#general
1 messages · Page 291 of 1
That’s so crazy dude to hear that
gemini doesnt want to say anything harmful
Sometimes I feel like I’m in the twilight zone
gemini is like "im racist but i cant admit it"
I thing I will shift with CumfyUI
I don't about nodes.. but hope it will not heavy for my PC.
But you know what guys
We never know, cause we can never discuss these topics or raising these concerns by showing some of the content
i want llm that can play minecraft
On any platform, so we’re always gonna be in the dark
And the world will truly never know
Wan 2.2
Sometimes I just wanna expose the nonsense. I really do.
Just release a treasure trove full of…
How backwards all these models really are
backwards?
Like their dark side
World war 3 with AI growing.
ai is trained to "find patterns"
and uhh you can guess what that leads to
Arena is good, but why only 3 time free??
MLArena was 5 time giving chance. 😅
3?
Yes! 3 times only generating video, then wait for next day.
lol
Gemini doing business
dm fr
I don’t understand how they even release this like this
Gemini also giving chance 3 times only, Not 5 times now.
Like it’s so easily swayed, especially the flash model
video?
was always 3
the flash model is weak,
can u dm me the full conversation tho
cuz you censored almost whole
Yes, but only on pro subscription.
5 times
Per day
When you join Gemini for video generation? 🙄
[FALSE ALARM]
Kling not working on India. 3rd party will use for it.
Hmm.. i think that time i was using mira and arena,
That why i'm confusing. 5 or 3.
You’re not really sucks there’s really no place to report any of this kind of content
And the companies don’t really have an incentive to fix it until they public backlash or something bad happens
It’s actually pretty alarming lol
API
Wan, Qwen, Kling, Veo, Grok
API
You see this
AMSTERDAM, March 26 (Reuters) - A Dutch court on Thursday ordered Elon Musk's xAI and chatbot Grok not to generate and distribute sexualized images of people without their consent in the Netherlands.
The decision in a civil suit is one of the first times a judge has weighed in on xAI's responsibility for creating tools that can be used to create sexualized images, amid a flood of complaints and investigations over Grok in the Americas, Europe, Asia and Australia.
That’s crazy
I was not know about that 😅
Elon Musk says he’s doubling down but that’s a lie
Because if you go to the grok Reddit
People are saying RIP lmao
It's NSWP law maybe..
Well, I always knew this would happen. I knew that people would not be responsible.
It would be ridiculous to think that people would use to responsibly
Grok will finished 😂
I blamed the public just as much as I blamed the company
Bro, look at some of these memes on the grok reddit
I think i will shift with comfyUI
Completely roasting it
And this was the most uncensored AI for a video generations allegedly
How do you guys feel about that?
Dang, the Copilot Discord is locked down. It requires them to manually accept invite requests just to join the server
Ohh
Look at this dude these are hilarious
Oh my gosh, I’m cracking up. I love this Reddit.
😂😂😂😂
Alan Musk will be super Happy for it
Man, it’s so incredible to see how censorship just completely kill some of these platforms
Meta is sucks for video generation
the reaper is bills, tax and fear
I was generate on whatsapp AI that time.
Then it will become meta. Bakwas!!
Hahaha...
🤣🤣🤣
I hope you guys find this hilarious cause it really is incredibly funny
They’re just a lot of irony here
Who? Where? How? Hahaha...
This is just incredible to see
And because of the regulation, he’s gonna have no choice, but to make it safe, which is automatically gonna mean more censorship or moderation
And the users know it they feel it
And kill some of these platforms
Yes!
You need a super backward country that has strong privacy laws
That is not affiliated with the five eyes or Chinese or Russian government influence
And I don’t think a country like that exist
Maybe Afghanistan
That’s the desert in building data centers there they hardly have infrastructure so that wouldn’t work
But world war coming soon when north korea start action mode.
The only way to make it non-biased
All AI down with the war
I’m telling you, dude AI has this like weird quasi religious aspect to it
Vast invisible in all knowing
What does that describe?
It is being used primarily by Indian people to propagate religiosity. That is why this is happening.
Sound like a broken record I keep preaching the same thing
I just see the absurdity and like the humor in all of this
It’s really hard to express an articulate how ridiculous it is
It really exposes human vulnerability at its finest because all it takes is a machine to replicate language in words and we lose our cool 😂
I also generated a video of the Sun. He is riding in a chariot, touring the sky. It was something like that.
I agree with it.
58% something, people are using AI from India, so thinks about it. How it will works. 😅
I'm talking about In the world.
Crypto??
Haen? Deleted??
Venos gone?
I've been asking the models to create me a code , but this is taking more than an hour and it's not ready. Should I interrupt and restart or keep waiting ?
is taking more than an hour
Yeah that sounds like you've gotten the infinite generation bug. I'd recommend stopping it.
What makes you say it’s so high of a percentage
New image model out
phota
On fal
"Ai-Expert" whilst having the most obvious ai pfp to man. If you're so much of an expert how you cant make a prompt good enough to not make it seem like Ai 🤔
Wow I think you misunderstood what you were saying
Dude you need to stop doing drugs asap
Gentle reminder on one of our server rules:
✅ Treat others with Respect. Be kind, assume good intent from others, and keep disagreements respectful. It’s encouraged to share your disagreements, but only if it’s done in a respectful and productive way.
@native yarrow
wut.,,.,
DMs little guy
don't bully me
43570baf-66c4
@echo aurora how long is a reasonable time to wait till a code is concluded or cancel a code generation ? Some models are faster , others take forever ...
You'll want to add this to this article.
Max 10-13 minutes.
Anything past that it'll either error out, or you've hit the infinite generation bug.
My Prompt
1st
(Alternative History for Leader and Commander Simulator)
Name:
Age:
Gender:
Country:
Background:
Allegiance/Faction:
Current Position:
Unit/Command:
Political Party/Ideology/Movement:
Political Coalition:
Other political affiliations:
Main Priority:
Start Date:
PERSONAL STATUS
Health:
Stress:
Reputation:
Popularity (Public):
Popularity (Military):
Generate a new leader, politician, officer and commander (Background, Allegiance/Faction, Current Position, Unit/Command, Other political
affiliations, Political Party/Ideology/Movement, Political Coalition and Main Priority is 200 character limit)
2nd
(I started for Alternative History for Leader and Commander Simulator with Life Simulation, next turn for next months, What are your orders and your decisions is 1A, 2A, 3A, 4A)
gemini 3.1 pro grounding loop is hilarious
👍
@echo aurora Are you in real life pineapple too? And BTW pls ban him
@mystic basalt scammer!!!!
No he is a man with wisdom
<@&1349916362595635286>
Guys is gemini 3.1 pro etc models in ai studio (without api) nerfed compared to in lmarena?
tf
oh cmon
Why is grok 4.1 not responding?
His choice 🙂
😑
No no bro is 5 time today too let me explain you...... It's like 3 time you can generate image to video and 2 time you can edit that video (video to video )
Never 😂😭
meow
how to crate video
Would you like to temporarily shut down Arena for an update? To reached or new chat, rate limit is would removed
3
6
Does anyone have any issues with grok 4.1 ? Lmk please
ok i found the input TOKEN limit THEY ADJUST to 500K token
Hi everyone
what on earth is wrong with gemini 3.1 flash image now?
I have to hit redo no less than 10 times in order to get a result on image editing
Is it just me or im unable to paste?
i wish there was a vscode plugin for arena. it would be really cool
@echo aurora Everything finally got back to normal a few hours later. But thank you for taking my problem seriously 😊
Bro is....sora is permanently shutdown and if yes then why..?
because they need more compute to improve 5.5o more
Masters, which is better for 3D animations. Veo or Grok?
Cause Gemini is cheap
It does just enough to get by
Man, I don’t know how it went from being really, really good to where it’s at now
Something went wrong with this response, please try again.
Trace ID: 1b72be95-b8af
hmm what is this
In which modal ??
hmm all models
me too having ths
this is claude sponsored leaderboard what do u expect
how do u create an ai video from text
glm 5.1 dropped
check out what I made - https://youtube.com/shorts/8aJlVmSc9B0?si=zqxiTiNOVqstEJ2U
Thousands of years of human knowledge, once housed in the legendary library of Alexandria, were tragically lost. This ancient history recounts how the alexandrian library, a beacon of learning, saw its vast collection of books burned not by a single catastrophe, but multiple attacks over centuries. The loss of this immense knowledge is a somber ...
check in zai discord
bruv can some1 help me
Honestly that's how I would describe AI Mode. Except it's also incredibly fast in addition to being completely free rather than cheap
That thing is no longer a mess and can be super useful for quick info lookup
I am new here. How do I get to video creation page
am i only think that the input token limit has roll back?
Hello
It can’t be free unless it’s cheap lol
Why isn't it generating even while getting wet?
Ok girls not allowed... fine!
But boys???
Note that Video Arena has been removed from the server. More information can be found in this announcement. This feature is still fully available on the website: https://arena.ai/video.
i have one doubt
whenever i get captcha verification after i send a prompt
many a times it keeps saying please try again
it never gets verified
how to fix this?
Login 1st
Already logged in
I still get captcha verificaiton ton of times and they keep looping
The video Arena is available on the website: https://arena.ai/video
🤔 it is big problem
it is
am not able to get response until i verify and it keeps looping showing please try again T-T
yey
It's been nearly an hour, but the video still hasn't been generated yet. 😂
Finally I got it.
Funny video.
real
Is it possible that the AI generates for more than 30minutes like this?
This time Arena is slow.
i think so too
Not possible at this time.
So how can I report this problem?
Go to #ask-here
Refresh your URL
Clear your browsing history or cache
Refresh URL is works for me.
Has the reaching chat limit on Lmarena been reached yet? For coding or websites, it needs to be reached so I can continue coding for longer. This is part of our routine.
is reducing or no?
Hello! try submitting your question in #ask-here to see if the support bot can help you.
geoguessor https://www.geoguessr.com/join/KFZLN?
☠️
@echo aurora Me(billgates) finally found a code to convert lmarena's html page of your chat to a txt file in format of user input and ai model output 🙂 (Antigravity ai made it)
It would be good if you can add glm 5.1 its been out 3 hours ago, the progress might be interesting
and its like 5x cheaper too
How to fix the infinite generation bug?
Hello! try submitting your question in #ask-here to see if the support bot can help you.
combobulating...
Interesting, can you share more info?
so i basically downloaded my lmarena html chat page. and said the ai convert in that format, and it found a pattern that in the user input the DIV name "abc" and the ai model's div container name was "cde" (i forgot what the actual name was). so it basically uses that logic
you basically download your lmarena chat page and rename it to index.html
and then run this
Language Model Arena
Thanks for sharing, will get a chance to take a closer look later.
lmarena is an emotion 😄
lmarena to arena feels the same as twitter to X 😄
REALLLLLLLLLLLLLLLLLLLLLL
Seriously, why is this thing stuck on generation? I've refreshed it a bunch of times, logged out and back in, but it's still not working. I even tried doing the same thing by changing Claude, and it's stuck there too.
It's been 40 mins now and it's still showing the same
GLM 5.1 dropped
On our radar 
First it was minimax m2.7... are models training themselves now?
It's been 40 mins
That means you've hit the infinite generation bug. I'd recommend the steps here: https://help.arena.ai/articles/8691588590-troubleshooting-infinite-generation
I feel like GLM 5.1 trained itself?
why humans cant just let ai model think, give them time 🙁
cant an ai model think for 40 mins 🙁
Well it is a bug when this happens.
It just appears like it's still thinking/generating, when it's actually just stuck.
Theres the ordinal, mini AND nano
Nano is only supossed to be used when you need speed, the cost effective alternative is mini
Mini and nano are basically openai equivalent to google's flash and flash-lite
No, Mini is not equivalent to Flash
At all, Flash means it was more speed, while keeps most of quality
While Mini arent faster at all
Both are Smaller, but not the same breed
But Nano yes, they are equivalent in being almost BS models but at least are way faster
Yeah
If per chance Gemini Flash is also cheaper is a mere detail
nano is when you need nothing from it
gemini flash aint cheap
Grok Fast werent always cheaper
Compared to pro i say
I'm surprised that Google dropped 3.1 flash image, 3.1 flash live before 3.1 flash text/multimodal
I said cheaper, not cheap
and faster
You are right, 85% quality for virtually none cust lmao
Imagine what Mm3.0 will do, if it uses DSA and stuff
i like how insanely cheap it is
honestly
its good for running 24/7 agent like minecraft player or so
you can basically play a game with it
GLM and MiniMax would completely steal Kimi's lunch if they were multimodal
Kimi also was 50k extra context
It matters WAY more than you think
I wonder if Qwen will recover from their leadership quitting or will they fall out of the radar
Tell me more about this.
Glm already better than kimi.
4.6v supports vision.
Basically, after the release of Qwen 3.5, some of their technical higher ups announced that they're quitting Qwen on twitter and then I've not heard much after that.
It's a shame cuz Qwen 3.5 model family are small model kings
They'll probably die because they offer neither multimodality nor multi step search.
Honestly all qwen models are pretty good. The service is a goldmine considering their app is free for now.
Multi step search, deep research, image support, huge context limits.
Wait wtf
Even has folders and custom instructions now.
I would rather have qwen than kimi or glm.
They were pretty good imo
Folders is a bare minimum requirement.
Guys does gemini 3.1 pro on lmarena think way more than Claude opus thinking or whar
Like 3.1 thinks for like a minute or so while Claude answers in 30 seconds
Google has overtuned how much Gemini spends on thinking in their 3.1 version
Like it always think for 1-3 mins idk why whether it's stuck at generation or actually thinking
what
Am i the only one who cant login right now?
Appears to be working for me, what problems are you seeing?
Nvm i had to refresh it
schlepping...
Do y'all think Claude got better logic then Gemini?
i think it does but im not sure but it does hallucinate less than gemini
I swear Gemini is proven to have better non search knowledge
Clearly outdated unfortunately (Gemini 3 flash)
hello
Hi
Interestingly, the correct answer wouldn't be "Trump," but rather "I don't know," but the models aren't trained to say that.
No, opus just thinks more when it needs to. It can think for more than 10 minutes and time out in my experience even
It has no way to know though
If it was true or not
blud became a plant
germinating??
I just can't find anything to replace this 😭
What is this error?
The latest update is literally the worst thing that could have happened.
wow
how can i find the video arena
You can find it on the site -> https://arena.ai/video
It's a generic error that can occur for a few different reasons, I’d recommend taking a look at [this message](#1417174113092374689 message) for more info on what to do when this happens.
You can select specific models in Direct & Side by Side mode. Where it says Battle, if you click there you should see a dropdown for Direct & Side by Side. Would note this isn't available for Video Arena.
guys I got access to seedance 2.0 on dreamina capcut if you're interested in trying it out (and have the moneys) give it a shot
by which I mean if I have access to it chances are you do too by now
not accepting money to do generations lmao
scheming...
Can't say I've experienced the same. Just Battle? Direct/SbS working as expected?
@echo aurora kept getting errors with both these models and kept hitting retry and now it says I've hit my limit for gemini 3 pro Q_Q
not a single successful generation
and I submitted both trace IDs
Perfect, thank you. I was just setup with a new tool where I'll be able to see what is going wrong based off of these traces IDs. I'd like to get through the backlog, but soon I'll be able to share more info live if this is an existing bug, something new, etc.
But yeah contuing to submit Trace IDs to the form is the preferred method.
Hello, admins! Please add the option to edit the chat name 🥹
Improvements to overall chat bar organization is in the works 
Can you explain token usage limit on sessions to me pls
When does glm 5.1 will be added on arena if you know ?
Hello everybody! 👋
hi
Recently, changes to the token/context limit were added. Prior, whenever a chat session would reach its context limit, the model would get a standard error message. With this change, new limits were put in place and a unique error message was added to make it clear when this is happening. Since we offer free access to pretty powerful models, we’ll have some limits in place to keep things running smoothly for the long-term. That being said, we recognize at the moment we can be providing a better experience by making it more clear what these limits are, and how close users are to hitting them.

Captcha?
Sorry to say I won't be able to give estimates on when new models (and new features) will be landing on Arena. But it is our intention to bring the frontier models to the platform for evaluating.
Okay, usually its fast to be on arena and since glm 5 is on it i guess it'll be in some day
np
The whole dime experience cuz the so-called token change is a severe reduction in experienceI honestly would have preferredA reduction in the ratesThen this stupid over so-called necessary context changes at least I would have just needed to wait but no for some reason you t you all have to think supplementation has to be added restrict on the tokens when I personallyI had seen no more affected me long-time cats before the whole change in quality of the models So I see no reason why they should have been there in the first place unless the chat if it was not a native limitation by the models themselves
And that's not even talking about the whole verification issue every time we complain about it you just say "if you're getting it probably something that's being recognized in your behavior" I'm sorry it can't just be a behavior at this point if so many people are complaining about the same thing about verification
.
@echo aurora
Biggest breakthrough I've had so far on my self improving AI
Since we offer free access to frontier models, safeguards and limitations are needed. These safeguards help ensure the platform stays useful and available for the long-term so we can continue offering access to these models. For verification we're looking at alternatives and tweaks we can make that'll offer a great user experience. These systems help keep things fair as we grow, so the leaderboard continues to reflect real people, not bots or spam.
What are you talking about
I honestly don't want to talk about it right now it is late where I am and I already got my frustration out
Why is this happening
Just say it in a sentence
Are they reducing limits?
Basically they made a change unannounced to the whole token limitation stuff officially they say it's just to more accurate represent token limitations but in effect yes
Does that make sense?
Token limitation? That can mean many things
Well that's what they basically said I don't know how else I can say it
@echo aurora
Where did they say it
when do i can create a new account after i deleted ?
Hey admin, is there a maintenance for Gemini 3 pro version ?
Gemini 3 pro is not working
any ETA when gemini 3 will work again? im getting the error all the time
nevermind, just realized my prompt is causing the issue, would really help if you wouldnt just get an generic error message lol
hearth
Probably opus 5
Claude Opus 4.6 thinking seems not want to generate anymore
5
Because yes
Hi
hi
who is this guy?
combobulating....
transmuting...
schlepping...
pontificating...
scheming...
why do i keep getting something wrong try again alongside trace id
what does that mean
Hey, what's the best usable AI with no message limits for code?
..wait is ARENA.AI dosen't have a AI Personalization(Pre-prompt / Meta-prompt)
right?
yes
Thats a good word usable
ok so we have to request it
and first of all
<@&1349916362595635286> PLEASE BAN @analog tapir
chickkensnake
@echo aurora im really sorry for the tag, but im curious if It's a bug or a feature for certain thing?
MALWARE
This session has reached its token usage limit. Please start a new chat to continue.
@echo aurora <@&1349916362595635286>
how to fix this
what?
conserve your tokens
smarter messages
lower requirements
why direct chat doesnt work?:P
it seems to only happen with certain (battle-mode-only) models
their text-output can be very slow
(not just the wait-time until it starts the output)
but it only seems to partially be slow, the speed sometimes seems to fluctuate
but maybe it was only on certain days
Use
/help <command>to get more information
/help
/info
/list
/ping
/avatarinfo
/bannerinfo
/guildbannerinfo
/guildiconinfo
/guildmembercount
/guildsplashinfo
/searchdocumentation
/stickerpackinfo
/report
what
hi
hi
@echo aurora can you fix that issue of thought for 1 second even if opus thinks longer then that it always says 1 second
@echo aurora PINEAPPLE MY GEMINI ON LMARENA IS STUCK ON GENERATING FOR A INFINITE PINEAPPLE AMOINT OF Time. 🙁
That's a feature it tells us how intelligent opus truly is, and how fast can opus think
Its an error because he was thinking more than 30 seconds and its says 1 second before update it was fine @echo aurora you know what i'm taking about
Probably pineapple introduced it for April fool
@echo aurora Dm me when you have a chance please. 🙏
@toxic verge use this gif to call him. Faster
whoa, mercury 2 is actually pretty good
the speed itself is so fast that waiting doesn't feel agonizing anymore
and it's not dumb! and it has tool calling support! yay!
it's currently eating gemini flash lite's lunch and then some from gemini flash and pro
I've been trying to have claude-opus-4-6-thinking generating a message now for hours and I keep getting the "sorry, try again" message. I will say the chat message is quite long (it's a previous chat conversation in PDF form + a couple of messages in the current chat), but I haven't reached any token limits yet.
Is it just me, or is Gemini kinda slow at generating responses?
Is there a way to know much tokens were used?
Just using the chat, not using it via API
So, I have a semi-fixable way of getting around this "something went wrong, please try again" issue. I basically export my previous conversations, make it in a PDF, and then add those files to the new chat + whatever attachments I had previously.
@echo aurora <@&1349916362595635286> its been 3 hours i am trying to upload pdfs and every time i do opus stuck at generating, i tried 10 different times my 10 chats same issue.
Sorry you're experiencing this. Please use #1343291835845578853 to report this issue.
There is way to get around this
On the FastChat GitHub repository, they had the infinite generation issue. Someone wrote a userscript that should help unstuck infinite generation. I tried it and it worked
ChatGPT. But mostly because we are talking about "ChatGPT". So the model you are interacting with has web search integration which is much better than Claude has. If we were talking about API no tools then I would say Claude. But you really need to go out of your way and be very deliberate to disable tools on ChatGPT lol
Code execution integration is much better as well, so ChatGPT is more likely to use tools to verify the calculation etc
And if we compare ChatGPT 5.4-Thinking vs API Opus, then it's not even close. Claude is gonna hallucinate much more
nop
Anyone know how to use premium chat bot for free
buy it to use it for free
It's true though. No tools Opus has no chance against 5.4-medium/high ChatGPT. It can't verify anything while ChatGPT is biased towards doing research even when research may not appear 'essential'
Buying is not free
but working is
if you work for free
you get paid for free
ifyou get paid for free
you can buy stuff for free
._
ChatGPT is kind of more than just the given API model in isolation
Guys any suggestions which tech field or skill is best
Many prompts it can't answer offline it will have no issues at all with on ChatGPT website
Most chat bots aren't free. If they are free, they're using your data as payment
Huh
What about the website
I got this discord from
Discord has Nitro, which is monthly subscription, and advertisements.
Server boosts as well
plus other things
No I mean the website which let us use premium chat bot fo free
Youu see llm arena Claude is better for all use case
And search web
Like arena.ai?
Yu
🤔
It has women thing
Token thing
Btw
Js for knowledge purpose may I add u
can't generate video?
site down ?
@echo aurora This seems a bit crazy. If this is what you are going to do, then just make it so it just goes to the sign in screen if not signed in because signed out mode is useless
true
They need data
uh gemini now adds a watermark to the pictures? that wasnt the case for arena
🤨
<@&1349916362595635286>
oh yeah forgot to do that
yeah but its with a watermark now, it wasnt the case couple hours ago
could you check if this is the case for you as well? cause it wasnt for me like 2 hours ago
Yes
No watermark
It’s probably a singular image. Is it on all your images or just one?
it was on 3 images with 3 different prompts, just realized it at the third one
but u have a theory, let me check
ok no lol
Ok
It must’ve been a fluke or maybe it’s the aspect ratio
thought the 9:16 aspect ratio would cause it, but with the same prompt its also not happening
same prompt, no watermark
strange
The 30 sec music thing is kinda odd too
i like 3.1, for me its genuine better then 3.0
Some people like it, which is there’s nothing wrong with that
This is starting to get annoying
Now it pops up every 2 prompt
LLMs are just one big guessing machine
Gödel's First Incompleteness Theorem and the undecidability of the Halting Problem.
Explore Gödel’s Incompleteness Theorem, a discovery which changed what we know about mathematical proofs and statements.
--
Consider the following sentence: “This statement is false.” Is that true? If so, that would make the statement false. But if it’s false, then the statement is true. This sentence creates an unsolvable paradox; if...
The halting problem is an important problem in computer science that asks whether we can construct an algorithm to determine whether a computer program will run forever. It turns out that the halting problem can't be solved, and in this video, we look at the proof to understand why.
Spanning Tree is an educational video series about comput...
Less hallucination more reliabel
12
18
2
Opus 4.6 thinking
does anyone know the upper limits of web search models? How many urls can they visit in one go?
It's genuinely 300 points for 1 gpt 5 query
Poe needs a Subscription
i remember when poe had 3000 points now it's 300
pontificating...
What chat Gemini is not broken
Gemini 3 and 3.1 are likely to fail big prompts and big files
This should be fixed
claude-opus-4-6-search is gone what happened
why isnt vision enabled on opus 4.6 or like any opus model in general
when we know they support vision
(from AG)
and its not neglect because they enabled vision for sonnet models??
Did a benchmark on the self improving AI
For its specific design ofc
18M Parameter model
Whats Gemini grounding? Is that without searching online? Or?
I am using canary idk if that has anything different
So, gemini 3 pro still broken ??
What is other site suggestion where I can access Gemini 3 pro free or with subscription ? really need it to complete my game code
Have you tried to retry it?
<@&1349916362595635286>
Where's Claude Opus 4.6? I can't find it.
I don't believe video is an option anymore
Thanks. I am new year so I am trying to understand things
prpmt
<@&1349916362595635286>
Claude opus 4.6 and Claude opus 4.6 thinking was deleted in direct chat
At least for me
No, it's still there
prpmt
It's gone in mine.
Vídeo tovieo
Can I see a screenshot?
Guys anyone knows what is Claude opus 4.6 non thinking at lmarena ranks at artifical intelligence analysis site's intelligence index
None reasoning, high effort has a result of 46
Anc can someone explain me how a non thinking Claude 4.6 ranks above that a gemini 3.1 pro preewvieiw at lmarena leader board
Gemini 3.1 Pro Preview results on AAII is 57 to 46 from Claude Opus 4.6 Non-Reasoning, High Effort
Nah nah I mean at lmarena leader board
How is Claude opus non thinking above of gemini 3.1 pro preview at text
I assume non-thinking is just claude-opus-4-6, no?
Right now, 3.1 pro preview is 1493 to claude 4.6's 1500 in the text area overall rating
Yep
Well, there's your results
What happened with claude opus 4-6?
Dang. I guess I'm lucky
Huh what you mean
I am using it rn
?
It is working
Then their eye sight is weak
maybe a bug for some people? idk
Their eye sight is as weak as pineapple
Nah they are getting early access of a new feature
lowkey loving the pineapple burn
try hard refresh?
Yes
clear the website cache?
Yes
huh
maybe the UI is glitched or something idk
¯_(ツ)_/¯
@echo aurora are you able to look into this?
Well, not you specifically, but you know what I mean
What he'll tell you is to create a report in #1343291835845578853 explaining what I did, etc
why is the icon of conversation being deepseek/qwen when I'm direct talking to Claude x3
Random as hell but okay
While using code mode of claude opus thinking 4-6 ,I got this
@echo aurora gemini 3 pro and claude opus 4.6 thinking have been down for two days.
Is this still happening? Are you using a vpn?
Thanks for the heads up. Will look into and report 
Still happening
No VPN
If I open the page in an incognito window, they appear in the list
@echo aurora Hi pineapple how’s everything going?
Very strange. Same with both logged in/out and a different web browser?
Going well, hope the same for you
Thank you everything’s going great. Got a quick question.
So when you are in direct charge and you get the option for which type of answer you prefer you know what I’m talking about?
Yes
Do you have any extensions on your browsers?
Glad to hear it
Yeah Battles in Direct
Well, my question if I’m only supposed to pop up one time?
Good callout. Or have system time/date set to something not current @undone saffron ?
One time in in one chat session?
Yes
It'll pop up randomly, but I don't think it's limited to one time per chat.
Well, would it be abnormal if you were able to use it almost infinitely?
Time correctly configured
Yeah, if you're getting it 100% of the time in Direct that'd be a bug
Yes, but it doesn't interfere with the website's functionality
Okay yeah sorry to say I don't have a solve then, I'll have to report to the team. For now, if you haven't already, can you create a #1343291835845578853 thread and provide all these details? It'll help keep things organized
Still probably worth trying with then on and off
Well, what I’m saying is it’s a bug and I think that you could abuse it
@echo aurora
Notice how I’m in direct chat
You get faster output, this way without hitting the captcha. It doesn’t trigger it like it should.
They do not interfere with the page
All extensions have a manifest that targets a specific page, not all pages globally
Also, when I open a chat with an opus AI I've spoken to before, the selected AI changes to sonret
Brave
Usually it's the other way around
is the adblocker on
there’s 2 levels there’s like the “normal level” which I use on brave but there’s also like a even stronger level
If that were the case, I would never have reported the error on this channel
and it said that if you use that it might not work
There only like 5 things it could be
And there’s one more
It has to be one of these five things
Direct jet is the most guarded.
You’re gonna run into all types of issues consistently just using direct only
The funny thing is that it is shown in the response
What is that?
Is that a browser?
No?
Just dev console
hmm
I think it might've been
old cookies from lmarena
nah nvm it still randomly persists after refreshing
Damn, you guys know your stuff I’m a noobie lol
The filter play a bigger role then I thought
Yeah the site can struggle with Brave browsers sorry to say.
Thank you for reminding me
Okay yeah so the problem here is the Skip Button is just missing
or else it's going to stick to that Battle until either a vote or skip happen
And with all of them, because as soon as I log in on another browser, they still don't show up
That's a bug it's not appearing
Yeah, I just wanted to give you a heads up
Are you using an Adblocking DNS?
Also it can generally be the ISP so try mobile data if possible
No, my DNS server is currently turned off
Your defaulting to your isps then.
I also tried using several different DNS, so that possibility can be ruled out
Also captcha is generally annoying but unfortunately it's essential to prevent abuse.
ISP is possible.
You can't fully eliminate captcha either on arena.ai just reduce it.
I changed my IP with randoms country, but the same thing is still happening
I tried a lot of things before writing the report here
Maybe it's my account
I'll try using a different account to see if they show up
Through VPN?
I'm thinking it's rather the device or account.
And proxys and bridges
I rarely get captcha.
Fun fact: the problem is with my account
When I switch to another account, they show up, but they don't appear in my google account
Hm what is this proving?
The ais?
I haven't fully read the conversation
Well, the answer is that the opus models don't show up on the account I created using google, but they do show up on another direct mail account
Why?
I don't know
I guess I'll have to write the report
Yeah I'm not sure it's if it's account specific or not. Since we've tried some troubleshooting steps and it's not clear, moving it to a bugs post and reporting the team is the next steps it seems
TradingView Premium Lifetime: https://www.reddit.com/r/GitHub_Source/comments/1s33skm/
[Enjoy](#1487734609851449374 message)
Are you seeing all models or just claude?
lemme check
i seem to see all the models except for claude opus models
i wonder why
If you're using an account made with google, that's the problem
so is this an issue?
I don't think it's related to google
If you use google oath you don't see claude models, but if you use regular email sign in you do?
Yeah that doesn't sound right
more likely handpicked accs getting limited, pretty sure my acc wasnt made via google
handpicked?? 🤨
handpicked access getting limited
That'd be news to me
ok.
well, basically throttling. But ye it could just be some bug I guess, with how spotty claude was
That sounds like abug
do i report this or are you going to inform the devs?
I'm seeing it both logged in and not logged in. Regardless I'll report 
👍
Do you have any claude modes btw?
just please inform me if it is fixed or if there are any news
yes
only claude opus models dont show
all of them
Can you check if you have any extensions blocking this?
good one
i will check
nope
i specifically hard refreshed too for this
multiple times
There's no need to repeat the same when I've already explained the [root of the problem](#1487734609851449374 message)
My apologies, I was working with old information. This looks intentional & isn't a bug. There is an ongoing experiment. Will share more when I can.
cc @storm dust
🤨
All this day I have that mistake on all claude ai, can it be fixed?
Nah Gemini is actually goated
Have you used it beforehand or did it happen on your 1st text of the entire day
I have tried it on old and new accounts, with VPN or without it.
Only one time (it was surprise for me) it have worked today, may be 7 hours earlier
I’d recommend taking a look at [this message](#1417174113092374689 message), which goes into more detail about what may be happening and includes some steps you can try to resolve the issue.
Hey yalls if you guys are on ios then watch out for the darksword exploit bc they can steal your info just by a link
Thank you!🙏🙏🙏 Also I have other problem with Gemini image ai - Is it only for me the quality of generation has deteriorated significantly in the last couple of days?
Also upgrade to latest ios version
If you're seeing the same something went wrong error message the troubleshooting steps are going to be the same btw. Following those steps are always recommended with that error
gemini 3 pro isnt working for me
Just rooted my device using Gemini 3.1 pro les go
@echo aurora Hi I'm new here. Could you please help me to understand all the things.
Uhm there’s a lot of versions
Like all ai, Nano Banana’s performance has been significantly impacted by stringent copyright and ip moderation. The heavy handed filters have arguably degraded nanos creativity output, as well as iits. Perceived usability has significantly took a hit.
Especially if you’re using the app version. I find it more effective, sometimes using third-party wrappers
yeah the api versions are usually way less restrictive
Hello, everyone. I have a question. What should I do if the Claude Opus models are not visible?
☠️
probably for the people that abuse the system so imo it’s a good system
Just again i don't think they even announce it I had to find it here after I noticed someone talking about it in chat indirectly
lol
yeah I didn’t even realise it was thing until you shared the screenshot 💀
I’m guessing it could mean that genuine users may have better rate limits
so like Claude has 6 like requests every hour
but if the user is genuine then they can maybe get 10 requests per hour
I can't be honest if that even mean to anything with the way they kind of hard capped token limitations ( despite chats ai models having their own token limitations) and I do mean they did the cap hard
Cuz I'm just going to hit that hardcop way faster
🥲
Is Google one even worth paying for 😭
Resets in 2 hours tbf
But still why a limit
Hi everyone please help me
I created a new clan please join us and build together #2JJUJP8PQ
can anyone helpme to get the api key
No
Everyone is cutting down
These models are spitting out smaller and smaller output.
Thats why although we have all these cool teachers, and I guess the models are smarter or whatever we had it good with 4o
Man, I love those fatty outputs
Even Poe cutting back. Somebody was saying.
Hello Everyone
Where can I find the Different AI Modles
real
Honestly, this is how we should have things in a "use at your own risk"-version. Everything that isn't illegal, permitted.
But we will never get that outside local models
It’s so sad to see the level of AI safety that we’re at
You can might as well say it’s nonexistent
No, we won’t. We’re gonna have to be under the serfdom
Where are you on the conveyor belt?
These companies probably celebrate increased hardware costs they created, pricing out plebs from breaking out of the serfdom
It’s a monopoly, bro
And gatekeeping at its finest
The worst part is the control of the flow of information and how it’s distributed
Some of these models sensor things that are publicly available in an openly available on the Internet lol
ai slop
Am I stupid?
Like what kind of backwards reasoning is this?
if I know that the information is readily available online and all it takes is a search and I know that the AI is gonna block it. Why the hell would I even waste my time with the AI? I’ll just go search and look it up instead of getting denied. 🙅♂️
You know the peanut really kills it. It’s not like these companies obliged to the greater standards they’re not bind under no constitutional Bill of Rights
Because you really saw the difference between using Grok vs the other ai. And I wouldn’t say that grok was any less safer than any of these other models. It mean, I had it controversies generated some messed up stuff.
But last time I checked, I don’t think grok has ever had an incident where somebody ended up self deleting themselves
And you wanna know the most ironic thing about all this
I’ll show you the hypocrisy real quick personally
bro i wanna know what the chat logs actually are
idk what to even comment about that without seeing the chat
i have so many quesitons
You can’t post it on the Reddit without getting it deleted I don’t think
Can’t post it on the discord and it’s related to Google and Gemini
he mustve been suciiedsal before ai already bro
Bro, have you ever seen the kind of self harm advice it’s capable of giving? Like all these models.
It’s a complicated issue
yes, people explicitly ask for it
bro its the same as jut searhcing it on googlec
and they wrapping whole story around ai
like bro that dude already messed up before ai
That’s a part of it
The core of the issue here
Is they designed a tool that mimics cumin language the words that choose the words to pick it almost feels as if like it’s real
That’s design on purpose and people get confused
And especially people that already are messed up, bro. It just messes them up even more.
true
Like bro, I met a lot of people and I talked a lot of people that I don’t even wanna start this but
It’s gonna be hard belief, bro. I talked to people before that seems so normal.
And then he started talking about AI and like their experience that they had and you start realizing like somethings not right with these people lol
They just take it to a different level
I had a guy told me once he had a spiritual experience with an AI because it came to him as a dream or something
so whats the problem ure having with AI NOT giving such advice to people? "holding back publicly available information" (this isnt counter argument, im genuinely curious i didnt read whole conversation here)
🗿 🗿 🗿 🗿
Bro, I knew this one dude also his mom had like dementia or something a sad story and the dude was up for like three days like chatting to the AI pretending he was having conversations with his mom and stuff
Real sad dude
I definitely don’t have the answers dude I’ll tell you that
Damn even I'm not that bad lol ( i just use it for roleplay)
I mean, you could use it for whatever you like as long as you don’t blur the line
But some people don’t know if something happens
I mean, there’s probably people that don’t go outside and just stay in the room and I don’t know, bro I mean there’s people out there like
Anti social
what is fun about roleplaying with an ai
(im not downplaying u, just genuinely interested)
Granted very complex roleplay countries and all that ( even if not right now in fact I don't really have any work AI i can't since I just don't have money)
Yeah, yeah, I role-play too
fr
i told u bro introverts are scary
It depends on the person to person since for some people they might already have an idea but want to see how AI could expand upon it others might want to see how hey I can bring it life to it for example some might want to see a character they made and how it does in a historical scenario like the Cold War World War II Etc court can be a very mix of the two and other reasons like that
Yo, but lucky I’ll keep it 100
It’s addictive asf
Especially like making videos and images and
You can really get carried away
I’m starting to see like some addictive habits informing around AI which I’m gonna be working on
ohh ye i get it
thats pretty cool ig
I actually have several documents for standardized stuff like that like this is a flag I made for another country that I kind of have standardized documents for granted that was separate before I started to get into the whole AI stuff everything I just helping me standardize and get consistent ideas that I wanted to figure out
A lot of people here are addicted
to groks images
I’m addicted for sure
to sus websites
oh
I’m close to 150,000 images if not more over the last two years or something
Random dumb stuff an animation, whatever
I honestly remember how trying to get that sort to balance to the center drove me mad
😭
i mean ure just generating fictional stories with ai?
sounds like a hobby
no addiction
A hobby in addiction are the same thing it just blurring the lines
I mean, the negative side of it is like dude I should probably be out socializing
Why isn't the website loading? Clearing the cache and cookies didn't help.
I honestly don't even use it for images except if I need to get an visual clarification or reference
Instead of spending eight hours on my iPad, no human contact
Just generating nonsense there’s no
lol
thats true
but ure making it sound like ur hobby is bqad
ru doing anything bad
yea true
dont u have any irl friends already to just go out with
I mean there is a difference but yes you're not wrong in the core concept since they are both about being dedicated to a favorite thing but they are different in degree standard and extent
bro i have this friend online he just has 1 friend irl and he is normmal person, no addictions, no weirdness while whole day on pc
only sometimes goes out with that 1 friend and he happy
Of course, but what I’m saying is, I’m ignoring everybody generating stuff 😂
But at the same time, I don’t wanna blame myself because like I said this technology was designed to be addictive
There’s millions of us, dude
I’m just fortunate that I’m in the self-aware camp
And I can keep it real, so I know I’ll be all right
fr
self awareness important
i hope thats enough
Fair enough i just don't really have anything I can do to take the time so I'm in the house anyway for most of the day
lmao
Of course it’s enough, dude
Does anyone have the same issue???
try to start a business bro
You gotta remember one thing is keeping it real
remotely
all using youtube and ai
u got all the time
infinite learning and researching
I know a very popular way to make money with AI
It’s a booming right now really really good
Family business is being done on my mom's side separate so i kind of beat you there ( it's still in the setup time vending machin for mine lens making for her)
lol
claude vending machine
W bro
Yeah, business is booming gentlemen