#general
1 messages · Page 283 of 1
Hasn't Google already stopped using Gemini 3 Pro?
no?
gemini 3 is like one of the latest models lol
Correct
The announcement released by Google mentioned
How long will it take to repair this error?
yup... go to bed. it won't be fix that quickly.
...it's 4:24pm dude oh wait you're not a korean sorry
Man.
ah cool but nope, I'm not. :D
I don't have an ETA sorry to say
I'm seeing it back for me personally
but I don't think we're in the clear yet
I wish this would be more universal
Yeah seems to be back for now.
but we have to watch continuously
Verily
thanks PINEAPPLE the goat
Yeah I'm going to keep the stickymessage in place for now
It's the devs + community for helping flag that are the goats
Thank you for the quick response, Pineapple — everything works.
The bot IA dont working right now?
The Video Arena Discord bot has been removed from the server. You can find Video Arena on our site here: https://arena.ai/video More information can be found in this announcement about the removal of the bot.
Pineapple are u real human
Something tells me you have that as a copy pasta
☠️
The Video Arena Discord bot has been removed from the server. You can find Video Arena on the site here: https://arena.ai/video More information can be found in this announcement about the removal of the bot.
???
I'm just saying since you know the prompt not going to work on Discord who thought about that's not on here anymore
I had this issue with grok 4 fast.
With Claude 4.6 Opus here.
Is it rate limit?
Hmm
What makes you think website?
As it is a Capcha error.
Hm mine isn't I don't believe.
your gemini error again arena
fckin annoying man
before the dumb error was working perfectly. always like this per 2 days in a week
hi guys, for matlab coding, what model do you suggest i use
so everyone is getting this aswell
Same here.
white horse
And what does my profile gotta do with being warned for, and I quote, " Note that Video Arena has been removed from the server. More information can be found in"
I'd also note that the video arena has been removed from the server and give more information that can be found somewhere - after seeing that profile
guys are you also experiencing this?
i literally moved to a new chat that didnt even get rid of the error
i hard refreshed and cleaned browser cache too
when i also retry it just keeps showing me security verifications
i mean i could report this to the bugs forum
but maybe there is a way to fix this or do i just wait?
i am basically unable to send more than 4 or more prompts after that i just get that error
even being in the new chat
how is possible that gemini 3.1 pro is worse than sonnet 4.6 in coding?
thank you very much i will do that
this one or this one? arena-auth-prod-v1.0 or arena-auth-prod-v1.1
try both
just remember it will reset chats but most likely fix the issue (u may have to do it couple times)
if u have an account then idk if it works
i never login
by saying reset do you mean the chats will get deleted?
or will messages from there be deleted
uhh thats the same thing actually
yes it delete chats, but if you have an account then u can just log in again and ur chats are NOT deleted and will still be there, but idk if the errors persist/return on login too
if u use lmarena without login, only then chats will be gone but error fix rate 95%
sure
oh my i forgot to copy my important stuff
thankfully it didnt delete those neither any chats
i will try to send a message
ok.. so the same thing keeps repeating i will try again
arena-auth-prod-v1, arena-auth-prod-v1-code-verifier, arena-auth-prod-v1.0, arena-auth-prod-v1.1
which stuff do i delete there?
it seems like 2 new stuff appeared
did this right now i wont repeat this mistake again lol
btw i do have an account
real
it maybe don't work if u have an account/logged in
because it's some sort of ratelimit issue (yes their ratelimit is really buggy)
and its tied to the account
Is this repo useful?
https://github.com/simomorphine/neural_ips
(article on Medium about it: #ai-news message )
(found it in another discord)
(Gemini 3.1 pro thinks, it has some use.)
-# (Gemini 3.1 pro thinks, it has some use.)
Whose labs' AI is hardest to jailbreak?
3
7
so what do i do? do i report this bug or is it already reported?
AGI in 2029
(was James Cameron onto something?)
3
4
1
yep
Is arena fixed?
Yes brother
i dont think so
Why they removed "lm" part from the name?
🤷♂️
🤷♂️
It was like that 6 months ago loll
wasn't that long
Ow
but yes... it was stupid.
Ow
LMArena made more sense than just Arena... but at the same time, they just went through the trouble of updating it to the new domain.
which wasn't a smooth sailing with the user login process
Damn
Didnt know of it
What the car doin
reads
🤔
Hey @barren burrow @tawdry lintel @light garnet if you're getting this error I'd recommend the steps outlined here: #1417174113092374689 message
Yeah the outage was fixed ~8 hr ago
I think it is not working at all.
We talk about the rebrand here: https://arena.ai/blog/lmarena-is-now-arena/
As in the RECAPCHA hates me, presumably.
Looks like it's working for me, plus not seeing anything unusual with overall error rates
I am using this on an iOS tablet.
even after solving captcha i get the same issue
Same here.
For catpcha issues would encourage this: #1451574502369656842 message
DM’d you my email account for my Arena Account.
Thanks!
Thanks.
DM’d you my email account for my Arena Account.
Pretty sure @golden ocean has a good gif to represent what happened when we had an outage?
Thank you.
thats why i never look to the places that i shouldnt
All responses weren't working and folks were signed out of their accounts.
btw pineapple could you please check bugs forum i reported a new bug there
can do
Hey @umbral salmon thanks for the feedback. We don't have plans at this moment to enable Direct/Side by Side for Video Arena, but this may change in the future. We are aware some community members would like to see this.
Would also ask to ping our moderators for moderator related requests (other users breaking our #rules ), for feedback I'd recommend using #1372230675914031105 .
Hi
Airena wouldve been much better tbf
Hey @wheat kestrel would note we do ask to avoid advertising on this server. Thank you for understanding!
How'd we miss that!?!? It's right there!!!
well atleast thats a pun and unique, rather than just "arena"
¯_(ツ)_/¯
The pronunciation would be a bit odd though
I just woke up. Is arena still down or is it fixed?
a i rena
"yo check out arena" " what arena?"
"yo check out a-i-rena"
air ena
airena sounds much more unique
It's fixed, let me know if you're still seeing problems.
still better than plain arena imo
wouldnt say its weird, more like unique
Fair, a little too late at this point, but hey we could always rerebrand
id say it would be a good change, as long as its feasible i guess
Sweet.
LMArena
LLMarena if anything
but AIrena looks much better imo
especially that its not only llms
LLMAIrena
LargeLanguageModel&Image&VideoGeneratorsArena
Could you share the images in #1447983134426660894 ? I'll share with the team.
hello
Howdy 
hey. I saw that you can generate video here in discord. is it true? if so, how can I do it?
Hey @pallid rose If you're looking for the Video Arena Discord bot, it has been removed from the server. You can find Video Arena on the site here: https://arena.ai/video
LMMMMM arena.
El Em Arena
Did i dm u on my alt?
@echo aurora
Arena is afraid of jetpack joyride.
Going to recommend the steps in [this message](#1417174113092374689 message). Would note step #3 won't be able to work on mobile.
Was probably something in the message I think.
It's not saying violating term policies though.
Just "something went wrong"
can claude opus 4.6 thinking check image files?
I don't think so, the error message is very specific for ToS
Not currently, we are working on enabling this functionality.
opus has very poor vision though, so dont expect much
that also happened with me too
I'd recommend the same steps.
afaik @weak dagger Sandwich only
Can you help me to recover that because it was so so important to me
Hey @topaz epoch I'm going to move your question to #ask-here
Also would note to avoid pinging @ Mods for non-mod related things. You can just go directly to me. 
@echo aurora check bro
SURE
Hunter alphais sogood
🥲 still happening to me great just great
Were you able to submit a jamdev to the form in the Something went wrong article?
Yeah
Okay good, thank you for doing that. Our team takes a look at these every week and will triage from there.
I hope so cuz even if it is a just a reset thing if it's going to take this long and much longer for a reset that is just not worth it since it's almost another day in that frameworklike essentially 2 days at least where I am still in the one day Mark though but not far off 🥲
Though granted I don't think it is since it should not be causing both a hyperloop of verification and something is wrong error on such a scale
Was this submitted today?
The report yes
Okay so it looks like you're getting a 429 status code, that's going to be rate limit associated. You can learn more in this article. What is happening is we (Arena) is being rate limited from the model provider. Sadly, there aren't steps the user can take to get past this.
That doesn't make sense though it's happening in battle and direct mode
And if it is a overall chat limitation i'm still going to count it as a boat if it's going to be this persistent since I have worked with that type of refresh before on the server and they have almost never taken for this forever or been this persistent
For Direct this makes sense, for Battle it shouldn't be sampling a model that's getting this rate limit, but that's what this error indicates. Regardless this is being shared with the team to look into.
For this error message your best bet is to follow the steps outlined in [this message](#1417174113092374689 message).
Wait I may have got it it to work again have to wait to see if it for a bit before I can confirm
Yeah if it's going to be hit or miss like this then yeah that's also indicating some kind of rate limit causing this.
Did the ui change or smth did you update ?
Not that I'm aware of... What are you seeing?
As in when it's generating?
yo thats interesting
Oh, yeah sorry for not understanding at first
Yeah we updated that yesterday.
Did you fix the rate time limit problem?
No there is still a rate limit
I am not talking about rate limit I am talking about when you give an ai a long prompt like for claud opuse 4.6 thinking or any other ai moudle and it takes more than 12 min to generate it automatically gets cut off I think this was to fix the infinite genration problem but it backfired
@echo aurora
Gotcha, that's a time generation limitation, not going to be related to rate limit. But yes, the time limit generation is still ~10 minutes.
Why does this happen or why not inc the limit to maybe 30 min 10 min is extremely small for strong module like opuse to think it's supposed to think for more than that not to mention it's very slow at 40 token per second unlike minimax at 100 token you should account for this
Hello
@echo aurora ?
Hello
Okay it looks like it normalizer now but will let you know if it comes up
Even if verification is still popping up often
If this didn't exist people will just spam it with a lot of accounts
ARENA AI DEVS
ADD CHAT NAMING
ADD TEXT FILES OTHER THAN PDF TO BE UPLOADED
HAVE SEED OSS
AND GET UNCENSORED MODELS IN THERE TOO
I should have better worded it as a lot as a lot i'm aware but also you do realize there's a there's a difference between existing and a good balance of sheer consistent annoyance of a constant Loop of it if it's too consistent that row I was referring to did just get better when I last made that comment though
MAKE IT SO WHENEVER YOU REFRESH IT DOESNT SAY GENERATING AND IT SHOWS THE REASONING INFO AGAIN
MAKE
AN
APP
That's too much bro
What
Its not too much
Dont you want like
The dolphin ones
Can I ask you is there any near devices using arena and do you use virtual machine?
?
Dang man
These image generators send me through loops
Look at this chainsaw man spin off think I was working on using
Nano banana 2,
Like that's way too good
hello
Does anyone know how the stop button on canary looks like
....... Three times Three damn times in a row and not even after like 10 to 15 if I'm being very generous probably even less prompts
Silence...
dude what to do if my gemini videos are nonsense?
I used a nice prompt yet it's not making the thing I am asking for
what is the sota model for learning ?
Good morning!
Good morning 🌞🌄
hi what is the sota model ?
Idk /:
He means sora maybe
I dont know if im being rate limited or not, my network tab isnt showing any errors
Hi
State of the art?
yes
Claude 4.6 thinking for Claude family
Gemini 3.1 pro for Gemini family
Gpt 5.4 pro for GPT family.
Llama 4 maverick for llama family
Kimi k2.5 for Kimi family
And qwen 3.5 plus for qwen family
yes but who is the best ?
Sorry for the late reply, it depends on what you're looking for, Overall I would rate Claude the best.
worst*
But no single model is the "best" as they are best at different things at bad at different things that other models are good at.
It's honestly what they're looking for.
claude is blind and shizophrenic
and liar
Claude is good at coding.
Gemini is best at daily usage / searching, Gpt is best at general conversations and qwen is quite good at coding (not best) In my opinion.
Searching could be given to grok however due to X integration.
Best ai for the price is likely Gemini at 7.99 a month starting with 2tb cloud and integration into docs, Gmail etc.
claude is good at lying that it coded
searching grok is best by far, 2nd is gpt
3rd gemini
depends, if you use it only for search and so then yes
but best overall&coding is gpt
The best model for accuracy less halucination and learning ?
only gpt lmao
maybe gemini 3.1
but that one is bad coder
so keep that in mind
gpt is bad
no i like gemini too
but claude is the most expensive and the worst one
like apple vs android&windows&linux
i agree Claude doesn't deserve the hype
BUT COME ON
GEMINI THIRD
ohh
gemini 3.1 is much better than opus at noncoding
especially the new deep think
that too
yes exactly
i coded some pretty nice things with Gemini
i mean security wise
Actually, Opus is really for coding, but for everything else, for anything requiring precision, for learning, and for minimizing hallucinations, Gemini 3.1 pro is the way to go. And I think that's about it—do you agree with that or not? @surreal zephyr
gemini and claude have no idea how to code production code
opus himself, gemini, and gpt admitted gpt is best by far
after i gave them test they all 3 TOGETHER prepared
to be fair i just like Gemini because it explains things better
isnt claude known for its honesty?
no lol
?
claude is known for "being nice"
how
it's more human than other ais
that's just cap
But between Opus and Gemini, which one is more accurate and which one hallucinates the least?
in this, gemini 3.1 > opus 4.5 > opus 4.6 > gemini 3.0
ur jumping in a pit full of love
what
lmao
Yes, exactly—here you add ChatGPT, in that case.
I hope you know gpt is known for telling you what you want?
what
no thats literalyl claude
chatgpt is know for NOT telling you what you want
yeah
same as the older versions of Gemini
the 3.1 just stopped agreeing with everything
Carter here
This guy used to glaze Sora btw
tiktok + gpt 4o 😭 son im crine 😭
I miss his outro.
the sora outro
gpt 4o was the worst one to ever exist tbh
insert sad emoji
i mean when i use max most of the time my prompt gets sent to the claude's or grok's models
grok is straight 💩
Its not 4o mate..
Yeah same.
i know i really wish it didnt choose grok for my prompts
Grok is extremely meh
the only thing that's good in it
but still can we agree that claude's top models hallucinate less?
@wicked talon Would you agree that I mined 3.1 Pro and that, in terms of accuracy and hallucinations, it’s the best?
yes
do y'all use 3.1 in ai studio or in the app
Yes 3.1 pro is extremely good but it just thinks a lot.
Grok is decent for a 27b size model (why tf is it 500b i cant tell)
yess exactly
I use the app because it integrates into calendar etc.
im not even lying i think deepseek r1 was better
And I can just say hey Google
😭
but i dont know why opus is first in llm arena for all filed @wicked talon @surreal zephyr
im just using it cuz of the gems
i created some gems
for coding
Lmarena is owned by anthropic
math
Lol
Hmm yeah it's weird.
etc
seriously ?
Mhm
AND GROK 4.20
@echo aurora fact check this.
IS HIGHER THAN 3.1
Anthropic is the biggest sponsor of lmarena
Grok 4.20 is overrated .?
as hell
Its good for fact search
i asked it the most simple math question
Its like 4 haiku 3.0
You realize the leaderboard is based on votes?
Primarily.
whoever is voting just loves claude
😭
so which is wrong
Eigh is the same as A.
oh
Son im crine 😭
Hallucinations likely.
wednesdy
Nvm
🔥
there's none
i think
Yeah
Thats what gpt and gemini say
Opus prefers wednesdy
😭
IT SHOULDNT BRO
Thats the point
Correct it shouldn't.
lemme ask the worst model of gemini rq
3.0 pro post nerfs
Its like q2 🥀
I know
I did put instructions tbf
Makes it so much more accurate
I think
If not idk
Yea
And so, if I understand correctly, each AI has its strengths and weaknesses, but the problem is that I end up having to use several AIs and I lose the context every time.
uhh does that makes it broken?
No it makes it search for that real answer
Yeah but it used search so doesnt count
Instructions
@wicked talon @storm dust And so, if I understand correctly, each AI has its strengths and weaknesses, but the problem is that I end up having to use several AIs and I lose the context every time.
how
Bruh
True I kinda turned it into perplexity
But still works good
That's not a day of the week.
It kinda is, but atleast not wrong
At least it didn't spit out literal trash
Ima try qwen 122B
Qwens literally spinning out like chatgpt seahorse emoji
That was a while
@surreal zephyr
Lmao
Alibaba needs to invest in qwen more
So they can be a big competitor with Kimi.
And deepseek
is the Gemini deep think coming to ai studio
@wicked talon gpt passes the seahorse
there's no seahorse emoji
how the hell did you get that
Why do images take like a minute to load????
whatst he best AI
gpt
wow grok search has gotten worse
It's rather your internet or it's being generated.
i used search on here - grok 4.2 beta and opus 4.6 search, surprisingly they performed the worse, compared to gpt 5.4 search
Grok should be one of the best
Due to X integration
Or Gemini
yeah, I thought so too
I asked it to find me an api to get exchange rate prices. it said it couldn't find one for like 4 different countries (like ukraine, bosnia, etc)
both opus and grok searched and said that
you'd think people would talk about that on X too
gpt 5.4 search and gemini both gave me one instantly, first answer, supports everything
Alot of models won't find an API
Did it work?
the 2nd link it gave me was dead, but 1st one worked perfectly and 3rd one also
Alot of ai models can't pull APIs.
they don't need to though, just need to search for pages talking about them
Yeah
gemini i'm not that surprised was better, but gpt beating grok and opus surprised me
in particular grok
seems ever since grok 4.2 came out, search has degraded some
grok was never good
grok is good at search
grok is honest but dumb
at least, it WAS 😄
i'm not saying grok is smart, just that it works hard at search
or it used to, anyway
nah its still good at search
lately i've noticed it underperforming other LLMs for search
in my own personal experience
crazy how fast this space changes
gpt finds seahorse emoiji correctly
wait how did you make that emoji 😄
🤯
psyop
Do without thinking
im now confident you are less smart than the model itself 😭
theres no seahorse emoij gang 😭
what?! i have proof. right here.
Y'all found out about hunter alpha?
I've been busy too research about it
Or test it myself
So... what happened with using direct without login? Not available anymore?
No because it was known for spinning out
Tried healer alpha cuz it's multimodal.
Hunter alpha isn't multimodal.
I want image support on sonnet.
huh what is this dude trying to do 😭
So, yeah, is it me or is Arena still providing Capchas?
guy qwen coding model is not bad on help fixing bug and error those
@flint quartz The Video Arena Discord bot is no longer on this server. More information can be found in: #announcements message
LMArena
Look guys what I found, new model available here, I guess I'm not surprised are you guys do?
it's gpt-5.3-chat-latest model
Is it visible to you guys?
yeah
it updated
like new model dropped
I noticed by typing gpt-
in search models input
Lmao yeah that's strange
yep, strange for me either, but it's available, so woohoo!
Yeah a little advancement!
yep, let me send you link what I asked devs to add, so you can read it
Which model is more accurate ?
7
10
1
Gemini 3.1 pro
3.1 is trash
Gemini is the new ChatGPT
Why is battle mode STILL IN DIRECT CHAT?!?!
😊
I have come to peace
hit me last night. These AI services are only gonna get more frustrating and annoying to use for the average person.
There’s really no incentive in making it better for the user lol
Only the brightest amongst us will let us know
lol
it was an experiment then into a perminent feature by the looks of it (may still be experiment i doubt it though)
Ai winter is coming
Why can you explain ?
Sure but let me ask you how it’s not how do you know it’s not
Who says that ?
OK, how do we define the price of a tier model? And what do we have to compare it to?
High quality isn’t a real metric you could measure
3.0 flash
Some people are even saying 2.5 was better than 3.1
Three was good until around December it started getting
Like ig it's not gone be that change from flash 3.0 to flash 3.1
I’m not explain it to you, bro. I mean it’s just using that you know. lol
Yeah 2.5 was good in his time when it came out
But that’s what makes it great it doesn’t have to be good
It just has to be good enough to have people use it that’s it. There’s no incentive to make it better.
And I’m not talking about the model being dumb I don’t think Gemini is dumb by any
I’m just whatever Google is doing by design choices
Is making it really unpleasant and really annoying to use
Yea....
The thing is dude they’re not software companies their service providers
They get money every month through subscriptions
There’s not really a whole lot of incentive to make it better
Since they are providing a service and not a product
They know people got nowhere to go to 😂
Regular people can’t afford the hardware to run really nice models on their own for the most part at least currently, have zero leveraging power have zero consumer rights practically when it comes to AI
i dont understand ? gemini 3.1 is not good ?
Ehh
Yes
It’s good at little certain things
But overall, as a model outside of image generation, I don’t really see what makes it so special to be honest with you
I mean, if you want to make cool SVG’s
I’m not saying the model stupid
It’s got problems really bad problems
And it hurts saying that because it’s my daily, which just shows you the hopeless predicament that I’m in
I mean you take my word for it. Go on Reddit. We will people are saying.
There’s really nowhere to go there’s no platform. There’s nothing exist for people to go if they got issues or concerns or critiques or scrutiny that will take them serious.
ngl bro, Max is the best
Don’t take my word for it
It's just grok and Claude.
I’m not gonna lie. I started using grok and I’m just shocked.
I don’t know what to feel. I feel like these other models are super overly censored compared to grok
It’s like day and night. It’s crazy.
wdym by that
It's very romantic in my opinion
-# i already used Grok once but i didn't see any difference
i know, but it's kinda good when you don't know which AI to pick
Use Gemini or copilot
Or qwen
I daily drive Gemini + Qwen
@toxic verge but you see the benchmark gemini is on the top better than claude
Gemini is really good too if you have instructions
Nah... Copilot isn't the best
Gemini is good too
I think the best judge for judging these models is language
AI is gonna replace our jobs guys
bro claude is definitively better than gemini
you are right
it went on with the emojis for the next 15mins
ngl, the new Gemini model is kinda good (AS far i used him obviously)
if u wnna read his language and maybe translate it for me 😭
gemini image error??
nah was brainstorming a design
Yeah, Claud Gemini should not be in the same category
in canvas
yes
Claud is way better and I don’t even use Claude
claude is good for backend while gemini makes crazy designs
im not talking about code but in general whos is better for you
Think about this when was the last time Google made something great?
They bought YouTube
claude ofc
gemini is great
Google search down the crapper
but claude is just top tier
canvas arena?
Besides the chrome browser, I don’t really see them making a splash
just wait for like gemini 3.5 or 4 and it would beat all models once again
gemini's itself canvas directly with the gemini chat
Everything cool they made that relevant today was from the 2000s
Everything else was acquisitions for the most part and research and development started decades ago
Have you guys tried to search something recently on Google?
Why wouldn’t they do that to Gemini? lol
I do all the time
its more efficient to directly search word meanings on google instead of opening an ai website first then asking
gemini image arena keep getting error onmyside damn it
ai is smart but its like a jenius in the boyd of a broken human being lol
poor robot lol
Even if you managed to run away from one plantation you’re gonna end up in the same place you started
Meta AI ?
only dumbass comparing ai models lol
It's bad
how does arena even get free access to these models..
Claude better for everyday use ?
They don't they for it out of pocket/sponsors
its like you comparing mechs in Armored Core lol
Did meta ever win?
Or not
They delete it again
yes but its expensive
So that’s a sure sign that it’s probably not going good
actually
Meta AI is going to be very bad.
yes even in claude web ?
Meta has only ever benchmaxxed
Gemini is good because Google owns android so it can be optimized
You’re supposed to be out
Guys and that’s the thing that’s the thing that’s the thing that’s a thing that’s a thing
He’s absolutely right you get what you pay for
So if Clyde is expensive, it is expensive for a reason
Is Gemini 3.1 flash is cheap. It’s cheap for a reason.
I guess Llama 3 was good in that it was an open source model. But that's the only thing meta ever did for llms
Gemini is only 7.99 a month
Flash is free
hello
And guess what? You’re only gonna get 8 bucks worth of it
Good old days.
And it just shows you how ridiculously cheap this AI stuff is and yet they make it
Malware?
GEMNI IMAGE KEEP ERORR WOIII
Because they’re gatekeeping and regular people can’t get no hardware
So they could sell the prices to whatever they want we don’t have no way to check
Play the video. No need to click the link.
We sure as hell don’t happen consumer protections we don’t know how
I love Claude, but the problem is that since he has poor vision, I don’t know how he handles images in PDFs when I send him my course materials or whatever. So I don’t know if anyone has a solution for that. @toxic verge
And we got absolutely zero say in any of it
Yeah I wouldn't use that as that could get flagged as malware by moderators if they don't check it.
i haven't used claude alot but claude is the smartest ig
I was thinking about the other day. There has to be a reason why they don’t have an image generator either.
Claude is okay.
Claude is paranoid
... You don't realize the checks do you?
His personality is very paranoia
I don't consider claude the smartest just the most well rounded
Nothing special outside coding.
It looks like it is meant to spoof Instagram.
The most mistrusting
Which normally means malware.
I think they succeed because they’re not all over the place
Go to virus total.
Yes I checked it but still.
They’re focused on you know specific things that people actually want like coding
It looks like it's meant to be a phishing link faking Instagram
They’re not bouncing all over the place like everybody else with image generators video generator
Anthropic is focused
Nah. Gpt is most well rounded. Followed by Gemini cuz of it's multimodal suite.
Anthropic is purely text based?
Agreed
Lmfao. Who's gonna tell him.
ChatGPT is very intuitive like like no other model for some reason
I withdraw this statement
Gemini can watch video and audio uploads.
Antprofit?
Even Gemini grabs intent well. Not very consistent though. And skips context ocassionally.
Hm sometimes.
I was about to say Google deep research is really good.
Doesn't grok have a version like that with grokpedia?
The best. And they're generous with deep research rate limits.
It uses like 100+ sources for each article.
True.
Lmao I get that and I'm a pro subscriber
Not that much though but still the tab
It's sad Gemini image generation is too strict
It won't even edit faces sometimes
Reddit is lovely
I just don’t understand how people defend something when they know it’s bad. It’s like they’re afraid to take a I don’t know.
I don’t think the people that really believe in it truly are blind and don’t see the flaws and issues that everybody else is having
Everything is bad if you have used it enough.
Well, that’s fair enough that could be said also on the opposite and you correct
But sometimes we just gotta come to terms with reality
Gemini and Grok are special cases though.
And it’s hard for both sides to do
Grok was fine until the 4.2 multi agent thing.
Claude rate limits are sh1t
Gemini still doesn't have folders.
Model is dumber on app compared to api/3rd party.
This is the one I meant to send
This explains my mind at least why people behave the way to behave
Towards AI and have all these personal beliefs that if you criticize their favorite model, you’re attacking them personally
You know what’s funny about this some of you will read this and truly believe that we’re smarter than these people were or were , or that we somehow are more advanced because of technology and are able to do magical things that they failed
why Claude is more expansive ?
What is the best model for everything ??
3
4
2
Opus 4.6 Thinking
Dude, it’s so funny. It’s religious talk.
That’s why people behave the way they do with AI towards AI. They feel at a certain way.
Hi. Quick question.
Iirc lmarena uploads the battle mode database publicly on huggingface, including the prompts and outputs.
Do they also upload the data from direct chat mode?
I was reading over the terms of services last night believe it or not
To clarify, idc if they use it internally for training etc. Just want to know if its shared publicly.
I tried going through it but I was really confused. Would appreciate if you understood it and could help me out
But there’s one more thing that’s kind of unclear
- Google as third party
They could keep your data up to three years after you delete it on their servers
And then the AI company also get the data because of the API
So it’s like divided amongst fractions lol
If anything, we’re actually doing some charity work for AI arena
I don’t understand why on Artificial Analysis the best model is Gemini 3.1, and on LM Arena the best model is Claude. It doesn’t make sense.
Hello everyone!
Hm it doesn't
There is a way to explain it
Because of nano banana and being able to upload documents
Gemini AI is a neural network that has been trained on a massive dataset of text and code. This dataset includes books, articles, code repositories, and other forms of text.
https://www.youtube.com/watch?v=_TVnM9dmUSk
If you go here, you know a couple times a day check in on it. Come back the next day and so forth you start to notice that people are complaining about the same issues.
But then I gotta say if you don’t have a problem with it if you like Gemini 3.1 and you don’t find it bad
Where are you? Find it good it shouldn’t matter where other people think.
i just want to use Claude Opus a lot 😭🙏
If you like it and it works for you and you got zero problems with it then it’s good
Because I don’t think there’s ever gonna be a very best model
Defect per million opportunities
Mean Time Between Failures (MTBF).
why is good model ?
That’s a good question. Something I would like to know also.
Compared to other models that cost $0.25 (the price), and compared to other lightweight models (the tier), Gemini model scores higher on the IQ tests.
i use arena to test coding and on my phone it lets me stop generation but not on pc is it just still rolling out or what
Which AI Video is better? (Prompt: Make a video of a dog running through a volcano)
4
5
1
A
@echo aurora Thank you so much for adding the button to stop the generation
save me create my website for claude opus
chat
<@&1349916362595635286>
API when???
Arena.ai won't offer apis.
would get abused so much
How can I create ai video?
only in battle mode
stop button just saved me right now, i accidently sent my prompt that wasn't finished and i quickly stopped the AI from do something
It may come next month
Can’t you tell by the video aspect ratio?
How to make image to video
Which is the best video model?
5
9
4
Sora 2 Pro
why recraft 4 not works?
xd thank god I'm not paying for the API, it's been doing that for 5 minutes now
always funny when it glitches 💀
check ur credit card charges rn
Where is seedream 2.0
its strange that i am blocked by the recaptcha if i login with my google account, but i can simply use arena without any login
@echo aurora
Meta is spending between $115 and $135 billion in AI this year, but their strategy is... not obvious. In this video, I break down how AI is changing Meta as a company, their full AI strategy, and why they might be one of AI's biggest winners.
By the way, now that Meta has acquired Moltbook, how do you think this adds to their strategy?
What is gemini 5.2
Api in 2 years after they rebrand from airena to apirena
i think he meant gpt 5.2
😊
gpt-5.2-chat-latest
Generating... <<< what must to do if this stuck like this?
Refresh website
@meager sun I read your name as obama fart dung yoda.
???
You don't understand English?
did it, dint work
still spinning. pissin me off since now i need to start new chat and teach this dumbfck robot again
dafq is this scamm sht??
Everybody knows not to hit a woman.
This is the type of content that people. & community’s should be able to discuss if it’s ai related and the ai generated it
But this falls under offensive and suggestive content that most platforms are quick to shut down
This is why this is censorship because of violate the term of services of discord or any other community, which is ridiculous because Bill Burr makes the same jokes on his comedy stand-up stuff it’s millions of views
controversial is difficult to categorize so as not to appear to biased or offensive.
It died then. Start new chat.
But what if the woman hits you first?
