general

[sterile tartan]

Like u?

[zealous sparrow]

its a flash model so obv its worse than 3 pro

[sterile tartan]

Just kidding

[zealous sparrow]

Here's the thing

we are testing the EARLY versions

the final flash could be worse or better

[whole sundial]

but apparently it is supposed to be as good as gemini 2.5 pro

[torn mantle]

yea

but expected better tbh

[zealous sparrow]

OpenAI already put hazel as a new image model into LMArena

[torn mantle]

lets see the cost first

[zealous sparrow]

its OpenAIs so its prob gpt img 2

[whole sundial]

officially called gpt image 1.5

[sterile tartan]

@torn mantle do you remember we were used to chat in pplx server?

[whole sundial]

there are two editing and two image gen-only checkpoints on lmarena, but more exist on OpenAI's side

[sterile tartan]

The competition is Cutthroat

[zealous sparrow]

Here's the thing with it

It had a style that identified it as gpt img 1

When i saw it i knew what model it is

[whole sundial]

it's good at making studio ghibli edits, reproducing album covers, putting a yellow filter on images, and not much else

[sterile tartan]

@whole sundial is Nova Actually Good?

[whole sundial]

not really except for the unreleased pro model

nova canvas image gen is terrible, editing is even worse, stable diffusion img2img would do better

[zealous sparrow]

hazel-gen-2 looks like a solid model ngl

[sterile tartan]

Oh well it will get better in future

The Benchmarks are kinda good on artificial analysis

[whole sundial]

it's the smaller one, 4 is better

[sterile tartan]

What did u test?

[zealous sparrow]

seahawk and skyhawk models are experiencing deployment failed errors, so we wait for pineapple to fix

[zealous sparrow]

wait let me find it

[sterile tartan]

ok

[quartz light]

unreleased pro model.. you mean stealth-nova? @zealous sparrow 👀

[zealous sparrow]

?

i think you meant silentnova @quartz light

[sterile tartan]

@whole sundial did u test Tensor 1.5 More?

[whole sundial]

raptor

[whole sundial]

not yet, but i will

[zealous sparrow]

there is like a new model in battle textarena called silentnova

and we suspect its amazon

because of nova in the name

[whole sundial]

sounds like google or grok to me

[zealous sparrow]

Why would we need 4 google models

on battlemode

[sterile tartan]

So what's your opinion so far?

[zealous sparrow]

google usually drops one or two stealth models for battle

never 4

[whole sundial]

it's obviously still a chinese model finetune but it's not a bad finetune

[atomic lagoon]

Is movementlabs good? idk who its done by and what its best at

[torn mantle]

wasnt nova amazon?

[sterile tartan]

It's

[torn mantle]

i see

if its from amazon then its not that good

[sterile tartan]

I thought is good because of Benchmarks

But seems like it might be not after all

[zealous sparrow]

im tryna get silentnova to get the info of who made it

[torn mantle]

the issue is that they are training for the benchmarks

not for making a good model

[sterile tartan]

Is a paradoxical situation

[torn mantle]

:/

[sterile tartan]

They can't afford to lose on Benchmarks

[zealous sparrow]

@hardy swallow

[sterile tartan]

The world basically judges by it without experience

[whole sundial]

<@&1349916362595635286>

[main nexus]

Dude every time I ask claud 4.5 opus what model it is it says they don't have a model 🙏

[zealous sparrow]

W mod

pineapple thanks

[main nexus]

https://tenor.com/view/kitten-cute-cat-pet-small-gif-9791329512534104666

[sterile tartan]

When i ask GLM it says im claude

💀

[zealous sparrow]

wait a sec

[main nexus]

Then when I say is it 4.5 opus it says there isn't a 4.5 opus 🙏😭

[zealous sparrow]

nvm

wanted to check if silentnova is an img model

it aint

[main nexus]

https://tenor.com/view/cat-kitty-pussy-cat-cute-kitten-gif-16158951

[zealous sparrow]

lmao sonnet what

[pale rain]

#1397655624103493813 create video of dog

[zealous sparrow]

why is silentnova so rare

Ok, be honest. Anyone got silentnova in textarena yet?

[old garden]

https://tenor.com/view/genius-zeus-hoverers-zeus-genius-hover-einstein-gif-18134355799312387018

is silent nova gpt 6

[zealous sparrow]

bro i cant even get it

its a textarena model

[old garden]

i haven’t used it

[echo aurora]

You'll want to head into the #video-arena-1 channel and type out /video then provide your prompt.

[zealous sparrow]

I still refuse to believe holo-scope is google..

Why would google make 3 models for Battle??

[whole sundial]

baidu? minimax? some other chinese ai company?

[zealous sparrow]

So like

frame-flow used to say its google right

later changed to be ERNIE

this migh be the same case

[sterile tartan]

U can't predict by asking them

The Ai Identity Crisis

[zealous sparrow]

I know that

Their coding results are what reveals them

[hardy lion]

this is fake we have not released a score for gpt-5.2

[austere sundial]

Metter btw

[sterile tartan]

Yup

[hardy lion]

do not spread fake information about us

[weary galleon]

I created it via Nano Banana Pro

[weary galleon]

Relax

[fickle venture]

https://tenor.com/view/cat-thinking-baby-thinking-cat-baby-thinking-cat-baby-baby-philosophical-gif-17428742062241334001

Is that fake guys?

[weary galleon]

Nope

[fickle venture]

Looks real to me

[hardy lion]

ok, sorry if it's a joke that's ok, just I've seen a lot of people spreading fake LMArena scores around recently and it got on my nerves

[fickle venture]

I mean who would believe it

[fickle venture]

No way is it free on api

[weary galleon]

Sure!

[hardy lion]

sadly people do 🤦
Posts like this got a bunch of comments and retweets https://x.com/teslaownersSV/status/1997008559469650261

Tesla Owners Silicon Valley (@teslaownersSV)

🚨 BREAKING: As of December 4, 2025
Grok 4.1 has taken the #1 spot on the LMArena Leaderboard
Highest Elo ever recorded.
Thinking mode dominating. Fast mode already breathing down its neck.
The rest are playing catch-up.
xAI just dropped the mic.

X · Dec 5, 2025, 6:21 PM

[quartz light]

yes

[fickle venture]

Tweet by Official account

[fickle venture]

Yo have you tryed Baidu

Is it that good?

[quartz light]

i know how to check for sure

[quartz light]

not yet but will try

[fickle venture]

I remember I went to their official website and there was no model

[weary galleon]

Elon Musk signed up this account.

[quartz light]

? https://ernie.baidu.com

they revamped ui i hate it

bloated

[fickle venture]

Yeah usually ai companies lies about their model being "top 1"

[weary galleon]

Only xAI lies.

[fickle venture]

I hate when ai use Chinese language on thinking very annoying

Some times they just write it on output

[quartz light]

for me it doesnt

[weary galleon]

Z.AI does for me. And if I use VPN, ChatGPT and Gemini think on the language of the country of my VPN.

[obsidian spire]

I accidentally deleted a chat. Does anyone know how to fix it? Thanks.

[weary galleon]

No way.

[whole sundial]

<@&1349916362595635286>

[obsidian spire]

oh god, thanks

[echo aurora]

Sorry to say there isn't a way to get this chat back once deleted.

[quartz light]

@zealous sparrow

☹️

[sterile tartan]

Grok 4.2?

[zealous sparrow]

we are not deada-

[zealous sparrow]

WAIT

nickname

use the note steg on silentnova

[lusty tinsel]

when will be able to upload files other than images in lmarena for coding etc...? is that in the plans?

[quartz light]

@zealous sparrow

@zealous sparrow

SAME EXACT RESPONSE AS SEAHAWK

[zealous sparrow]

okay jee-

[sterile tartan]

Flash 3 is definitely coming

[zealous sparrow]

ok but like

what can holoscope be

seahawk and skyhawk are flash models

then what is holo-scope??

[compact sleet]

it's a media trick innit? the fact themselves are not quite wrong, grok 4.1 is either 1-3

what do you expect on X anyways

it's full of bots and elon praisers/haters

[sterile tartan]

Perhaps GPT 5.2

Grok 4.2

Possibly GLM 5

[compact sleet]

Grok itself aint a bad model from the pure perspective of a casual user, Iunno about programming

[weary galleon]

There'll be Grok 4.20, not 4.2

[sterile tartan]

It could be a new qwen model too

[hardy lion]

we changed our ranking so that rank column is now without ties, they are not #1, they did not have "Highest Elo ever recorded" and they did not have a score of "98.7/100" it's total nonsense and lies

[sterile tartan]

Why the extra 0

💀

[craggy cloak]

hello does this site contains subscription ?

[zealous sparrow]

no

[weary galleon]

Musk confirmed it several times on Twitter

[compact sleet]

yeah that's media trick, 1 truth 3 lies, 5 embellishment

[sterile tartan]

I know but why a extra 0

[compact sleet]

it is what it is

[sterile tartan]

💀

[weary galleon]

Ask him

[craggy cloak]

so if i have an account do i have unlimitied chat?

[weary galleon]

No

There are plenty of limitations

[sterile tartan]

Kinda but there are rate limits

[weary galleon]

Expacialy for opuses, and nano bananas.

[quartz light]

new lmarena feature

in testin

[craggy cloak]

gotcha so the limits in chating wise is a bit low?

[sterile tartan]

These are very strict others not as much

[sterile tartan]

I mean yes

[weary galleon]

I said the same

[sterile tartan]

Mainly for Opus 4.5 and Nano Banana Pro and such expensive models

[craggy cloak]

ah yes

many thanks guys

[sterile tartan]

https://tenor.com/view/same-different-but-still-gif-18224441

[weary galleon]

All opuses, and all bananas. There are plenty of them.

[sterile tartan]

True but wait they count differently?

I thought they count together

[weary galleon]

No

[quartz light]

@echo aurora this should not override the last used setting if set manually, please tell team to fix this before they push this update ☹️

[sterile tartan]

That's Actually Good

[golden ocean]

grok imagine 5

prompt: "turn his frown upside down to turn it into a smile and make him purple and make his team green"

[burnt sinew]

Is it on lmarena

[torn mantle]

is this a new model or what

idk whats the prompt here, but if its google then we can assume its flash lite

[quartz pike]

yall. how good is ernie 5.0?

is it ass or nah ass?

[celest lark]

Idk if you notice, but does it seem like that gemini nano banana pro on lmarena is down almost every day?

[quartz pike]

meh

[quartz pike]

ooo where do i try this?

[fleet lintel]

Any news on new gpt model?

[stray aspen]

what new gpt model

[stray aspen]

it sucks lol

why are you even asking

[quartz pike]

idk

just wanted to know the opinion of someone who tried it

[fleet lintel]

I expected it to be released by now. Wasn't today the release day?

[cloud zinc]

no its thursday

[fleet lintel]

It's an expensive model and I believe lmarena pays for its use and it's not free from Google.

Lmatena may have a quota on number of queries per day on it.

[hardy lion]

(generated with nano-banana-pro-2k on lmarena.ai)

[slim spire]

What is this new gpt garlic model thats being talked about

[zealous sparrow]

improved gpt 4.5

[slim spire]

ok

[supple vector]

LMARENA API WHENNNNNNNNNNNNNNNNNNN

and free (pls)

[zealous sparrow]

never

LMARENA pays for other apis already

[sweet topaz]

👋

[zealous sparrow]

it will never get an API

[supple vector]

i gotta stick to openrouter then

n

also

[shell crypt]

👋

[supple vector]

anyone know any free unlimted video gen apis

or do i have to selfhost on comfyui

why are there so many bots

[zenith scarab]

👋

[upper locust]

👋

[pale reef]

👋

[supple vector]

👋

why everyone spam ts emoji

[quartz light]

yes

its gemini 3 flash

[half pelican]

👋

[quartz light]

???

[surreal creek]

what was the prompt? is holo-scope another ERNIE then?

[quartz light]

say pineapple if you are real

[quartz light]

say pineapple if you are real

[quartz light]

say pineapple if you are real

@zenith scarab @proud sail @shell crypt

[warm zodiac]

<@&1349916362595635286> the hand wavy guys are bots It seems? at least some of them

[zealous sparrow]

any luck on uh integrated-info textarena yet

[torn herald]

👋

[quartz light]

yup

[quartz light]

LOL 100%

what are they for???

[warm zodiac]

really weird

[quartz light]

they're all bots

[ornate ginkgo]

there are three video arena groups how i know they use which ai model ai tool

[quiet spire]

Fast is fine but accuracy is everything

custom chatbots for support, onboarding and communities
automation agents for daily tasks and workflow operations
RAG search tools for documents, knowledge and internal data
speech-to-text pipelines for meetings, summaries and reports
content automation tools for writing, planning and rewriting
AI integrations with major platforms and APIs
small custom AI tools for everyday use

I am an AI Developer focused on delivering stable, high quality results.
feel free to reach out anytime if you need support on your AI project.

[quartz light]

https://tenor.com/view/kurt-angle-thousand-yard-stare-stare-tiktok-wtf-gif-12206044702998985154

what the #### <@&1349916362595635286>

[stray aspen]

what

[neon idol]

<@&1349916362595635286> autopromotion

[quartz light]

and bots

[stray aspen]

lmao

[neon idol]

Fr

[knotty fable]

Oh look at this, using AI to make a scam video. #video-arena-2 message

[stray aspen]

thats messed up

@echo aurora

[neon idol]

@torn herald imagine not knowing how to use a self bot. Looser 🥀😭🤡

[quartz light]

LOL

[zealous sparrow]

check dms 🙂

[knotty fable]

If anyoine don't read spanish....."you have a transfer of $100,000 at Banco Pichincha and it is completely secure." Typical scam phrasiology.

[zealous sparrow]

If this is actually what they are using it for, then yeah imagine commiting crimes in a public discord

[neon idol]

Fr

[knotty fable]

Well we're all anonynous here so - if the account is spotted, just use another.
Just like that guy who have been using a dozen accounts to generate scenes from the same film for weeks now.

[neon idol]

Ehehehe

U forgot a thing

Ip grabber?

[compact flame]

Guys did chatgpt 5.2 release yet?

[neon idol]

Nop

[knotty fable]

Well I pointed out that multi account guy to Pineapple and the other mod, what I know they did not do nuffin.

[compact flame]

I guess the 9 December is gonna be on 2026 atp

[zealous sparrow]

I think its hard to ban someone from videoarena

[compact flame]

Instead of this December

[knotty fable]

Yeah, when the user is question most likely use both VPN and perhaps other tools. [I'm only on VPN myself, making me appear 800 klicks from my actual location.]

[compact flame]

But anyways I hope 5.2 is gonna be good

Though it seems kinda rushed so idk if it's gonna be better than gemini

[proud bobcat]

no

theyll probably say "oh its delayed because we need to refine it"

or theyll release it last minute

[zealous sparrow]

only certain undertale players will understand who this is
[skyhawk]
https://019b04b0-5fe9-7b27-99b5-74e806887f1e.arena.site

im sharing bc

of cool design and 3d

also there is no victory

you just die

[proud bobcat]

is that undertale bossfight made with ai

no way

[zealous sparrow]

yes way

do you know who dis

tho

walmart on the left
original on the right
admit its identical kind of

[proud bobcat]

it is

in a way

[fickle venture]

https://x.com/i/status/1998470209066160405

Elon Musk (@elonmusk)

Grok 4.20 is coming in ~3 weeks and then Grok 5 in a few months

X · Dec 9, 2025, 7:10 PM

[zealous sparrow]

4.20 will flop

i can tell

grok 5..

If it does good somehow.

[short ermine]

whats up fam

[torn mantle]

grok 4.2 will be sooooooooooo bad

we can already tell

i mean he usually hype slop models but i havent seen him sayin anything like 'IT WILL BE THE BEST MODEL'

[zealous sparrow]

seahawk can have a stroke

i just saw it

[stray aspen]

elon musk yapping again

[proud bobcat]

please dont be ass

please dont be ass

[stray aspen]

only thing xAI is good is at is UI design for ai websites

[zealous sparrow]

skyhawk is my beloved

[fleet lintel]

how are you finding this model? Good or meh?

[queen veldt]

Gpt 5.2 just broke lmarena elo rating leaderboards

And grok 5 imagine is #2 in image editing

And also the gpt image 2 is in top 5

[neon idol]

@queen veldt wth are u talking about?

[queen veldt]

People spreading misinformation in the last few days

X is full of posts

https://grokipedia.com/

This grokipedia is terrible

[knotty fable]

'He' considered wikipedia to be too left leaning [wiki as quite bad but not for that reason.]- so he launched an alternative.

[weary galleon]

Grokipedia is a piece of crap

[torn mantle]

agree

its so bad :3

[neon idol]

Never tried

And idc

:3

[echo aurora]

Hey everyone - sorry I was in a meeting and couldn't respond to the 👋 reports. I've gone ahead and let our mods know to start removing the "hellos/waves" content from the #leaderboards channel, so we'll start to be more active in keeping the discussion in #leaderboards about leaderboards. In terms of scams they're making with Video Bot we also want to be removing this content, but would note this stuff is a bit more difficult to spot (compared to the other stuff we're modding out), but yeah if you do come across content that has bad intentions don't hesitate to ping our mods.

For those that are a little suspicious we won't be moderating them out, unless they start breaking server rules; being disruptive, etc. What this looks like can be a bit subjective, so don't hesitate to let us know if you think that's happening. But overall I don't want to start booting people from the server just because they're sending a 👋 emoji (unless it continues to escalte into something that we find annoying). Hope this helps. Let me know if you have any thoughts or feedback on this. At the end of the day we want to build a server that benefits the community, and you all sharing your thoughts will contribute to that.

[neon idol]

Elon musk stuff = low quality stuff

[viral cedar]

starship = low quality?

[neon idol]

Never listened about self bot?

They were spamming with self bot

I tracked all of their messages and never sent a message different from the wave

[weary galleon]

Agreed

[neon idol]

Also the member doesn't have any role that increase the % of bot user

And also no pfp

And we are at 78% of a possible ai user

Then, do whatever u want. U are the leader here not me :>

[echo aurora]

Yeah I mean these are all things that point to bot, for sure. However, my concern is if I start banning these folks, even if a small percentage is actually a person, I'd hate for them to be collateral damage.

Now if these bots were causing harm, that's a different story.

But at the moment it seems like a lot of 👋 in leaderboards (which we're now going to be actioning).

This can change though, I'm very very much open to changing how we do things here if it's what the community watns to see.

[neon idol]

Try to go on the moderating discord section. U can see the all messages sent from the member. If all messages are the wave that means that is a bot

[weary galleon]

We should talk only about models in leaderboards, right?

[neon idol]

Hmm yeah

[echo aurora]

I'd say anything AI leaderboard related is fine. Doesn't ahve to be LMArena's leaderboards, it can be others.

[weary galleon]

And future models too?

[echo aurora]

Yeah speculation about where models will land is totally cool to discuss.

[echo aurora]

Would note too these server rules can bend a bit depending on the context. If two people are having an in-depth convo about leaderboards, and the convo starts to go into a new direction, for the most part we're not going to step in. However, if two people immediately hop in there and start discussing something unrelated, that'll be treated differently.

A lot of the moderation stuff we do is going to boil down to ✨ it depends ✨ .

[echo aurora]

And yeah I agree that it's very sus looking, but unless it's becoming disruptive I'd rather the mods (and myself) focus on other things before we start prioritizing these kinds of actions. I hope that makes sense.

But again want to reiterate - let me know what you all think. I'm happy to make changes.

[compact sleet]

This is a bit ironic innit?

I just noticed

And the top one was also the same

[neon idol]

Ok bud :>

[weary galleon]

I edited it with AI for better image quality.

[neon idol]

???

It come out???

Oh

[compact sleet]

Why would you do that?

[weary galleon]

2 minutes ago

[neon idol]

Naaaaah

[burnt sinew]

No its faken

[neon idol]

Lemme cry

[weary galleon]

It doesn't matter, I wanna bring a joy to good mood to people.

[compact sleet]

There is no joy in this, fake things are fake... But you do you I guess?

[misty vault]

Don't tell this guy about The LMArena-discord gork incident

[weary galleon]

Tell me that

[hollow flicker]

why does LMArena ai have rate limits?

[viral cedar]

because rate limits are to prevent abusing the ai testing?

think thats common sense

[hollow flicker]

they are to much 😭

the rate limits are insanly high

[viral cedar]

high or low?

[hollow flicker]

low

idk

honestly

[viral cedar]

your testing ai not using it 😐

[hollow flicker]

uhhh

oppsie

i kinda use it so i dont have to pay for claude

[viral cedar]

kiro

[hollow flicker]

i send 5 messages to claude and

Rate limit reached 😭

[weary galleon]

Opus or sonnet?

[hollow flicker]

opus

[weary galleon]

There are two opuses with thinking and without, you can use both.

[whole sundial]

<@&1349916362595635286>

<@&1349916362595635286>

[queen veldt]

https://youtube.com/shorts/oJbOklMAFEA?si=3xrHQdrUloCTeH4X

Anyone heard of this?

First time hearing

[cloud zinc]

stop with ad

[queen veldt]

It's not an ad 💀

It's spintronic memory

[cloud zinc]

what do u know about spintronic memory?

[queen veldt]

But it's expensive and impossible for regular humans to achieve

[cloud zinc]

sounds like scam

[queen veldt]

I've just read about it

[cloud zinc]

https://www.wired.com/2004/04/spinning-out-faster-better-chips

its been since 2004

[queen veldt]

It's super expensive tho so i don't see it in near future

But we got the ibm quantum computers

You can get 10 mins of ibm quantum computer for free 😂

Datacenters aren't the future

They must find some other way to make ai power-efficient

What would i do for 10 mins with quantum pc

[surreal creek]

just found it myself

voted for ChatGPT-4o over it though, lol

[golden ocean]

Ten minutes on a fault-tolerant, large-scale quantum computer capable of running Shor’s algorithm at meaningful key sizes would be enough to collapse most classical public-key systems if the attacker had already captured encrypted traffic. RSA and ECC would fall. Any stored handshake, any archived encrypted session, any intercepted key exchange becomes readable.

Such a device does not exist. Current quantum machines cannot factor RSA-2048 or break ECC; they are orders of magnitude too small, too noisy, and too slow.

If a future machine reached the required qubit count and error-corrected stability, preparation by pre-recording traffic would make the ten-minute window sufficient to run the needed quantum circuits and extract private keys, which cascades into broad compromise of internet security for any system not migrated to post-quantum cryptography.

[sand trail]

For how long does LMArena host a website that was generated at my behest?

[echo aurora]

The preview link from Code Arena?

[undone ravine]

How can I create a one-minute video using AI for free?

Please 🙏🙏

[unreal hatch]

What year do yall think OpenAi will release Garlic?

[pearl drum]

...This has happened to me ever since today, every model says "something went wrong" even for test messages. Did anything change overnight?

[whole sundial]

#1397655624103493813
<@&1349916362595635286>

[quick jackal]

@knotty whale Hello! please read https://discord.com/channels/1340554757349179412/1397655624103493813 to learn how to use the bot properly and https://discord.com/channels/1340554757349179412/1397655695150682194 https://discord.com/channels/1340554757349179412/1400148557427904664 https://discord.com/channels/1340554757349179412/1400148597768720384 for your creations.

[fickle venture]

https://huggingface.co/spaces

This one has free models kinda

They might have a video generator

[fickle venture]

When you send message?

Maybe refreshing could fix it or there might be a bad word in the prompt, So lmarena refused to send it.

[fickle venture]

What's that?

[astral blaze]

holy
What did they do to make the site even more unusable now

[native yarrow]

what does NB pro on open router even do?

its obv not uncensored right?

[stiff nymph]

hello

[obtuse smelt]

hi

[lapis imp]

So this might be a stuuipd question someone mentioned token limit? After you get the retry how lognn do you agve to wait 1 hour or 2hr or? Is the toke limit aka the rate limit?

Hope you'll understand what I'm trying to ask lol.

Also forgive my spelling mistakes

[burnt flax]

Hello Andre and I'm all here cuz I want to learn how to edit in create videos and edit photos and images I'm trying to create a podcast and I want to be able to edit my own material as well as create some material myself for skits

[keen beacon]

https://youtu.be/rfHfPxGReCE?si=tpwWNeO6HOQ57E2Q

Told u

[jovial sapphire]

hi guys

have u ever seen an edit like this?

[jovial sapphire]

like this style

[keen beacon]

No

[jovial sapphire]

like tracking style

[jovial sapphire]

what do u think

is it like beautiful

[keen beacon]

Ya

[proud bobcat]

No

But

It has a little less guardrails

[native yarrow]

ah

[timid yew]

how to generate video in LM arena?

[desert abyss]

Hello! Please check ⁠how-to-video-bot to learn how to generate videos.

[keen beacon]

I figured out how to unlock Pokémon

[jovial solar]

Why grok don't have images button

[cloud zinc]

how

[plucky sparrow]

did gpt 5.3 come out or was it fakenews as expected?

*5.2

[vivid coral]

Has anyone found any anons in the search arena? I haven't. Weird there's 1000 anons in text arena but not search arena

[hot pebble]

Hello guys, is anyone else experiencing “ Something went wrong with the response, please try again later “ error mid chat ?

[digital forge]

Has a 50-minute waiting time been added to the regular Claude Opus 4-5 session?

[hot pebble]

I am using Claude - opus-4.5 and Sonet 4.5. Same problem with both models. But works when open a new chat….

[hot pebble]

Waiting time ? After how many prompts?

[vivid coral]

Claude has the worst rate limits on this app, by far. They are literally run by communists. That's why they are going for regulatory capture. Government approved monopoly and they will be able to charge insane rates for unlimited or increased limits.

[digital forge]

I don't know exactly how many requests, but I created a new chat with Claude 4-5 and, after a few messages, I was blocked and can only use the chat again after 50 minutes. Until an hour ago, this waiting time didn't exist in this version, but it did exist in the Claude 4-5 (Optus Thinking) version.

[vivid coral]

We need a full on boycott of Claude to send those commies a message we won't bow down to their crap

[hot pebble]

I was using Sonet 4.5 20250911 yesterday night and got “ Something went wrong with the response, please try again later “ error mid chat….. then i started new chat using opus 4.5… now same error again…. But after your message, i went back to my old chat and still same error….

[bleak lake]

Clear browser cache

Then log out and log in back

[hot pebble]

I agree. On official website, Yes. On Lmarena, No.

I used the website and it allowed only 4 prompts. But Lmarena allowed more than that…

[hot pebble]

But i also tried on my phone, same error. And this was the first time i used on phone, so how come same error again?

[bleak lake]

are you only experiencing these on claude models?

moreover log out and log in back

usually it works for a ton

[hot pebble]

Honestly, i haven’t use any other models yet. I am only using Claude for a coding project so…

[digital forge]

I can't start any chat with any artificial intelligence on my computer, so I open lmarena on my phone and create a chat there, sending just one message. Then, I try to access that chat on my computer and the error disappears. Sometimes, the error reappears during the chat, so I send another message from my phone.

[hot pebble]

So i can go back to my old chats and try again after clearing cache?

[bleak lake]

I see

[bleak lake]

of course they are not related to cache memories

[bleak lake]

log out from your account in lmarena and log in back

and clear cache

[hot pebble]

Good to know. But for me its been 12 hours and yet i couldn’t access the old chat… but according to pranav it works so imma try that

[bleak lake]

you should try both

[hot pebble]

Same thing….

I cleared my cache memory, logged in again…. Typed a new prompt….

@bleak lake any other suggestions? 😶‍🌫️

I also forgot to mention, switching to different models yet pops up same error…. So is this a problem with the Lmarena itself ?

[bleak lake]

Ping moderators ig

[hot pebble]

should i just @ moderator ? Or any specific person?

[chilly nexus]

Number of captcha is insane, i have to fill a captcha for each new message

[hot pebble]

@quick jackal bro you here?

[bleak lake]

@echo aurora

I did it for you

[hot pebble]

Thanks bro

[vivid coral]

He's probably asleep, its 2am, lol

[hot pebble]

Ah okok. My zone is IST so it’s afternoon for me 😅

[thorny schooner]

You know a glitch is bad when even after a full-on reset it does not go away ( was turning on developer mode)

[grave plaza]

hii

[hot pebble]

Now this is seriously bothering me alot….. after every few prompts i am getting same errors….

If the limit is reached, it should mention that.
If i switch to another model, it should atleast respond and not show same error….

[compact flame]

My friend had the same problem honestly

Idk if he fixed it or not

Probably could be due to bad internet maybe

[valid dust]

hello, came here for the fun video making

[hot pebble]

Nope…. Starting a new chat always worked…. But this error is coming mid chat…. And wont load even after reloading, logging in again, clearing browser history/ cache

But every time i open a new chat i have to load the data from previous chat again, and if there is a limit of prompts, then i am already using multiple prompts just by typing old data

[hardy lion]

hello and welcome check out #1397655624103493813 for instructions on how to use the video bot!

[coarse cradle]

How can I create a video with sound?

[hardy lion]

only some of the available video models support sound, so it's up to random chance if you get one with your request

[coarse cradle]

OK ty

[valid cobalt]

The overlapping brand icons is a chart crime and LLMarena should be ashamed of themselves. I can promise you, nobody is going to be fooled. You should fix it or lose all credibility

[fringe fern]

hello

[hardy lion]

lol are you serious?

[valid cobalt]

grok is second or it isn't. if it's second, it's brand icon should overlap anthropic. Simple as that.

[hardy lion]

it was generated using https://app.flourish.studio/ we didn't set that deliberately it's probably like alphabetical or based on the placement at the beginning. The video has the names in order the ranking is clear at all times.

[atomic lagoon]

Wth is this?

oH SORA

[whole sundial]

why does the z.ai graph only begin with the glm 4.5 release? i get that it wasn't called z.ai until around the time glm 4.5 came out but couldn't the previous glm/chatglm models that have been on lmarena for a very long time be included in the z.ai graph for continuity's sake?

[tulip tree]

https://tenor.com/view/ishowspeed-speed-reaction-try-not-to-laugh-bruh-speed-gif-15909423665204532371

[obtuse smelt]

future is coming

[nova pivot]

Please why is it I can't choose my video model

[brisk turret]

when flash and flash lite?

is flash lite being tested atm?

[queen veldt]

Why are y'all so hyped for flash?

It's a smaller model

[golden ocean]

fr

last time too when gemini 2.5 flash released

everyone scking it off with youtube thumbnails like "the new best model" like tf its a smaller model

i think its the poor people community celebrating

[neon idol]

Fr

[flint fog]

Hello, I'd like to report a problem with the LM Arena software, specifically with the Nano Banana Pro model. Sometimes the model stops responding and displays an error message. This has happened quite frequently recently. I hope this issue can be fixed. The same problem also exists, but to a lesser extent, in Cluade 4.5 Opus Thinking.

[elder merlin]

general

[queen veldt]

Yes mods are aware of it, for a while now

It might be the problem with nano banana itself

[sterile tartan]

Those are rate limits because this models are expensive

[flint fog]

Okay, but sometimes, even if I didn't use it that day, I get an error message.

[sterile tartan]

That would be due to server overloads i predict

[flint fog]

Fine, thank you

[sterile tartan]

Sry and yw

[quartz light]

[quartz light]

@gaunt spade

"its a wrapper for glm 4.6"
"its a wrapper for gemini diffusion"

1 million-token context window

[torn mantle]

its a wrapper

for gemini flash

[fickle venture]

Gemini 3.0 Flash coming soon.
https://youtu.be/nn-lTCcMJWM?si=3d427lr2pFexH0ma

[fickle venture]

They have a official discord server?

But wait what is tensor

I have never heard this name before

[torn mantle]

this channel has the most annoying ai generated voice

there are tons of good free TTS and small models and hes still using this

[golden ocean]

and intro

[torn mantle]

yea

that intro is annoying as hell too

I know he thinks that the voice now is kinda his signature but he needs to c hange it

[fickle venture]

I mean as long he provide information, I fine with it

[torn mantle]

how are you 100k and you cant afford a good TTS

even if you dont want to there are gazillions of free TTS

[sterile tartan]

Give the List now

[waxen fern]

Is rate limits still there in Lmarena???

[sterile tartan]

Always will be

Unless Ai becomes absolutely cheap

[waxen fern]

Bruh I want answers!!!!! 😭

[sterile tartan]

Try other tools too

U can try Yupp Ai if you want

[waxen fern]

@echo aurora please remove lmarena rate limits

[sterile tartan]

He can't and he won't neither

Compute is expensive bro

💀

[torn mantle]

kokoro / edge-tts

like so many

you can even run them locally

[sterile tartan]

Give the entire list

[torn mantle]

brotha

[sterile tartan]

I don't have a pc

[torn mantle]

i dont have a long list

i gave you two

[sterile tartan]

Alright tysm!

[torn mantle]

should be enough

[vivid coral]

Happy Gemini 3 Flash day everyone. Is it on the arena yet?

[sterile tartan]

Are these best ones?

[sterile tartan]

More likely as anonymous

[torn mantle]

they are not

there are better but i dont have a list

[sterile tartan]

I mean in Free List

[torn mantle]

these are more lightweight

[sterile tartan]

Bruh ok

[torn mantle]

no there is a better one but the model is big

[sterile tartan]

No worries I don't need for now anyways

[sterile tartan]

I need online tho

[vivid coral]

It's already been on there as an anon. It should be coming out today, according to sources

[torn mantle]

edge-tts is online

its an api wrapper based on microsoft api

its good

[sterile tartan]

That's great

Gotcha

Tysm again

May you be more useful in future

As well

[torn mantle]

https://www.tts.best/

you can try it here

ava is a good voice actor + andrew

there are like some 4 decent onces

[sterile tartan]

Gotcha ty again

[sterile tartan]

Name them

[torn mantle]

i forgot about the other two

ava + andrew are good tho

[quartz light]

i agree

[torn mantle]

andrew

[sterile tartan]

Gotcha

Now you are more useful

[torn mantle]

ava

i kinda like andrew better ngl

[sterile tartan]

Yeah

[sterile tartan]

Better for females

[quartz light]

isnt there a webgpu one too

[torn mantle]

kokoru model

[torn mantle]

kokoru

[sterile tartan]

Not bad at all

I could sleep on this

There is elevenlabs and minimax speech too

[quartz light]

not only

[torn mantle]

minimax?

[quartz light]

https://huggingface.co/spaces/webml-community/Supertonic-TTS-WebGPU

[sterile tartan]

The Chinese Company

[quartz light]

https://huggingface.co/spaces/webml-community/OuteTTS-WebGPU

[torn mantle]

lol since when

[sterile tartan]

Years ago

[sterile tartan]

Ty!

Are these yours?

[torn mantle]

its fast but meh

https://huggingface.co/spaces/TTS-AGI/TTS-Arena

there is this leaderboard

for tts

[sterile tartan]

My eyes

💀

[fickle venture]

There was a great TTS model but I think they removed it

It was sooo goood

[sterile tartan]

Maybe that's why it got removed

Alot of people would had been using it

[torn mantle]

im still looking for that model i mentioned before, maybe its what u're talking about

[fickle venture]

Not kinda probably it was expensive to host

[sterile tartan]

True

[sterile tartan]

edge-tts?

[fickle venture]

Hugging face is just so good they provide free working open source models

[quartz light]

nono

but they're good

[quartz light]

u mean the microsoft one?

[sterile tartan]

Ok

[fickle venture]

It's was long time ago I think it's made by Chinese people

[sterile tartan]

Chinese are everywhere

[fickle venture]

Because they are smart

[torn mantle]

they provide free server/vps

i think its 12gb RAM

[fickle venture]

Yeah so you can host your own ai

[sterile tartan]

There is also colabs i guess

Kaggle

[torn mantle]

im already hosting my own stuff 😭

its handy

[sterile tartan]

But idk if they allow to host ai

[sterile tartan]

Where

[fickle venture]

Huggingface automatically configure your opensource ai

[fickle venture]

https://Huggingface.co

[sterile tartan]

Daym they allow Free hosting wow thx

[sterile tartan]

Gotcha

[torn mantle]

huggingface spaces

[trim portal]

Hey sorry to bother does the model in direct mode also hide their identities? I asked gpt 5.1 from lmarena and it claims to be gpt 4.

I used real gpt5 to evaluate the response from the model, real gpt5 said the response is clearly gpt4

Am confused

[sterile tartan]

Noice

[sterile tartan]

The ai models don't know their identity themselves

The ai identity crisis

The models are as stated

[trim portal]

I asked gpt5 to make some reli hard reasoning problems

give that to the model and fetch the response back to gpt5

Gpt5 claims that's clearly gpt4

Maybe that's not good validation?

[sterile tartan]

Because that screenshot you gave

Any ai will say that

Is not good validation

And asking a model for its identity is just not Efficient

[fickle venture]

Has this released to Lmarena?

This one provide free opensource models.
https://huggingface.co/spaces/hadadxyz/ai

[echo aurora]

Sorry to say removing rate limits is very unlikely.

[sterile tartan]

Afaik Free users get 4 minutes of daily compute for using this spaces?

[fickle venture]

Yeah I forget about the limit

[keen beacon]

why now in nano banana have limit for make images?

[torn mantle]

nice

[keen beacon]

im Lmarena

[fickle venture]

Because "too expensive"

[sterile tartan]

Can we have multiple accounts

[keen beacon]

ok

[fickle venture]

On hugging face probably but the limit is just so small

So it won't be helpful

[sterile tartan]

4 minutes yeah

Pro users get 25 minutes

[echo sinew]

@pure torrent Please have a look at ⁠⁠https://discord.com/channels/1340554757349179412/1397655624103493813 for a step-by-step guide on how to generate videos using the bot.

[floral quarry]

Why I got error when I upload image made from lm arena and I want to edit it with ai?

[neon idol]

Not here lil bro

[bright shard]

@echo aurora Nano Banana Pro is having many errors again

[echo aurora]

Appreciate the heads up, are you seeing this with the 2k version, both?

[bright shard]

For now, I've only used the regular version; I haven't tested the 2K version much.

[glass zealot]

this one little error is so annoying… i think it depends on time when i use lmarena. sometimes my generations are consistent successful and sometimes i get this error 9 times per 10 requests

[twilit delta]

what fix of this

[torn mantle]

antigravity gotta be the worst IDE ever

omg

[balmy mist]

i havent used it since launch, it was amazing that first week, but i stopped when i got hit with usage limits

did it get worse?

cause it was doing crazy stuff that I could not even do with other IDEs and ai clis

[torn mantle]

The file on disk STILL shows the broken syntax! The multi_replace_file_content tool is reporting success but the changes are not persisting to disk. This is a serious issue with the editing tool.

even opus 4.5 is furious lol

ive been trying to fix this issue last 2 hours

IDE tool call is so messed up

"This is a serious issue with the editing tool." lol

finally opus fixed it

took me like 40% of opus usage

holy

[proud bobcat]

bruh

they put gpt here and not deepseek??

you gotta be kidding me

[torn mantle]

cursor new model

probably released today

[hardy lion]

hi @tulip tusk please read #1397655624103493813 and use the video-arena channels for video requests, thanks!

[fleet lintel]

where is this from?

[iron glen]

is there a way to upload 2 images to generate one video?

[echo aurora]

There is not sorry to say

[wispy cloak]

HI everyone, hope to learn new stuffs here.

[echo aurora]

Welcome welcome

[proud bobcat]

openrouter

[proud bobcat]

its benchmaxxed

[hardy lion]

I think because there weren't anywhere close to the top 10 before that

[echo aurora]

Gotcha. Unfortunately, we have been experiencing high error rates with this model. The team has been made aware and are looking into solutions to lower this.

cc @glass zealot

sometimes my generations are consistent successful and sometimes i get this error 9 times per 10 requests
It's difficult to say what's causing this error as the Something went wrong is a generic error message which can be caused for different reasons. However, what you mentioned here points to rate limit being the culprit.

[neon idol]

<@&1349916362595635286>

[queen veldt]

Cursor is so op

I've just tried it

[dusk phoenix]

Hello!
Can someone tell me what this model "Hazel-gen-4"?

and how to select it in direct chat in the image generator? i found it only in the battle mode

[heady pine]

gotta follow the rules 😃

[zealous sparrow]

272k context

if this beats gemini 3 pro the bench is rigged

also new battle model that has a search output [searcharena only]

first battle model for searcharena in a while

[fossil fable]

burn openai

[zealous sparrow]

anyone heard of big sur AI

[twilit delta]

what fix of this

[zealous sparrow]

Can't

this is to prevent people from spamming to get models they want on battle

[neon idol]

I cant see it on cursor

[echo aurora]

When it comes to models that are using codenames, you won't be able to fidn them in Direct or Side by Side mode, they can only be found in Battle.

[neon idol]

Hai pineapple!

Hru?

[echo aurora]

Howdy

[echo aurora]

Doing well. We had our company holiday party last night so a little tired, but overall doing well.

Yourself?

[zealous sparrow]

Think it was in their API or smth

[neon idol]

Oh nice!!

[neon idol]

Chilling and doing homework

Nothing special (⁠≧⁠▽⁠≦⁠)

[echo aurora]

Nice nice, preparing for finals kind of thing?

[neon idol]

No

For nothing

[torn mantle]

its hidden

but its in the metadata

so its probably coming tomorrow

[whole sundial]

this is grok, menlo is the codename for grok 4 fast search

in fact, if you use that model right now it still identifies itself as menlo by big sur ai

[zealous sparrow]

hm alright then

why would it be on searcharea

arena if its grok 4

I meant battle.

[whole sundial]

...because they want to evaluate their new search model before release? like any other model on lmarena?

[zealous sparrow]

wait how do you know that is the codename if the model is new

[whole sundial]

because it's not a normal model name?

[zealous sparrow]

Think about one thing tho.

Will Menlo always mean grok?

[whole sundial]

if it says Menlo by big sur ai, it will likely always be grok

it's an lmarena-only designation anyways

[zealous sparrow]

Alright then

they removed a raptor model from textarena

[hushed gyro]

Uhh chat did chatgpt release adult mode?

[proud bobcat]

my thoughts exactly

5.1 is just a finetune of 5

theres no way 5.2 will be eons better magically

if 5.2 is a flop i think its safe to say openai will take a backseat in the market

[zealous sparrow]

the worst thing you can do is code red and rush models

than be google and make people wait

google took months to make a model and people were happy

openAI took a few days to make one and people were unhappy

[sterile tartan]

@whole sundial is this good?

[cloud zinc]

yes they will in december

[hushed gyro]

It's like the 10th of December rn

😭

[cloud zinc]

next week

[hushed gyro]

If they make it paid plan only imma crash out 😭

[cloud zinc]

its gonna be behind ID

[zealous sparrow]

most likely

[cloud zinc]

good, we dont need little kids using it

[zealous sparrow]

OpenAI takes the risky route for operation, gemini 3 pro releases, Oh no! Panic panic, we need a model that beats it so we don't get forgotten.
Instead of, oh congrats, now we take time to work on a model to beat it.

[meager harbor]

Does anyone else feel that LM Arena doesn't punish lying enough?
People are more likely to vote for an AI that makes stuff up than for one that admits it doesn’t know!
There should be an option to justify your vote by saying the other model lied, so LM Arena can add a “truth” category where the highest score goes to the model that lies the least.

[proud bobcat]

that would be cool

you could prob add it to feedback

[cloud zinc]

google has tpu advantage

[proud bobcat]

id give you a thumbs up

[proud bobcat]

doesnt matter atp

openai were the PIONEERS

[cloud zinc]

openai needs those datacenter up asap

[proud bobcat]

datacenters wont fix their ass architecture

[meager harbor]

thanks i will do that !

[proud bobcat]

gpt 5's architecture is plain ol ASS

ineffecient

bulky

too dense

[cloud zinc]

how u know

[echo aurora]

Yeah this is interesting, would be really curious what others think about this + how it'd be done. Starting a thread in #1372230675914031105 would be ideal.

[proud bobcat]

look at the diff between 4o and 5

4o was perfection

sure it wasnt a benchmark king, but it was a reliable model

4o consistently hallucinated less for me than 5

5 came out and despite them saying that it hallucinated like 50% less or whatever

[formal dagger]

still tiny

[proud bobcat]

for me it hallucinated around 80% of the time

and that is not an exaggeration

[zealous sparrow]

logan kilpatrick said that like they dont have enough TPUs to make their models even more powerful for now

he legit said it

[proud bobcat]

openai had such a good oppurtunity with 5 and they blew it

now deepseek is basically taking over their spot as the reliable workhorse

[zealous sparrow]

until they fix their awful frontend [chatgpt]

it wont beat anything

i would argue grok beats it

[proud bobcat]

its not even the frontend

the model just sucks

they need to do a good rerouting of their entire company

[zealous sparrow]

robin-high was a good model of theirs

frontend still sucked

but

[cloud zinc]

no it was about deepthink

[zealous sparrow]

ah

also btw

me and nickname figured out why google put 4 models into battle

2 are flash-lite [one thinks, one doesn't]
2 are flash [one thinks, one doesn't]

[meager harbor]

Ok i made it : https://discord.com/channels/1340554757349179412/1448392968653504604

[echo aurora]

Awesome, thanks

[meager harbor]

thanks, check this : https://discord.com/channels/1340554757349179412/1448392968653504604

[torn spear]

Is there anyone here looking for developer?

[torn mantle]

https://x.com/mark_k/status/1998830017300557983

Mark Kretschmann (@mark_k)

Gemini 3 Flash by @GoogleDeepMind can now be added as a model in Cursor!

Interestingly, the model appears with a 🧠 symbol, indicating a "thinking" model. Gemini 2.5 Flash was a non-thinking model.

This means that the release is getting near!

X · Dec 10, 2025, 6:59 PM

😖

[zealous sparrow]

already or did they mess up

[hot pebble]

I have been asking for a solution for this error in Claude Sonet and opus 4.5 but no mods are available I guess… 💀

I have 3 chats open coz after few prompts each chat gave this error, been 24 hours yet same errors even if i type a new prompt

[torn mantle]

i have no idea

[zealous sparrow]

people are also saying its today bc of this

wait

https://x.com/AdamHolter84937/status/1998718520893288651/

Adam Holter (@AdamHolter84937)

It's confirmed. Gemini 3 Flash today.

X · Dec 10, 2025, 11:36 AM

2.5 flash tts is being uh

removed

so people think its today

[rain drum]

Can anyone tell me how to fix this

[torn mantle]

i see

when do they usually release their products

[zealous sparrow]

thursday or a tuesday i forgot

when did gem 3 pro release

[torn mantle]

no i mean time

is it still possible to be released today or nah

[zealous sparrow]

um

wait let me uh

compare with g3 pro release

it was a tuesday @torn mantle

Wouldn't get your hopes up

[fleet lintel]

it's not necessarily gemini 3 flash

[frosty lava]

Why do cursor have access to new model before anything else ?

[zealous sparrow]

I know its a TTS model

[fleet lintel]

tweet is confusing between tts and flash model

[zealous sparrow]

btw google might drop flash lite with flash

because integrated-info and holo-scope might be flash-lite models

[proud bobcat]

done

[proud bobcat]

oh my god its peak

gemini 3 flash is gonna be so good

[zealous sparrow]

yeah but the thing is

tts models are seperated from the model arent they

[fleet lintel]

yes they are different

[queen veldt]

Why are people hyped about the flash model

????

It's a flash model??? Nothing special

[fleet lintel]

super fast, very cheap. not far from gemini 3 pro performance

[zealous sparrow]

guys did skyhawk and seahawk get removed from codearena or are they just rare

i haven't gotten it for a few gens now

[fleet lintel]

very good chance (in fact willing to wager) that gemini 3 flash is going to be better than 2.5 pro

[zealous sparrow]

prob so

[queen veldt]

It doesn't have thinking tho?

Or just the flash-lite one

Doesn't have thinking

[fleet lintel]

gemini models (since last version I think) have thinking by default . gemini 3 flash is definitly going to have thinking

Lmarena needs to remove these useless Nova models. Total waste of tokens and energy. 😠

[fiery vault]

i want to generate ai videos

[queen veldt]

#1397655624103493813

[compact flame]

Any news on chatgpt 5.2?

Cuz I think it was supposed to be released yesterday

[sweet topaz]

👋

[zenith scarab]

👋

[upper locust]

👋

[pale reef]

👋

[gleaming roost]

👋

[sterile tartan]

Let me break this chain

And make it rain

Gemini 3 flash will be SOTA 2

[half pelican]

👋

[frosty lava]

is flash like a lower cost version of the pro version and less capable of course ?

or this time it will be something else

[sterile tartan]

My prediction says

Google is Cooking

Like they cooked Gemini 3

[neat apex]

Ok i will stop now

[hollow ivy]

-# still 3 weeks of 45² left..

[neon idol]

<@&1349916362595635286>

[frosty lava]

honestly they should all focus on hallucinations

[neat apex]

Overreacth final boss

[neon idol]

@echo aurora Come immediately here

[compact flame]

Why did the chat just suddenly revive

[sterile tartan]

True

[neat apex]

Someone did a chain, but we stopped naturally

[torn herald]

👋

[neat apex]

Nahh, lets stop

[compact flame]

[neat apex]

We are giving slop to the new vicuna model they must be training

[compact flame]

So guys any news on chatgpt 5.2?

[echo aurora]

Yep I'm seeing

[frosty lava]

I don't think we actually need very capable ai if they hallucinate like that i want an ai that WON'T hallucinate cause its time wasting and risky

[neon idol]

Niiiice :D

Mass ban?

[neat apex]

They could easily do a IA that says "i dont know" everytime, but it would make it 5% dumber and they dont care

[echo aurora]

Not really sure tbh, I did remove the 👋 from #leaderboards

[neat apex]

Since fine turning does that

[sterile tartan]

💀

What's going on

[echo aurora]

But yeah I don't want to go on a mass ban for people that 👋

[frosty lava]

I think they're working on it but its not what they focus on sadly

[fleet lintel]

i recommend 24 hours ban

[sterile tartan]

👋🏻❤️

[neon idol]

Bud i really respect u with all of my heart but these people are toxic and are just destroying YOUR community

[weary galleon]

There are good news!

[neat apex]

Oops, you are now in the 👋 chain, you will get banned too

[sterile tartan]

#Pen + Apple =

[sterile tartan]

Fake

[compact flame]

I know this image I'm not getting fooled again

Well maybe

[echo aurora]

[sterile tartan]

What if is real

[sterile tartan]

U should not mess with the mods

[compact flame]

I've seen this image before

It had Gemini watermark

[sterile tartan]

Yeah but what if it was real this time

[neat apex]

Yeah, i edged too much here, stopping now

[sterile tartan]

🤔

[echo aurora]

I don't think it's too disruptive tbh

[weary galleon]

I removed it

[compact flame]

He cut it off and it's low quality

[zealous sparrow]

synthid is prob still inside

[sterile tartan]

Maybe you should apologise

Maybe is alright

Idk tbh

[neon idol]

Fine

[sterile tartan]

💀

[weary galleon]

Not all people download it

[neat apex]

I already know pineapple, he is also joking a little too (most times at least)

I think Garlic is Gpt 5.5 not Gpt 5.2

[echo aurora]

I'll start removing folks that do 👋 multiple times + if it's their only contribution to the server

[neat apex]

Since they need do a lot to suprass Gemini 3 pro confortably

[slim bobcat]

Does anyone know what is holo-scope?

[echo aurora]

But for folks that are just 👋 they can stay

[sterile tartan]

Garlic will be new name of GPT

[sterile tartan]

And it will be renamed as Closedai

[neat apex]

Garlic Pasta Temperoni

[neon idol]

Oook

[compact flame]

Does garlic mean like chatgpt will be stinky or something

[sterile tartan]

Is just a codename

A ai can't stink

💀

[weary galleon]

Garlic has wonderful smell

[compact flame]

Okay fair point

[sterile tartan]

https://tenor.com/view/good-question-you-are-a-good-question-deepak-dobriyal-tanu-weds-many-gif-21433901