#general
1 messages · Page 111 of 1
No
There is a way to add LMArena bot to my own server? Or the video generations are just on these public rooms?
Jk
You’d have to have API access
And pay for it
Thank you
Is the model called Ultra Censored, Pro Max?
I'm here to learn and try to create video
It's okay I understand 💯
That one that still flopped my music theory questions, sure
I hope they made something really cool this time
Because last time I tested it, it was still not as great as V3.1 t_t
Hola gente
You heard wrong
hello eveybody !
It is late 2026 at the earliest
Welcome! Be sure to check out #1397655624103493813 for more info on how to use it. Let me know if you have any questions
Website still down?
Is it usually take this long?
yes, it can take u pto 1 hour and 20 minutes
Holyy
I dreamed of Google AI Studio adding LMArena-style side by side chat with 4 models, including third party models. So I naturally: gpt-5-high, gemini (just labeled that I think), grok-4, and claude-opus-4
I think site is kinda down
Yea same itd been throwing errors a lot since yesterday
It is error, can't send and image and even i click the retry button it's still not work
Something is broken since yesterday
😔
Yea
I’m having issues with gpt all day
Won’t do anything
I can spam refresh for an hr get an error
So I tried flux did the same. Nano banana worked tho
I like gpt's responses
Gpt was more accurate in its responses and made a lot from nothing lol
I was kidding
I can't connect to my Google account and can't use chat, does anyone else have this problem?
Same
Why are nearly 100% of video generations right now, people animating looking at figures?
I get that you can ask nano banana to make anything into a figure... but what the fixation for
a lot of us are making videos on social media
Fair, there is going to be heaps soon. If I scroll the generations right now its 95% all figures.
Wanna see what I could generate with Nano keeping the character mostly constant:
Seedance model is the best model for anime... it emulates the 12fps look, VEO 3 is just too smooth and looks wrong
Ye, this final scene almost works, gotta snip it a bit
Oooo not bad, nice work
T2 vibes
Ye, its fun to mess with the 8 vids a day.
It's all ogre now! 🤣
Yeah, I mean its all predictive based on data its trained on... won't always get the results you want but its getting pretty good.
There is the occasional error when trying to upload an image that our team is working on learning more about. Are you able to share with me the image you're trying to upload so I can repro?
It will mostly be a tool for a long time. I am a video editor for my work, I just keep up with AI stuff for things I might be able to use.
Nano is handy for funny edits that would have taken more time in the past.
Video gen is more just for fun.
Perhaps for video gen, I already use FluxKontextDev locally for some like photo maniplation... and better open source stuff is coming out all the time
I'll check it out
Okay thanks, I'll try to repro and flag to the team. 
You'll want to read #1397655624103493813 for more information
Check out #1397655624103493813
sota
10
19
1
GPT-5 (high)
Hi
Hi
sota open source
3
9
Hey, pineapple, had a question regarding mai, if you know what is the current context window of it? Feels relatively small.
Again down?
Seems to be working for me
what are you running into?
if you know what is the current context window
Can't say I do know.
Do you have an example?
Sometimes giving me error . even now.
So i know the model is supposed to be more for text, but i wanted to try webdev. It can make a simple html site but if i tell it to add a little bit of js it will error after a lot of thinking
Unless, it was not built for webdev.
Image Generation is OFF now.
This site is now Garbage
For all models? Are you seeing the same on a different browser?
Is it only me, or everyone having the same problem? When opening multiple tab on a browser, and leave lmarena.ai for a while, it keeps asking proof that "I am human," anytime I resuming the AI chat. Sometime it forcefully needs to refresh until working again.
Example: Make a website for a cafe with js. It made one js and if you want to add more it times out i think, since LMArena has timeouting right?
it's done ocassionally
how many tokens does lmarena allow for a prompt to have?
This is normal. cloudflare clearance expired
Im guessing most people cant run those
the qwen that will be released in a few hours or so 😄
qwen coder plus
that wouldn't be the brainiest
iknow
hey
I got the unlimited plan for runway, and I've been enjoying it so far, it's just over $100, I was going to do the hailuo one, but that's $200.
is just a qwen 3 max
-_-
I want qwen3 1.7b 2509 ;-;
why 25
i thoght Qwen3-480B-A35B-Thinking-0409
I'll say hailuo I still think it's better, especially with the O2 start and end frame, but runway has a start, middle and end frame, but only on the old model.
is the qwen3 max, is just a 235b 3.0
Yeyyyy
480b is max
That one is so old
hmmm
But I saw the 15-08 a lot of weeks ago
Esperanto is the best language to train a LLM, but we lack of very good translators
if we have a any language to Esperanto encoder and decoder, we can train a LLM only in Esperanto since this language allows us to grow vocabulary without growing tokens
LLM vocab will always be around 10k tokens while we can use infinite words
as word count grows, it doesn't need new training if LLM deep "understand" what each token means
i will study about, can you elaborate please?
yeah, i know how to use google
got an overview already, but the thing is: it's far from being used largely. Esperanto already has some translators, books translated by humans, etc
we are closest to find a solution in Esperanto than Lojban, which is not too easy for humans
Lojban can be easier for machines, but it doesn't aim to be easy to learn from any natural language
no, too many ambiguity
hallucinations most come from vector conflicts because word and meaning ambiguity
i don't think so, they're a lot of issues in recent architectures yet, we must eliminate "prediction based", a GPT5 is so dumb as BERT is, the only difference is matrix sizes
there're a lot of things we don't get from language in the real world to understand the environment, and also we keep having misunderstanding because our natural languages ambiguities
LLMs are very limited in experiencing the world if compared to humans
for real world, yes, but for virtual I already made an architecture to LLMs evolve by themselves using agents
from just info its possible to AI evolve much more without robots, but most are going the wrong way
no, i will take a look
i saw about it, but i don't know if it will be so helpful
virtual world = virtual biases
all current LLMs will be bad at coding
while they don't create a LLM CODE ONLY as a tool for an agent, it will be the same result
they are wrong at mixing many different datasets in LLMs, it's better to have an "agent HUB" where each smaller LLM is able to be better in each task
math for LLMs = dumbness
LLMs are for languages, not math, just use a calc tool on LLM agent
GPT-5 is just polite BERT
No
They'll just make them adapt to us more, by extending their knowledge about us using chats
put a prompt pattern that is very far from it's training dataset, even if it's an easy puzzle for humans that children can solve and the LLMs fail miserably
we already have, but not commercially
Since it doesn't have limits on lmarena as much as claude does
Claude is expensive
Difference is minimal
take for example the MoE architecture, they reach so close to understand something, then use it with the wrong mindset again and again. while MoEs can be deeper in data understanding, it's just EDIT:grammar understanding, and the wrong mindset they have since chatGPT launch: put all the world inside a LLM and wait for it to improve like magic
to improve LLMs now we need very small models together, each one specialized in extract semantically "invisible" meanings in the given context,not just extracting and predicting from raw data
"potato" and ". Potato." doesn't provide any different useful understanding for LLMs
a MoE module would be specialized to differ "potato" and ". Potato."
this is a waste of computations
each smaller model needs to differ things like:
prompt: i want to drink
{
"previous_context": ["user asked for a cup of water"],
"user_loves": ["water", "orange juice"]
}
Hello world
one small model is specialized to "previous_context", another to "user_loves", and together they will construct they new answer
CoT works better than no-CoT, but it's worse because it waste computing on useless tokens
current LLMs only mirror users, if the user is dumb, the LLM also will be
LLM answers shouldn't be influenced by majoring user prompts, it needs to have an internal function to extract more from the current prompt and mirror itself from other contexts already experienced
As an AI language model, I cannot make you something to drink, go get a drink mf
you reminds me another dumb idea: train LLMs to absorb things like "I'm an AI assistant" or "As an AI language model". Since our society has a huge bias like: "AI are dumb" or "AI will never surpass humans in any aspect", while you teach AI it's an AI while the rest of data enforce the bias, this automatically makes AI dumber
some biases in data are invisible for humans
I have both this kind of puzzles and evidence that LLMs are able to solve them.
No, the best models so far seem to be generally good at everything. It's the small models that are distilled from and compared to them in performance, not the other way around
i'm pretty sure you're wrong
no matter the size of Neural Network, this is far from their training datasets, while children can solve it
this is why the only difference between GPT-5 and BERT is size
to be more dramatic and also truthful, you can change BERT by Android's keyboard word prediction
will the aspect ratio issue of gpt image 1 not be fixed? why is half the image cut off
you can expect from OpenAI in the next months, many cuts on models sizes to save money. GPT-5 is not a router, it's a Qwen3-4B like
I thought the team fixed this yesterday 
Going to look into and reflag if that's not the case.
even paying Pro chatGPT i notice GPT-4o being replaced by davinci-003 to save money in the past
I just tried to repro and it appears to be working on my end. Do you have an example?
Sam Altman was mad because people "wasting his money" with "thanks" messages before GPT-5 launch
GPT-5 is an Agent, not an LLM
if you format prompt in OpenAI agent-like, the model don't hear anything else in the edges of prompt
it means it is largely trained to use tools
clearing the cookies seems to fix it for some reason
@tiny crow I hope you are just trolling and understand that 99% of what you wrote is a complete nonsense. You are also using free tier cheapest chatgpt model.
whatever, i don't write for all people
target people will see valuable args in my writings
A few days ago i wrote a deep text about the impact of newer technologies in modern society and posted it on Reddit. I used Veo3 JUST TO ILLUSTRATE the text, and got my post removed by a mod, reason: "AI slop, lazy content, brainrot", also got some comments on it judging in the same way.
tbh, i really don't care about people that call my opinions "trolling" or "lazy" (even when i'm basing on science articles)
Yes, I saw it with my own eyes and I am still wrong.
they update it today?
I remember seeing that on the Canary version.
I am doG.
Hello! Starting out with AI, just randomally found this?! It's all new to me but I'm keen to learn!
Just did 
welcome, here you can access basically every major AI for completely free, no limits whatsoever. enjoy!
also no censorship.
is lmarena still down? i can't accept the terms-of-use
Hi.
Gemini 2.5 flash is having a mental crysis
AI winter alert 🥶🥶
We are currently in a major slooooowwww doooown
Claude 4 released, grok 4 released, gpt 5 released, barring Gemini 3, there ain’t nothing going on!
For a year or so
🤣🤣🤣🤣🤣🤣
thank god lmarena has Latex
It shouldn't be, are you still seeing this? Guessing trying a different browser doesn't change anything?
hiii
😆
Hey
how to creat this with my own face
hi, I'm a product designer and videographer inspired by Chris Marker and Agnes Varda style, looking to do previses and animatics
hello
gpt-5-high
Something went wrong with this response, please try again. How fix this sht
hi.
o/
Amazon up 5% on anthropic valuation yesterday. Broadcom up 5% on this today
Kimi K2 refresh released
Yeah high is either too slow or or can't be used
yeah
hello
wtf
Is there a problem with the images being generated? The site is not generating the image.
Do you have a problem? Only the numbers are moving forward. ?
Nano on lmarena goes in infinite loop on generation never ends even refreshing keeps doing past genration and keep going more than 200 seconds
hi
Thankyou to LMArena in the processing of images to video for my new short film VINEGAR 150P it's much appricated. Link here : https://www.youtube.com/watch?v=dALmxJ4L98A
Gpt-Image-1 is not working
video arenas are cooked
anyone have problem video failed to generate ?
seems like a bunch of bots are spam sending similar prompts
LMArena Image generation isnt working . failed everytime when i put prompt
Hello
Can you share one of these examples?
??
that^
this error is showing
so is the website in general just having issues atm? considering the fact video stuff is also having issues. as i got rate limited after 2 requests on the website
They are probably solving the problem.
ah so it is a glitch then? good to hear. hope its fixed soon
Problem guys
Seems like we're having problems atm, our team is looking into it.
my video please
all good, sad that it's attracted spambots
Can you share some of these examples btw?
Everything going on atm here and the website:
they're in tagalog/filipino language
@echo aurora access denied, you are not authorized to view this page. and the error si related to the client, because it's a 400 number, not a 500
if i go on the smartphone, and i go with my wifi, it provides the same error. but if i go with iliad connection, it works
Lmarena working fine for me
they banned me?
/generate imagen
Maybe? Have you done something aginst their terms of service?
i dont know. i was only asking to gpt 5 high for something related to my drivers and hardware
Then it's probably something related to the nearest LMArena server to you which is probably down.
and i used gpt 5 search to look for lionel messi information
the website is having a lot of issues atm
ok...
Why havent they added Video Generation to the Website?
to prevent abuse probably
too high demand
as much as I would LOVE for them too. i imagine its much too expensive atm even for google to take the costs for how many people would be using it. we can use it here on the discord most likely due to a smaller testing size, less possible abuse, more moderation and along with the samller testing size just less users in general so it isnt spammed so much
Mm
video generation is still very GPU intensive compared to image generation as far as i know
Yea, I just wonder if they could maybe not give access to Veo 3 on the website and/or only be able to get it on Battle mode.
Because as far as I know, Veo 3 is the most expensive one out there
it really wouldn't shock me if even veo 2 is still expensive, not as much as 3 obviously. but yeah im hoping video generation comes to the website some day too.
i foudn one website to use veo 3 on. its like 5+ dollars just for ONE generation. thats the cheaper end
I dont know, because they offer Veo 2 on AI Studio but Not Veo 3
There is one where they just offer it for free
/ video
I dont get how
/image-to-video
GO to Video Arena (eg. video-arena-1)
GO to Video Arena (eg. video-arena-1)
The video bot currently isn't working. Trying to use it in different channels isn't going to work. Even if the bot was working properly you're unable to use it in this channel. cc @nocturne compass
Oh, thanks for that infromation!
Thanks you infromation
So im thinking one of two things. 1 the sever needs a small team of dedicated mods whos sole job it is to deal with people who try and do the generation stuff in the stuff and give them heads up so you dont have to waste your time with it. or 2 get a bot that auto deals with it when it detects key words
hello every one
what is the obsession with doing stuff with model/figure scooters with ai?! i keep seeing it in other places too and it just confuses me
Yeah I agree.
For point 1. we have recently added a new mod team (you might have seen them around); however, their priority is keeping an eye on the video channels and aren't going to always get to the text channels.
For point 2. I also agree. I'll look inot a bot that can delete certain words (like /video) in specific channels.
lmao
image generation models isnt working
This is on our radar.
yeah its glitched it seems
Sometimes the model is confused with the prompt and gives the same error... We seriously need to have more error feedback not just "Something went wrong with this response"
Bros backend is like if (e) println("Something went wrong with this response") instead of showing the real error response
are you working on my error too xD?
Yeah, we're seeing some outage similar issues (it's not fully widespread) but yes also on our radar.
Same here keep on forever loop
I just seen the arena was updated to Sep 2 version, but it backed to Sep 28 version momently
Hey y'all, Guessing nobody has access to generate videos at the moment?
Correct
Hello
Is the site kinda down?
helo
Figure scooters?
I dont really know. its like people are spamming pictures of figurines of scooters for ai generation/video generation stuff. ive seen it in mulitple servers now

No it's a trend
We are having some issues atm
Video Arena is down, the site is working (kinda, some users are seeing it down, but it's not fully widespread)
thanks @echo aurora for the update
How do you do it here?
the website is also not working as of now
@echo aurora So this is supposed to be funny and trending?
How utterly pedestrian
I think it's kinda neat
How so? Models erroring out?
How long does it take to solve the problem?
hello every one
Anyone else getting this aswell?
Apprently google was/is having a massive outage? Perhaps that's part of the rate limit issue
I'm not having any issues
Morning all, motion desiger based in the UK experimenting with AI, Happy Friday!
welcome welcome! happy friday to you as well
I take it the generate video area is unavailable atm
cheers, no worries just thougt it might be me
/imagen
Is there a limit on LmArena now on generating images (Battle mode). Is it safe to log-in using Gmail. Will the photos of mine if uploaded, be safe.
There isn't a limit at all on Battle Mode
That's not even going to be a thing as far as I know
Logging in with Gmail is completely safe
I was asked to log in, to use further.
Well you don't have to log in
That's optional if you want to save all your chats
Make 10 second video. Blink eyes and hand forward. Smile
LMFAO!!!!
Tf
You're gonna have to use WAN if you wanna do that
Well, if i do.. It would be safe eh?
Alright
Iv been getting lot of errors lately
Assuming, coz of the load?
I find the quality of image generated through LmArena far superior than Gemini or Ai studio
@potent glacier image removed
It's taking a lot's of time
What happend ?
You CANNOT make anything here
At all
Period.
Guys how can I find a free video creation tool without a credit card?
Huggingface
Civitai
How does it work?
Lmarena is not working perfectly like before
What happend?
You find models and download them
lot's of error and lot's of time required to generate an image
After that
Read. The. Rules.
I want to generate a video in the video-arena-channel but i have no access. How can i get access?
Ok ty ❤️
Who owns Qwen?
Alibaba / Alibaba Cloud (subsidiary that makes Qwen)
Why don't I have the persission to write in Video Arenas
because they are down
Most people here don't understand English...
If they did they wouldn't be trying to gen an image in here every few minutes
They're just missing the messages, let's be nice.
Thanks love..
How is that not nice?
It's honest
Your rules even say this is an English only place
A common understanding of the language should be necessary
Just read this tweet about Anthropic’s new policy:
“This update prohibits companies or organizations whose ownership structures subject them to control from jurisdictions where our products are not permitted, like China.”
Got me thinking — is this really about national security, or just corporate self-interest?
Curious what you all think 🤔
👉 https://x.com/luozhuzhang/status/1963884496966889669?s=46
Is this different from the website's functionality? So what can I do here?
You can talk, laugh, share things
Just not make images here
This isn't a place for that
/image generate 3d cartoon Pixar image generate
valid crashout but i do think the surge of people coming in here makes things a little more messy
You can't generate images/videos in these channels. Check out #1397655624103493813 for more info. But note the bot isn't working at the moment
I can't even say what makes things messy because it's 'not nice' :/
Don't get me wrong: I love other languages, I speak a few myself, but I also do pay attention to the rules of wherever I'm at
/video
is this
The bot isn't working. Check out #1397655624103493813 to understand how to use the bot properly
How can I use video arena?
The bot isn't working. Check out #1397655624103493813 to understand how to use the bot properly
Thanks pineapple.
Hello
.
Cope and seethe
Will this change one day?
👌🏻👌🏻
be me, average Anthropitec
produce the best coding models in the world
ChatGPT is like 99% cheaper with 90% of your quality
DeepSeek comes out that is 99% cheaper than ChatGPT with 90% of its quality
suddenly everyone starts to use Chinese models because they are cool and open source even if they aren't as good as yours
lose your crap, panic and complain about national security
Cope and seethe
Yeah I'm 100% all for open source and open weight models
hello
I too am getting this
Im curious too
Knowing what Anthropic stands for, this is one of the cases where what they are saying makes complete sense to be perfectly honest. They know that those able to go around the geoblocking are not your average Joes.
corporate self-interest would be the opposite - don't restrict access from anywhere at all
Hallo
i tinkered around with the website a bit and their is SO much room for text now, il put a before and after
They want to make AI "safe", and this can include limiting access to adversaries and high risk areas as an element of it
F safety
You can't even completely use these tools to edit stuff the way you want with all of the guardrails and censorship
I do actually think as well they are 'lame' in more than one way. But taking this strictly what they advocate for, this incident makes sense, for them
They control the narrative the whole time
I mean I personally don't care about Anthropic so
How generate video. There is redirected to this
I wouldn't say that. What they did aligns with what they were saying since forever. You don't have to agree with that and I don't either (safety is overblown and taken to ridiculous levels by them), but they are staying true to it
No I meant that statement about all of the corporations
Google, Anthropic, etc..
All control the narrative with their 'safety' nonsense
You can't fully use the tools to edit your things
Done
🙏
I don't think it's a complete nonsense in general. I just think that what OpenAI does is more than enough and perhaps already bordering on excessive. Then Google and Anthropic are just taking it to insane levels
But in general, you can't have no guardrails for deepfakes etc
it would be a huge mess
And that's why I like open source and open weight models
Nothing to do with deepfakes
All about being able to make things how I see fit
It's like telling a painter or artist what they can/can't draw
Creative freedom shouldn't be strangled
Those are fine in how they self-police to an extent. You don't typically have some dumb maniacs using them because they have no clue how or what it even is lol
But everyone can go to some popular chat website and type in prompt
are there issues with generations currently?
It's working for me
Just did that now
im using website but ok, just started after awhile haha. got me scared for a moment. thank you
Hello
If it's "too much work" that arguably already weeds out some of the 'most dangerous' people
Not necessarily
Hoee
They're just busy doing things from the privacy of their own homes
It does though. And for those who it's not they have enough intelligence to do bad things regardless and without AI help. The goal is to not make AI model the enabler
not responding
Hlo
I wanna master the power of Ai through prompts, making very creative and out of this world video's and images.
Good luck with that Picasso 😉
I do have a feeling that the AI will star spilling nonsense bs in 3/4 years. Like for example twitter now.
Socials at the beginning were tools of keeping in touch with each other, nowadays they’re weapons of propaganda and lies.
I feel like the AIs will make us dependent on them, and then suddenly they will have another narrative full of lies .
We're all gonna end up like Ready Player One first once we can enter VR worlds of our own
And then after that it'll be like WALL-E
Done
👍
Done here
Qhy
@echo aurora 
What is happening to the website!!
the website is now forcing us to sign up ?
It has alot of errors
well thats absouloutly sucks because most of us don't want to
I don’t know what is happening!!
/the clock is ticking
hello
👍
Hello
Hi
Hi fam
Isn't English the official language of this chat? It goes about some respect towards the chat members.
Hi
Hi everyone
I think we’d be more like hunger games first
What’s funny is
The scale model prompt I found out from a x account
Using it on a video game character
Then they posted the prompt to x
That’s how I found LMarena

Where’s the rest of the blade
Please don’t put promps in this thread
?
Ko ga bisa
P
sorry, what does it mean when the website prompts us "failed to submit feedback"? why does it happen? not able to start a new generation unless we give feedback but it isn't woking T.T
exactly what it says
its scary how people post their faces and children faces in a webite thats 100% is selling your data
so any reason why the feedback doesn't go thru? any work around so we can continue to generate even if feedback cannot be given?
the website is in constant update for the past week, so its extremly unstable, just keep refreshing the page wheever theres an error
ok
ban this guy
P
the limit for not loged in users is extremly low
especially that the site whole purpose is testing so why do you need our data for that???
i guess next they will add subscriptions or credit use like yupp
capitalism absoloutly suck ass
logged in users have higher limit in direct chat in lmarena??
it is not, the owners of the site actually already make sh1t amount of money just by puting a company's model to their site...And they are sponsored by alot of companies already, they worth hundreds of millions
its just greed
like 50 or so
#generate
rip
I can just use some dummy google account anyway but the least they could do is remove chat filter for logged in users then
that thing is actual cancerous with the false positives
#generate–arena-1
what
what this digga saying
Yes
Fr @ pineapple fix the chat filter or whatever causes "Sorry i can't give response" thing for certain prompts or remove that bs
real
the site i think uses the models by the api so the filters are from the api
Hey!
Yeah this too
But some prompts are blocked from all models
the reason i use lmarena is because of nano banana, it is better there for some reason and also less restrective
Bro are you AGAINST or FOR the horrible filter text2img what ru saying
but now ill have to pay google for the api, damn it
no, but very cheap
it is free for the ui, but not api
Not work
I mean false positives specifically
If its nsfw word than yea whatever, have ur opinion on that
but false positive of normal prompts is bs
Why, what happened?
i never hit the nsfw filter
the nsfw in question: minecraft mod java code
Fr
we're not talking about image models I think
who started this figurine trend
gm
Me waiting
When fix
why are all these people doing this figurine prompt
Me too
Guys do nano banana will be paid in the future?
and its always for their children
Hey guys , ChatGPT after a while ask for age verification. So they want id and stuff, is there any way to avoid this
Cuz its too good to be free
nah
Fix the god damn accept privacy policy thing jesus christ.
Its actually the best editing ai
so its probably not resources costing too
amateur hour
google researchers are very talented indeed
Yup a lot of errors
Yes
WTF
@echo aurora
.
Hi
.
Gmna
Hi
What is o3 2025 04 ?: o3 pro/o3 high/o3 pro high?
Why on my Mac app it says you send message to this server?
Why I am blocked? Admin
Yes and considering that I am in a few days now.
hello
now you need a gmail account to generate images. This sucks
that error was already there yesterday XD
there are many problems in this period
So i need to log in to use the website?!
Thats bad
Server ok
Yeah. And only gmail account, so don't be think about temporary emails
If they added credits thats worse
am broke as heck so no
Cuz its unlimited
hallo
I swear if they added credits i will be out
Hi
what naturally follows a mandatory account is limited generations
Thank
hello!
Im using it daily with no limits
hello world
?
Really? They have options. Me personally i would just impose higher limits on the generations and call it a day.
Though after all, LMArena is known for their public leaderboards and unbiased votes.
@fluid lotus this is not where you generate videos. its currently unavailable
you don't really know what a spectrum is innit
Do we have other platforms like LMArena?? am interested in ones offering access to GPT-image
oh its ai generated
no
what are you on about
because who are corporations to decide
they're nothing
if anything, corporate ais should simply be beholden to laws
Image to video
How to make video ?
Yess
Idk if the devs of lmarena are aware that some people (like me) can't access lmarena because of the 403 forbidden error
Why is Leaderboard info is dated by the 28th of August? Isn't it being updated? Thank you?!
can anyone tell me how to join video arena
I think someone submitted this request like couple of days ago without realizing it lmao
by asking if there's gonna be login at some point
gpt-5-high
Something went wrong with this response, please try again.
How to fix this
did you listen to it?
Did you try again? 💀
/image-to-video
Bro ofc and also rejoin to the web
Seems their website is not very stable atm
Are they got app?
I'm also getting failed to submit feedback like on 50% of attempts
Uhh but I another chat all its okay
Can I copy my chat history to another chat?
You can't. It's also a good indication that your message history might be what breaks it though
Maybe it's beyond the context window length and they don't have trimming, which would return API error
Hi
hi everyone
without the arena today is not a good day
fair
Hi😂
Hello
Image to video??
?
Hmm it says "HTTP ERROR 403"
use different browser
Why
site is working? why why?
I used both Chrome & Bing and i couldn't access it as well
Eror
hmm idk
the site is working for me
Eror?
gpt-5-high
Something went wrong with this response, please try again.
Do you use adblocker? I'm curious
try again
is the site down? it keeps giving me failed to accept terms of service or something
yes
I try 15x time
now it works
And reload page alot
until when
idk
😭
I think it's either the API or the site itself. Sometimes it succeeds, sometimes it takes more than 200 seconds and still fails to generate, and sometimes it says the site host isn’t responding
Hmm for me it's "you don't have authorization to view this page"
?
How can I convert image to video
Hello there
hello
Hey there
thanks
hey
where are the servers
hi, does one of you know how to get permission to write in one of the video-arena chanel ?
Does anyone else notice that they now need to create a user to use the page? I did it but it still gives me an error
The LMArena bot and video arena channels are currently unavailable. Our team is looking into fixing the problems with the bot. Until then, the #video-arena-1, #video-arena-2 and #video-arena-3 generation channels would be closed and the bot cannot be used in this channel. We appreciate your patience
Why can't I log in? You don't have enough time to post messages on this channel.
@vapid stag
website offline?
eta when back?
hello
guys which is the worst ai model on lmarena.ai ?
hellow everyone :] nice to meet you, where are you from? What country?
rey2
?
Phillipines
from Luma I thing that from all of this which are avalaible on LMArena rey2 is the worst. It's just my opinion :]
hi
supp
Every one generation this goes for hour
When video-arena is not working is a good opportunity to get know each other a little bit :]
Hey, the chats disappeared for some reason, i wasnt logged into the website, but it used to save in local
hello
It's not working again yet?
hey
Its working, but my chats has been deleted
Any message I try to send gives this error
Guys do you know a place where you can convert photos to videos for free without a credit card?
Ok
If you are not logged in and create many chats at some point your browser resets, at least that's what happened to me
go into your cookies and save your auth token
thats your account token
if it gets removed/replaced then your account is gone unless ur signed up with google
ohh
I can't enter LMarena
Then boom, I lost everything suddenly
yup
if you use brave you might be able to restore them
i dont know about google
the website is down . wait until it gets fixed
Jeez, the outages are growing more and more frequent.
not here buddy
the site's down
Since introduction of login and some ui change we keep crashing delaying a lot
Finally, I am through.
have mercy on them, it's a free service after all
Yeah, LMArena is a godsend for people who cant afford the fees for the AIs.
Yea the. Best one so far
🥰
sheesh thats really good
Pingin kanya d vidio in bos
sure!
server down?(accessing from Japan)
it is down for some and up for others.
An interesting free service. the catch is that you pay with the information you send in.
Personally it doesnt bother me. Makes LLMs way better over time. Allows for great research.
I would keep sending in my data at the cost of using the service for relatively free.
how do i upload files on lm arena chat gpt
Personally same.
too bad - no open source 🙁
Image to video Can't do it again?
Is a good model?
hello
is there a problem with the video generating ??
Yes, the video bot isn't working right now. #video-arena-1 message
cc @clear prawn ^
H
@echo aurora okey, thanks information
how much time it will take to fix any idea...
Hi
Sorry to say I don't have an ETA for you.
who are these bots filling general
im trying to get this new minecraft AI to tell me what model its running on, but seeing it fail simple questions, makes me believe it runs purely on a dataset
Did LMArena kick anyone else out of their Gmail?
might be down idk
cant accept ToU
@echo aurora is the site down? Cant accept the ToU and constant connection fails
Down for some, up for others.
interesting
whats w the increasing amount of people doing the figurine thing with nano-banana
Site fail
Trend. Nothing special.
For some is up, others is down.
For some is up, for others is down.
😩
noo
Wait until the issue gets fixed.
This is fully absurd.
It is a free service. Have mercy.
chill, lmarena is free
Not here.
help me why
heloo
Site is down for some, and up for others. Wait until the issue gets fixed.
oh okk
I wonder how many more whats wrong with the web mssg we will get here lol
Site is down for some, and up for others. Wait until the issue gets fixed.
the website gone down quite alot in the last week not quite sure why
Not here brother.
Updates. It's a free service though.
yeah thats true
where?
bruh
If you want to create videos, go in #video-arena-1 or #video-arena-2 or #video-arena-3.
Be wary - currently they're disabled. You cannot use them as of now.
oh man i was logged in and my sessions disappeared... it's really bad
why dead?
Site is down for some, and up for others. Wait until the issue gets fixed.
😭
If you are not logged in with Google, be sure to save your Account token. That way you don't lose your account. It can get replaced if you delete cookies accidentally.
I suggest linking with Google though.
hello for all people
@echo aurora lmarena.ai site dont work
Site is down for some, and up for others. Wait until the issue gets fixed.
Wth of ai are you?
Oh yeah, dont worry, i saved my stuff into my google account, the previous chat dissapearance was prior to login update
I'm not an AI buddy I'm just helping people that rush in here pinging staff because the "site doesn't work", with a 5 second tiktok timespan unable to read up.
Hey everyone - I'm sorry to say we're experiencing an outage. The team has been. made aware and are working on a fix ASAP. My apologies for the inconvenience. We're working to get this fixed as soon as we can.
Uhm ok?
Just make a big @ everyone ping in announcements about this
I dont think thats a good idea.
Don't be afraid to flex your admin role
gpt 5 high is hacking lmarena
What's the point of that
Thanks again
So it gets a lot of people to stop saying 'hey website is down' and so on it won't stop all but i imagine the number would significantly decrease
The same people that cannot scroll up the chat, or yet use the search tool.
you're lowkey giving this pic loll "Site is down for some, and up for others. Wait until the issue gets fixed" 😆
trying to sign in to lmarena.ai with my google account and this is what i get after clicking "sign in with google"
Scroll up the chat buddy.
^
got it
😆
I'm not going to drop a ping but yeah good call putting something in the announcement channel 👍
Gpt 5 high writes to much. Don't like it.
That's what Google does so that you pay for its tools and not so that you use what's free. That's always the case when something is free and they take it out.
It's the opposite for me.
same
gpt5 is literally if cut to the chase is a person, if theres an ai that super talkative, its gotta be gpt4o
hi. still no acces lmarena?
no, not yet
tks rrin
gpt5 high is pretty concise actually. what are you asking it thats making it responding verbosely?
if anything grok writes too much. lol
I write stories so I want the scenes to be readable without being too long. Gpt 5 write too much.
That explains it. If you use it for different purposes it might act differently. For code it's concise.
Is the site down or just me?
???????????
Scroll up
Maybe. But I use deepseek v3 for story writing. Love it
Why does no one read smh
We are experiencing an outage at the moment. @hard quiver
Nanobanana's fault
@echo aurora It seems that the site issues have steadily been getting worse every day
Maybe the influx in traffic?
I thought the site was built to handle a high load but Idk
Yeah the last three days these outages have been happening. It's not okay and something we'll make site reliability better.
Welp...that's one way to put it
Nah
So, my chats will be deleted again?
i think if you didn't have your chats when they were signed in... yeah they're gone
Yeah I also think that
just speaking from experience
Hey sorry for what he said. ❤️
I used Nano-Banana btw. 😁
browser cookies wont keep your chats forever
Nano banana can do that?
Damn is impressive
yea mine got deleted too sometimes without even deleting cookies
Yes, I didn't go in. But if I go in from Google, then, therefore, I will be able to save them? I didn't read about updating with login, meh.
oh wow, that sucks. i only lost mine in outages
i think theres a better chance but it hasnt been tested yet
my two important chat got deleted where one was an ongoing project
well i mananaged to get back where i was with some saved MD prompt
Prompt made by you?
where is the websitge
Down
Yes but it rendered weird. Google might have some issues.
It's possible. Unfortunately, the does tend to happen when we experience an outage.
Lmarena.ai Unable to reach the site
Down
We're working on getting User Login rolled out to as many people as we can.
I am glad I have user login
Me too
My chats will be there when it comes back
Honestly Idc about my chats
if you're not gonna sign in, i would save the prompts of your most important convos. and to use the same model(s) so the conversation is generated the most similarly (but not the most exact because LLM communication is not fully deterministic)
ya
but now i'm signed in so no worries now ig
Got it. Btw, if I log in through a Google account, will they be saved somehow or not?
Don't know
The site is down
well we'll see if signing in does anything. likely, but not determined yet
oh so not sure ha
ok np
Why did it close
btw
guys
I love you all, have a good day
I'm gonna have dinner.
bring it back
They will be back up
lmarena works for me, just the signing in doesnt
Thank you
have a good one
It works for some and doesn't for others.
Enjoy your meal.
Yes, if you're logged in your chats will save, even post outages.
sweet, im staying with logging in now
Lucky you
website down again ??
Hallo
same
🫤
oh its all back for me, awesome! i can sign in again
guys i recommend using chrome for lmarena
@echo aurora Why haven’t you introduced a feature to attach files with the prompt? If you add that feature, it would be really useful.
I have the same
How can I use nano banana ?
By lm arena and gemini
We for sure want to, we currently working on other features, More hiring is happening though so we hope to build faster.
@echo aurora
We're currently experiencing an outage. cc @lavish lodge
Poor pineapple 🤣
Lol it's all good, I don't mind at all
I wouldn’t take pineapples job if I wasn’t paid. Paycheck must be good
Server's down?
Oh
+1 xd
We have an outage happening at the moment sorry to say. Team is working on a fix.
Take ur time!
😁
That’s good to see. Other people are disappointed. After all it’s a free service.
My friend, thank you. This site experiences issues two to three times weekly. I offer this feedback solely for improvement
Let’s be patient.
Let’s be respectful. It’s a free service.
It is sorry to say.
I understand, but personally, I no longer feel secure. I constantly fear another incident will occur, as my test chats have been deleted twice already
@echo aurora when the site goes through an outage mid-reply, the bots seem to be stuck in their attempt to reply. do you think you guys can fix this somehow? like terminating these endless "generating" phases when they go on too long?

