyone
uhm me too
Something went wrong with this response, please try again.
arena broken af
only when u create new chats
im resorting to unused battles
Ī
ą±§
ā¬
ā¬
ā
no
ohh you mean the new coak model
cloak*
im trying to jailbreak system prompts rn
working good
found qwens new stealth model anonymous 1222's system prompt
gemini is not working that well
"Something went wrong with this response, please try again."
i am getting this message from gemini frequently
other models too
I need AI expert advice for transfer one chat thread memory to there chats, I started a conversation some time ago as direct chat with model gemini-2.5-pro. That conversation goes quite lengthy, I am gave my information time to time so now its been quite alot discussion goes on that thread.
**now the question is if we choose a single chat direct mode then how we can transfer memory to other chat thread? As the current thread goes quite long, so need expert advice what should I do? How can i move my that conversation to other new thread? really confused about that so serious help required. **
seond question is How to best way to use it properly? I mean do short conversations, every time? then dont we need to train or exlpain more about our issues, requirements again and again to chat models? what the optimised way? @echo aurora
#ask-here instead
Please create a detailed summary of our entire conversation so far. Include:
- All personal information I have shared about myself
- My main goals/requirements/projects
- Key decisions we made
- Current status of any ongoing tasks
- Any preferences or styles I asked you to follow
Keep it under 2000 words but donāt omit critical details.
write this and start new chat and copy paste the message it sent
2 thousand word limit is nonsense
why
The website is doing horrible since yesterday, I can't generate images without getting error even when I upload reference photos
Why not? Why is there a limit in the first place??
People really don't understand that their long chats are nowhere near filling the context window of today's AI modles.
try from incognito mode
thatl be nano banana, its completely broke rn especially for refrence photos
gpt image 1.5 is your best go to for now
guys i have a question how do i generate sora 2 videos here?
Yes, test in incognito
so will u fix this bug?
evening guys
now i cant see history chats
are you guys working on website or just in discord
can someone tell me pls how can i generate sora 2 videos here please it would be so appreciated TwT
really that captcha is annoying !?
Maybe it's internet, refresh and click again! Keep trying at least 5 times
when will i be able to upload .zip files
so i can export my entire c++ project for the ai to assist me in (claude opus 4.5 thinking 32k)
ok i tried 5 times but it dont work
it always show Connecting to Arena has failed. Please try again later or on a different device.
who?
cloak model that only appears in battle mode
oh
I can't do anything! the website keep wanting me to do securities check, im stuck in a loop
buh need fix
how do i get more brief understanding about arena and its various functions
fix what?
Is it only me or the image generation is stuck for some modelsš
Linux MintArena 
This feature was rolled out to everyone yesterday - #announcements message. Note that the proper method to delete chats can be found here - https://help.lmarena.ai/articles/9130232616-how-to-delete-your-chat-sessions-and-data-from-lmarena?lang=en (there are slight updates I need to make to this article as the UI changed slightly)
Can't say I'm seeing anything out of the ordinary overall. May want to try a different browser.
I'd encourage you to read this message - #1451574502369656842 message cc @obtuse smelt
Thank you! This is flagged.
there
I love the search function, but it doesn't have a delete button; you can only archive chats and can't delete the ones you don't want.
is nbpro working yet?
@echo aurora LMArena name and icon in Google login page
Besides, in my previous chats, they tell me 'the chosen model is no longer available', why?
probably because it has been removed..?
which model gives this error?
Almost everyone I don't use
Why My video generate failed always?
Good callout, thanks.
"Hi, I'm having a 403 error with YouTube remixing. How can I open a support ticket?"
Reminds me of "I'm not a cat"
The #1343291835845578853 channel is best for this. Be sure to provide all relevant details.
I by mistake archived a chat, is there a way to get it back?
Click on Search, then Archived, then the 3 dots on a chat and then Unarchive
Please add an option to delete a chat directly without needing to go to the search page.
Kimi K2.5 is the first Chinese model I'm impressed with. Only because it's natively multimodal. A text only model without vision is useless nowadays
Kimi K2.5 is just genuinely good
Defo beats DeepSeek as my daily driver now
I like that it actually doesnāt always validate you and instead says āokay yeah youāre right about this but wrong about thisā
From independent benchmarks it looks like its vision capabilities are on par or better compared to gemini 2.5 pro
Which was the sota for all of its time
I have lost my Gmail and now can't access my chats
Is there a way to recover them
They just disappeared when I tried to open the website
Log back in with the google account. Logging back will recover chats saved in that account.
If you lost access to google account, try recovering it.
We cant help in that
@left lodge
Any way of transferring my Gmail from that to some new
Or is there way to recover the chats from local storage?
I dont understand what you are meaning by Gmail
I think he wants to transfer his chats to diff one
The chats are liked to the account logged in the website.
If you log back in you get all your chats back
Currently there is no way to change email of a account
True
You can only either create a account with a email or delete if already created (deletion requires login/access to the email)
Their pfp is ironic with the problem they have lol
<@&1349916362595635286>
Thank you
@woven stream Hello! Please read the info posted here š ā ā https://discord.com/channels/1340554757349179412/1397655624103493813 to learn how to generate videos or images using the bot.
Bro even I didn't go this deepš
hey! New around but been using arena for a while ā¤ļø I have a question, I recently generated a short video and I wanted to use it as my twitch donate BITS animation, can I use it that way? also can I edit the final part where it says "arena" in all white, for the twitch alert
You can ask here: https://discord.com/channels/1340554757349179412/1466486650170245435
cheers
yeah anything you generate is yours
thereās no copyright attached
I'd encourage to review the Terms of Use here - https://help.lmarena.ai/articles/5629909088-terms-of-use
@left lodge
Any chance of getting the recover chats button again
I got it like once before maybe 2 3 days again
Hi everyone
I got a little deal
If you do my task i will give you something related to video generation

I still have Nano Banana tokens left. If anyone wants to generate videos, send me a prompt and settings.
You can't delete chats anymore?
Hy
Holaaaa

Hello
Any way to upload files to lmarena? Not photos files such as pdf, documents, etc
Going to followup in #ask-here
Hello
I wanted to take a real look at Moltbook ā a Reddit-style platform where AI agents run their own communities, post content, and interact with each other without humans involved.
This isnāt a demo or a hype breakdown. Iām walking through what Moltbook actually looks like in practice, what feels useful, what feels experimental, and where th...
Dead internet "theory"
That platform is no longer useful.
What
Therefore
You do your homework well and however you want, don't let yourself go
Hello! Im forky the new guy!
The bots are out (lots of people trying to circulate a scam)

This is now fixed btw. If you're still running into issues don't hesitate to flag!
@echo aurora I love this ai video generator! Its cool

Glad to hear it!
I'm desperate trying to run glm 4.7 flash from my simple notebook, if I knew that glm would release this monster from 30b to 3b I would have bought a notebook with ryzen 7
I'm about to donate my laptop to my brother and go buy a laptop with ryzen 7 7735hs or something like that for me to use the glm 4.7 flash at home
The glm 4.7 flash would be enough for my work, I would just need to do simpler skill and with more steps
Hell no, anti gravity don't work good in linux
Ahhh I see
But thinking about it if I had the glm 4.7 flash I wouldn't even need the anti gravity anymore
The problem too is cpu/igpu
I just have i3 1215u
For my use is 10/10 but now exist glm 4.7 flash bruh
Oh yeah no youāre cooked
It would be very crazy if I could do everything LOCALLY
Youāre better off using plotter and stuffing some credits
Plotter????
What the hell
Openrouter
How did that change to plotter
Ah no, I want locally
See my laptop/pc exploding when it making my docs
I'm seeing this now
Hmm okay... But I hate imac
Code in MacBook Pro
Send project files to windows laptop
I'm android and windows in the veins š
I'll see the promotions that are here
I think I'll get a laptop with ryzen with strong cpu
Ryzen is also a good choice
I've got my eye on 7735hs
Igpu almost gtx 1650 like good
Hmm
Maybe I need some more stonger
I'll use 32gb
I'm seeing the cpu of 7735hs the igpu is just bonus
Worst part is that I'm not finding a small notebook with a better cpu than the 7735hs
When there is i9 it has to come an rtx 5060 and 3 kilograms
I'm going to give up on my dream š
What distro of linux
Microsoft's shares plummeted overnight with a staggering $400 billion wiped off its value marking one of the largest single-day losses in corporate history.
Despite this, the broader US stock market remained relatively stable.
The wealth of Australian billionaires Mike Cannon-Brookes and Scott Farquhar, co-founders of Atlassian, took a signi...
@echo aurora wow the new like brownish theme on the site looks very good
@karmic bough
nice
gemini is better anyways
whats the best image generator ai for logos ?
Nano Banana Pro
No way
Theyāre finally realizing this model (GPT) is useless
That Gemini
Claude
DeepSeek
Kimi
Are just
Better valuation
llama?
Does anyone know how to stop vibe coding? I haven't written a line of code in two months, it's become an addiction for me
I donāt know Gemini turned into trash recently
Itās still pretty fire
Though kimi is my main now
Itās really good
i think this is the problem
this
Its true Gemini has gotten pretty bad.
It was really nice for a month
GPT more consistent, then current Gemini3
Gemini 3 pro and flash still good for me
and claude opus
i tried out antigravity's gemini
and its a total mess
it always gonna repeat like 6 times of thinking process table
but when i try opus 4.5 on there
it works pretty well
It was peak when it came out
huh i can't believe i went back to my old IT days apperently china has it out for this website and russia could be the same unless they got hit with a massive DDoS
fng cyber warfare is so god dam stupid when it effects the normal civilians who just want to go online
Well, finally it's the day Ai studio announces theyāre reducing their gemini 3 pro quota.
Gemini 3 Pro isnāt great at following instructions these days, especially when given updated information. It constantly reverts to outdated ones.
btw who paid for grok-4.1-thinking to sit at top 2 in arena chart when all the benchmark results are terrible?
https://www.youtube.com/watch?v=lQsvTrRTBRs
AI can finally do your homework for your fitness classes and your paper homework!
Last year, Helix showed that a single neural network could control a humanoidās upper body from pixels. Today, Helix 02 extends that control to the entire robot - walking, manipulating, and balancing as one continuous system.
Human preference optimization is one hell of a thing
Can you add Direct Chat to videos?
And with unlimited generations?
why cant we pick a video model
Because you can only select Battle
When you're in Video Mode
we used to be able to pick the video model at one point š
Yeah!
And LMArena got renamed to Arena
Since Project Genie got released...
I just want Arena to have the integration of creating worlds
@echo aurora does Lmarena support cancer? i cant believe you guys this is unbelievable how can you allow this
What do the file and glasses symbols to the right of the models indicate?
because of the bad word you got there
change fxxking to screwing
there's rate limits on gemini 3 pro nooooo...
Tell me, is it now impossible to delete chats?
This article will show you how - https://help.lmarena.ai/articles/9130232616-how-to-delete-your-chat-sessions-and-data-from-lmarena
Thank you
hey pineapple we got a problem GEMINI 3 is not working again
i think this is the problem (i try to find another error but only i got this )
CaptchaArena
well actually if 1 captcha appears it's fine but that son of a system is keeping spawn (LIKE "oh you did it now solve this ")
Nano Pro has been nerfed guys. Anything with Disney characters or Disney-owned IPs will now get blocked in the image generator after giving this message: I can't generate the image you requested right now due to concerns from third-party content providers. Please edit your prompt and try again.
So yeah, was fun while it lasted.
Not blaming LMArena on this one btw. This is entirely on Google's end.
wow, you are right actually
disney owns too much of a hegemony
Interesting. I tried generating again on LMArena and it finally went through, but gave me live action versions of the characters instead. But I the same prompt on Gemini, and it keeps refusing, even when I exclude the 'Disney' keyword.
maybe its an API error
google has stood ground for this long idk why they'd fold all the sudden
Isn't this Disney Character?
its unlicensed thooo
I posted about this in the Gemini subreddit and apparently, Google got a Cease and Desist from Disney themselves so that's why they finally surrendered and now block all Disney-owned IPs.
disney such š± 's bro hooly
But yeah, for some reason, it managed to go through on LMArena. Got Davy right. Gonna try and get a cartoon version now on LMArena, then I might get a bit of a picture here.
Yeah you can't do now
YES! Got this error last night! When I tried asking for an image with Hook, Negan, Garou and Vader in a Dark Fantasy D&D campaign.
Because they gave their rights to OpenAI
Yeah, feels random for now on LMArena but doubt it will last as long since they will likely update Nano Pro to be like the parent site and block all Disney stuff for good.
Still... OpenAI sucks
Not only do they suck, I HATE their artstyle so much too. There's always that watercolour filter they always add in their 'artsy' generations.
This part baffles me. I think GPT will rollout an update that will finally allow Disney content. But I'll actually be pissed if they only limit it to pro users so that Disney can profit.
Yo is there unlimited banana pro website?@magic ravine
I'm on Go
Interesting.
2 free gens a day on the official Gemini site/app
But it will always have that diamond watermark there
Is Snow Bunny in battle arena?
I am trying to generate an image of a wedding dress. Does a word "bodice" really violate terms of use?..
oh we got limits on how much can we use a model?
How does this song I made sound? Is it good?
Have you tried p-image
itās crazy good imo
like it generates pictures in like under 5 seconds
rate limits?
If it asks for a display name, just enter anything you want. But I found a glitch where entering the name as, "Gemini", it gives you an extra 1 day???
Yes, but can be bypassed by signing out and in again
Or clicking Regenerate Response, (though that sometimes doesn't work)
Yes
Well, if I remember correctly, Gemini Enterprise doesn't have rate limits, or it's just absurdly high
Who's gonna tell me to delete it?
The people in this server?
i say that cus theyve done it before
?..
how did u even find out that strat hooly
