general

[undone geyser]

15 months jobless

[sly cedar]

i see, have you ever plan to make something for profit?

[undone geyser]

nah

actually would like to learn to make sites using ai

[sly cedar]

oh

what type of website you would like to create?

[undone geyser]

any client related

idc

nail salon, barber...

i use a program called mobirise but its plain jane

meh

i wanna use arena.ai along with whatever is on market that is best for site making

[sly cedar]

I see, and yeah arena ai is quite helpful, it actually helps my game project that i've been dreamed on ,

[polar wharf]

new image model is beating seedance 5, imagen 4, z-image and p-image
https://artificialanalysis.ai/image/leaderboard/text-to-image
(image-1)

[undone geyser]

i live in greece dude, it is like minecraft on perma hardcore survival

[toxic verge]

cluade 4.8 leak

https://www.ainexusdaily.com/article/anthropic-source-leak-claude-sonnet-4-8-undercover-mode

[undone geyser]

on may 19th i think we have google i/o...

nano banana 2 pro!!?!?!?!

[silver hearth]

https://tenor.com/view/pokemon-pokémon-slowpoke-glitter-sparkling-gif-9077674878617351176

Code name capibara...and mythos.

[toxic verge]

If it is Claude 4.8 would that make 4.7 a flop

[vernal raft]

Leak

[light sleet]

@surreal zephyr

[surreal zephyr]

Uhh

False evidence

[light sleet]

Nahh that was u

This you? @surreal zephyr

why your Jawline kinda fife

fife

Fire

u looksmaxxing?

[whole mural]

they said that gpt5.5 performed better than mythos

[light sleet]

Well it did

[whole mural]

should we believe them or not

[light sleet]

We should

[sly cedar]

claude ai is even breaking their company database

[river pendant]

some models work, some not, but make sure to check it out

[light sleet]

5.5 pro does?

[river pendant]

idk

but likely it can't due to no credits

eleven labs model working

tho

check out in tts section

[whole mural]

WTF is gpt 5.4pro with NO reasoning

so fake

[river pendant]

no

[light sleet]

no chat models work lol

[river pendant]

just open router allows to use reasoning and none

[river pendant]

yep

cuz no credits, that's why I warned

[light sleet]

no point then

[river pendant]

yeah I guess

but tho, soon as possible, you guys can use it when I pay, basically only developer as me paying, not users

[toxic verge]

https://tenor.com/view/let-freedom-ring-bell-big-bell-gif-14513327

[river pendant]

yep

[river pendant]

yep, let's freedom ring

[mint vigil]

Why everytime has a captcha to me solve and give error?

[toxic verge]

https://tenor.com/view/bro-did-you-hear-the-news-rhino-love-live-serve-have-you-heard-big-news-gif-19003928

[river pendant]

plus, in my project, there's no stupid captchas for models to chat with, it's better than arena.ai

[whole mural]

use VPN solves like 90％

[mint vigil]

Ok bro

Ill see this

[cedar citrus]

arena ai serves millions of users per day

[river pendant]

yeah, but soon as possible I'll hit same results when I pay for it

[mint vigil]

You are badass, thanks Bro @whole mural ❤️

[river pendant]

for the project myself, and models and stuff

[light sleet]

how will u plan getting money

[river pendant]

trough YouTube monetization

by uploading videos and stuff

to receive views and likes and comments that mostly be positive

plus, trough donations, trough Kofi

or some other donation website or app

by including one button to pay

[quaint dirge]

y

[light sleet]

@surreal zephyr @surreal zephyr dm me

[river pendant]

plus, my project will contact tech arena, where you can test different stuff with rom, I will might add the Google pixel to make it been added with custom ROM, so you guys can use it, but this happen when I need

[surreal zephyr]

3.1 flash lite was out since 5 years 🥀

[whole mural]

when will gemini 3.1 flash release

[surreal zephyr]

Its out since ages, 3.1 flash lite

[light sleet]

worse than 3 flash btw

[river pendant]

yeah, btw, PlayGround have Gemini 3.1 flash Lite, but Gemini 3 flash also there

but it's worse than Gemini 3 flash, so you right my guy

[light sleet]

yep

[river pendant]

yeah

[light sleet]

@river pendant look at this

rip anthropic

[river pendant]

lol

It does make sense, cuz gpt-5.5 and pro version of it, kills the Claude Opus 4.7 series

tho, opus models are in playground either

not like arena.ai who remove em

Ik opus expensive but you guys also could have none reasoning variant

[surreal zephyr]

Obviously

[light sleet]

sam altman mog +1000000

or should I say ampro

[surreal zephyr]

3.1 pro is worse than 3.0 pro-base

[light sleet]

ampro and Elon Musk in court leak

[river pendant]

yeah, but no, it's improved itself, a lot, than Gemini 3 pro

[light sleet]

see that's you

[surreal zephyr]

Lies

[light sleet]

nah

I have more

[light sleet]

Ok maybe that's not u

Because altman mogs

[tight meadow]

I got that using video arena

[river pendant]

damn! that's kling 3.0

15 seconds don't remind smth?

[light sleet]

ah there u are @surreal zephyr

@river pendant another anthropic base cooked

[river pendant]

holy cow, insane!

well, that's explained a lot

[light sleet]

Ty

[river pendant]

because Claude is sucks compare to gpt latest model

[river pendant]

yeah, no problem

[light sleet]

@surreal zephyr found u

[primal widget]

genrate audi rs 268*188

[river pendant]

that's not working here, only in arena.ai video arena

[primal widget]

give me link pls

[river pendant]

oh god, fine

here: https://arena.ai/video

[primal widget]

thx

[river pendant]

no problem

[glass echo]

Hey is anyone havining an issue with their chat list no longer updating with the latest used chats?

mine sometimes shows the most recent but right now its showing me chats i haven't used for the last day as the top.

[river pendant]

plus, I need this, if you asking why

[tight meadow]

its seedance 2

[river pendant]

just they look same because kling and seedance might have same quality and seconds

if I'm right

[tight meadow]

Oh

Its 1080P

[river pendant]

Yeah

[ornate cloak]

:54736ac3-8204-**:54736ac3-8204- i keep getting this error when using arena, can anyone shed some light on this or how to fix?

[tight meadow]

I will try with Kling 3

If i can lol

[river pendant]

yeah, try

but I bet it's not available yet, only my platform called PlayGround have this, but it's not working because no credits yet, it's have std and pro, (std - standard)

plus here's the url to it: https://playground-1.on.websim.com/?v=798

some models work, some not, but that's ok, you'll find out

[tight meadow]

IM ACTUALLY MAKING IT LOL

[river pendant]

Woah!

[tight meadow]

Its halfway the generation

[short sluice]

I try making an image

Basically my phone and PC are on different ips

Yet I still get limited

I am not connected to wifi

Did they change it from IP to location now

[tight meadow]

Over half now

[river pendant]

wait so you did that in PlayGround?

[river pendant]

or in arena.ai?

[tight meadow]

Im making iy using Luma AI.

Thank me later

[river pendant]

alright

thanks I guess

for letting me know

[tight meadow]

It has Kling 3.0,Kling 2.6,Veo3.1,Veo3,Ray3.14 and Ray3.14 HDR

First is good,2nd is slightly worser,Veo good,Veo 3 good,Ray good but if im sure has no audio or include audio option

Kling 3.0 is the 1st video,Seedance 2.0 is the 2nd video

Proof its kling 3

[river pendant]

Woah! I wish I can have API option to implement these, except veo 3.1 and kling 3.0, I need to implement those that not available in PlayGround

can I find it?

[tight meadow]

There

IDK

Did you make the playground

[river pendant]

Yeah

I did

[tight meadow]

Can yiu give me your api

I want to use it rq

[river pendant]

it's open router

api

[tight meadow]

Api key tho

Chat me

[river pendant]

god, create you're own API key

it's in open router available

if you want, I can give you step by step guide on how to

[tight meadow]

I've used AI for over a year, so I know a lot about Artificial Intelligence.

How about you

[whole mural]

starts to use AI since gpt4o

😂

[river pendant]

me either

[tight meadow]

Ohh

[river pendant]

yeah

I knew about ai around 2024, where I discovered what websim was, and I've started to learn about that

[tight meadow]

I used in 2022 actually

[river pendant]

well, that's fine

different years I guess

[whole mural]

???

[tight meadow]

When AI was terrible at images lol

[whole mural]

gpt3.5turbo?

[river pendant]

yeah, when basically it can't generate properly

[tight meadow]

YEAH

If im sure one of the models is called like Dall-E

[river pendant]

well, first generation I guess of dalle

[tight meadow]

It had so much imperfections.

[whole mural]

actually u could still use it in arena

[tight meadow]

I still use it

[river pendant]

anyways, about gpt-3.5 turbo, it's available at platform playground, some series of it, basically I've created playground just like arena.ai, did you need link? if so, then ok

some models work some not

[tight meadow]

I guess so

[whole mural]

how about gpt2

[river pendant]

alright

[river pendant]

It's unavailable there, plus it's not available to be published as gpt-1

here link for you guys to check out: https://playground-1.on.websim.com/?v=798

[tight meadow]

Minedont work

I dont have credits

Ye

I have zero credits

[topaz epoch]

@echo aurora failed to get email provider

[uneven cypress]

No credits

[topaz epoch]

I'm gonna lose my account no hell nah , whole months of data

[river pendant]

ik

[uneven cypress]

Bruh

[tight meadow]

Can you at least add 1 dollar of credit

[whole mural]

well maybe use google to register in the first place next time

[tight meadow]

I have sora 2

[wild blade]

@everyone give it a listen 👂🏻

[hybrid ridge]

Hi, what model you recommend for generating CSS?

[wild blade]

Claude

[tight meadow]

Claude

[hybrid ridge]

Okay thx.

[storm dust]

thats really good

did ai make that?

[wild blade]

Yep

[storm dust]

impressive

[wild blade]

50 percent

Ai is getting good day by day

Look at the flows in this even good rappers can't these days

[storm dust]

eminem kind of flow lol

[tight meadow]

Try veoaifree.com

[wild blade]

NF to be precise

[tight meadow]

It only uses veo 3.1 and imagen 4 tho

[wild blade]

The flow is heavily influenced by NF

[undone geyser]

have patience guys, when summer starts, june 1st that is, i wanna believe on arena.ai, all opus models and gpt and muse spark will come back on direct chat

and also we will have nano banana 2 pro

[light sleet]

E

[vivid oasis]

https://cdn.discordapp.com/attachments/920629083342270515/1497670659361341522/rapidsave.com_0te38506tcxg1.gif

[hollow cosmos]

Is there any other way to get opus free ?

https://tenor.com/view/zuh-cat-side-eye-stare-cat-staring-ca-gif-7404392670733848800

[wary nacelle]

yes but i wont tell

[plucky fractal]

anyone noticed gemini 3.1 and 3 flash have been insanely good recently?

i mean, on arena

[hollow cosmos]

Can you please tell me in dm's if possible 🥀

[plucky fractal]

no idea what they did but they don't feel like the same model?

[wary nacelle]

if u mean the design obviously yes

Gemini 3.1 always was good at design

[wary nacelle]

since they are preview models

they are actively being worked and changed

[plucky fractal]

nope. not only design. i got gpt 5.5 high vs gemini 3.1 on code arena and gemini did a 2x better job

im kinda shocked. gemini 3 flash is also better than it was

something going on here.

[wary nacelle]

obivously... GPT 5 was the biggest failure of OpenAi that even OpenAi understands just doesnt give up

[plucky fractal]

yeah maybe google updated their models. that makes sense. so not a new model but a big update on their existing models?

[wary nacelle]

thats literally what preview means-

[plucky fractal]

idk but beating gpt 5.5 high on code arena is kinda shocking and it wasn't a simple coding task

[wary nacelle]

If it beats claude code THATS going to be shocking

in terms of stable and scalable code

[plucky fractal]

it was a highly complex simulation task

[wary nacelle]

and optimized

not design

[plucky fractal]

anyway. good to see gemini getting better it's my fav model

[plucky fractal]

yeah got it thx for the answer 🙏 will keep testing them

[wary nacelle]

I use gemini mainly for design since if i tell it to do code its not gonna do it on the same level as claude code

[cedar citrus]

guys i just found new kimi model in lm arena

[whole sundial]

that's not going to work

[plucky fractal]

bruh 💀

[whole sundial]

that's localhost, we can't access your server like that unless I was on your network

[plucky fractal]

codename?

[cedar citrus]

solar-eclipse

i sent it at codename discussion

[plucky fractal]

oh 👀

just saw it yeah

[cedar citrus]

son

[wary nacelle]

Bro had job one

[hollow cosmos]

[light sleet]

gpt 5.5 so good

opus cant even do the same lol

[primal depot]

уберите ебаную капчу ишаки

[cedar citrus]

nah]

i mean not in my opinion

opus for me is alot better

[light sleet]

prove it

No advertising here.

[cedar citrus]

well gpt 5.5 produces unoptimized really bad performance roblox code also the code is worse than opus and opus performed exactly like a expert dev would

[whole sundial]

<@&1349916362595635286>

[cedar citrus]

<@&1349916362595635286>

[hardy lion]

we need roblox bench

[wicked talon]

Grok actually made supergrok lite lol

Fairplay

[normal abyss]

If claude is the best at coding

GPT is the best at images

Gemini is great at images and both coding (a duo id say)

Where does grok stand????

[little thorn]

hey
why aren't powerful models like Claude Opus, Gemini 3.1, and others available on the arena.ai site?

[digital umbra]

because they cost money

[thorn coral]

The limit is 5 images per hour right ?

[little thorn]

Are these models disabled for everyone now, or is it just me?

[plucky fractal]

it's not just you. those model are only available in battle mode. it's been like that for a while.

think they're working on a solution to bring them back. there are discussions.

[gritty dock]

Where can i create my video

[hardy lion]

https://arena.ai/video

[gritty dock]

You cant make them anymore in discord?

[hardy lion]

no they are on the website now

easier to maintain one platform than 2

it works the same way though, same models

just now other people can't see them unless you choose to share them

[next trout]

nowhere

[normal abyss]

nah, u made poor elon sad now 😢

[next trout]

he's just so behind

like what even is he doing

all that money and not a good model

[willow swift]

are gpt 5.5/claude opus models in search mode (battle) thinking variants or non thinking ?

[night fiber]

https://www.youtube.com/watch?v=ucK9cX71S1g&list=LL&index=5

[surreal zephyr]

bro gpt 5.5 is better

[tight meadow]

LOL

[obsidian cargo]

probably not better at creative writing though

[tight meadow]

I am creative guys

https://tenor.com/view/spongebob-mocking-imagination-rainbow-gif-15659331

[spring oar]

no lol

is cheapper yes

but not better

[surreal zephyr]

its better at cybersecurity, the only public benchmark shown

while over 20x cheaper

due to much less token usage

and its 5.5 thinking not pro

[storm dust]

what did i just watch

[cedar citrus]

no way

ts one aint right

[cedar citrus]

from my actual testing opus 4.7 performed way better code

https://tenor.com/view/dead-group-chat-waiting-cereal-eating-cereal-dead-gif-25969021

[surreal zephyr]

LMAO

opus 4.7 is one of worst models rn from sota

[cedar citrus]

nah

[surreal zephyr]

3x less efficient while worse than 4.6

[cedar citrus]

i made them both with the same prompt a roblox studio hard script

[surreal zephyr]

sky high sycophancy

4.7 is literally same as 4o

[cedar citrus]

and guess who performed a okay script but way less performance at runtime

gpt 5.5

[surreal zephyr]

skill issue

[cedar citrus]

opus used advanced roblox scripting stuff

[cedar citrus]

nah prompt was great

[next trout]

roblox 0 day

discovered by claude mythos11!1!1!!!

[cedar citrus]

roblox lua scripting is the one thing i am kinda good

so it was easier to evaluate

opus code was great honestly and its not because im a claude fan

i did actually blind review them

[surreal zephyr]

opus writes much worse code

confirmed by opus himself

lol

[cedar citrus]

nah

it its more expensive

that we can agree

but it performs better

[surreal zephyr]

5.4 beats 4.6

[cedar citrus]

opus 4.7

[surreal zephyr]

lmao

[cedar citrus]

i tested gpt 5.5 and opus 4.7 not gpt 5.4 and opus 4.6

[surreal zephyr]

4.6>4.7

[cedar citrus]

nah

[surreal zephyr]

5.5>5.4

[cedar citrus]

i like 4.7 more

okay bro in my opinion opus is better you are a chatgpt fan i respect that and gpt 5.5 is not a bad model is just that opus is better in my opinion

im not gonna do any fights no more i decided to stop doing that

[next trout]

enriqueeeeeeee

[cedar citrus]

https://tenor.com/view/benjammins-idk-this-person-questioning-random-person-this-who-gif-5295285761881556675

[next trout]

we can all just agree that gemini is bad

[surreal zephyr]

you dont even have opus access prolly

[cedar citrus]

i do

[surreal zephyr]

it costs 200$/mo for 10 prompts per day

lol

[cedar citrus]

i have global gpt and i have claude pro

global gpt my dad bought a yearly plan

so i test the models there

[surreal zephyr]

claude pro is not enough for 1 opus prompt 💀

holy liar

[cedar citrus]

hello i use global gpt

what did i just said

[surreal zephyr]

whtas global gpt bro

[cedar citrus]

also stop trying to make a fight here

[surreal zephyr]

bro got scammed

😭

[cedar citrus]

all the models

[surreal zephyr]

holy scam

[cedar citrus]

nah

[surreal zephyr]

xD

[cedar citrus]

i use it

[surreal zephyr]

bro thats not real 5.5

[next trout]

that's just qwen 27b repackaged as gpt5.5

[surreal zephyr]

real 5.5 is on codex

[cedar citrus]

son

it its gpt 5.5

[surreal zephyr]

(still better than slopus)

[cedar citrus]

i know the results

[surreal zephyr]

son they barely released 5.5 on api

💀

😭

[cedar citrus]

they did release?

son

[surreal zephyr]

yeah like day ago

bro

[cedar citrus]

nah

5 days ago

💀

actually even more

7 days

[surreal zephyr]

april 24

[cedar citrus]

today is march 2

[surreal zephyr]

so barely dropped

[cedar citrus]

smart

also why are you trying to cause a fight bro

i dont want to fight]

[surreal zephyr]

im not

[cedar citrus]

im just saying my opinion

[surreal zephyr]

but you are usign scam website

and think

[cedar citrus]

and i respect yours

[surreal zephyr]

that you are using real 5.5

💀

[cedar citrus]

nah its real

[surreal zephyr]

bro real 5.5 is agi

[cedar citrus]

it acts exactly like gpt 5.5

and i tested it in other platforms too

it has seedance 2.0 too

which is the actual model

[surreal zephyr]

seedance 2.0 is free

lol

on some websites

[cedar citrus]

yes

im putting these as temp cuz

[next trout]

that doesn't mean it's actually gpt5.5

[cedar citrus]

i dont want images floating around general

bro

it literally acts exactly like gpt 5.5 and i used to be a chatgpt fan for a long time

until the accident

[next trout]

just because they put oai's logo on it and the system prompt was to act as gpt5.5, it wouldn't be it xD

[cedar citrus]

i used to be a chatgpt fan from the 3.5 turbo era

[next trout]

i'm saying u rlly shouldn't buy from these aggregator "providers"

[surreal zephyr]

they dont evne have proper harness lol

[cedar citrus]

3.5 turbo gpt 4 gpt 4o gpt 4.1 gpt 5 gpt 5.1 gpt 5.2 (i left here)gpt 5.3 codex gpt 5.4 gpt 5.5

[next trout]

because like, how does it make sense to provide 30 models for $20, and like limits matching official providers?

[cedar citrus]

i agree on that but

my dad bought it as a yearly plan and

i tested them so it seems like the actual models

[surreal zephyr]

"claude mythos 5 - 100 prompts per day - 5$ - trust"

[cedar citrus]

nah mythos is expensive

[surreal zephyr]

if you think 5.5 loses to opus

[cedar citrus]

def only for 100 dollar plan and 200 dollar

[surreal zephyr]

then its not real model

[cedar citrus]

then tell me bro

[surreal zephyr]

bro mythos costs LIKE 5000 per REQUEST

[cedar citrus]

where can i test 5.5

[surreal zephyr]

its 200$ PER 1M TOKENS

[cedar citrus]

without paying

and without it being a sketchy website

[surreal zephyr]

you cant, thats the point

[next trout]

i almost choked and died to my death because of how this chat is enriqueeeeee

[surreal zephyr]

chatgpt.com

[cedar citrus]

chatgpt.com doesnt let me choose the model

if i dont pay

[surreal zephyr]

adrian

[surreal zephyr]

then pay

[cedar citrus]

why

[surreal zephyr]

https://giphy.com/gifs/slovakia-communism-socialist-XFaeLrHJGjtcmEmzcn

[cedar citrus]

i left for some reasons the model can be good but the system prompt

is just unbearable

and i tried to change it alot of times with custom instructions

it still kept lying it still kept being annoying

hallucinator

and when i used claude

it was open minded

[surreal zephyr]

bro what

bro

claude is the hallucinator

not gpt

💀

[cedar citrus]

nah

[surreal zephyr]

bro CLAUDE CANT READ ANALOG CLOCK

😭

[cedar citrus]

visual limitations

not hallucination

i do agree

its not good at visuals

[surreal zephyr]

https://giphy.com/gifs/mrw-end-state-7rJE4vH3ItKDu

[cedar citrus]

but i know it will be better in sonnet 4.8

[surreal zephyr]

it cant add 3 numbers

bro

[cedar citrus]

@surreal zephyr i know you are a massive chatgpt fan

and i am a massive claude fan

we arent gonna change each others minds

[next trout]

tung tung tung sahur

[cedar citrus]

we know that

[surreal zephyr]

https://cdn.discordapp.com/attachments/1018470269326729286/1470906684347253038/togif.gif

[cedar citrus]

son i dont want to fight

also

let me give the benchmarks

1. Coding benchmarks — Opus 4.7 wins
   Opus 4.7 takes the coding crown — 64.3% on SWE-bench Pro and 87.6% on SWE-bench Verified. SWE-bench is real GitHub issues in real codebases, not toy problems. LushBinary
2. Overall benchmarks — Opus 4.7 leads
   On the 10 benchmarks both providers report, Opus 4.7 leads on 6 and GPT-5.5 leads on 4. LLM Leaderboard
3. Reasoning — basically tied, Opus slightly ahead
   On GPQA Diamond, Opus 4.7 scores 94.2% vs GPT-5.5's 93.6%. DataCamp
4. Architectural reasoning — Opus wins
   Opus 4.7 performs better on tasks requiring broad architectural reasoning across large codebases — which is exactly what complex Roblox scripting is. MindStudio

[surreal zephyr]

benchmaxxed vs smart (5.2 wins)

[surreal zephyr]

son im crine bro trusts swe bench 😭

[cedar citrus]

opus 4.7

[cedar citrus]

okay what benchmark DO you trust then?

[surreal zephyr]

💀

i trust "not hallucinating adding 3 numbers" benchmark

and actual use

[cedar citrus]

bro those are Opus 4.6 screenshots not 4.7 lol and cherry picking gotcha prompts works on literally every model
GPT-5 fails basic math consistently, social media was flooded with examples right after launch and it still fails even with prompt engineering
GPT-5.2 literally cannot count the number of 1s in 11000 and cannot tell if a bracket sequence is balanced, even with chain of thought thinking enabled
your gotcha game works both ways bro. difference is i gave you actual coding benchmarks where Opus 4.7 wins 6/10 including SWE-bench which uses real github issues in real codebases
find me a GPT-5.5 coding benchmark where it beats Opus 4.7 at complex software engineering. ill wait

[surreal zephyr]

funny how they work on 4.6 but not 5.2

[cedar citrus]

be my wish master

find me a GPT-5.5 coding benchmark where it beats Opus 4.7 at complex software engineering. ill wait

please find it

[cedar citrus]

and i might change my mind depending on the benchmark

[cedar citrus]

source?

[next trout]

mythos mogged

[surreal zephyr]

ok a dare then

i give u two prompts

you put them into opus

if it passes, its as good as 5.5

[cedar citrus]

bro the solid lines are basically tied lol, the dotted lines are best attempt not average
also Mythos is an internal Anthropic model nobody can use, GPT-5.5 is barely edging something that doesnt even exist publicly
and this is one cybersecurity benchmark, not general coding. you cant use a CTF graph to prove GPT is better overall

[surreal zephyr]

(5.5 passed both in 3 minutes, on slow mode)

[cedar citrus]

bet

[surreal zephyr]

but its long so dm

[cedar citrus]

im gonna use global gpt to cuz i dont want to burn my usage

i do agree on the usage part

[surreal zephyr]

scamsite

[cedar citrus]

nah

website with alot of ai models

[next trout]

u know these aggregators?

[cedar citrus]

bro

i use it

you know what i dont even want to argue

[next trout]

u don't rlly know for sure if it's gpt tho

[cedar citrus]

you guys are arguing for no reason i respect yalls opinions

[surreal zephyr]

@cedar citrus i sent u the prompts

[next trout]

https://klipy.com/gifs/cat-laugh-2--k01KQN43QHFFY8NEDNEME63HZ4Y

true

[surreal zephyr]

we'll see

[surreal zephyr]

if he runs the prompts

[next trout]

imagine if it's all gpt 5.4 nano

[surreal zephyr]

are you going to run the test or are u afraid

[cedar citrus]

what test

[surreal zephyr]

well if it fails its not 5.5

[surreal zephyr]

i dmed u

two prompts

OH

LMAO

[cedar citrus]

im doing another test let me explore creativity and coding

[surreal zephyr]

so you are afraid its fake?

[cedar citrus]

i didnt even saw your dm bro

[surreal zephyr]

yeah theres only 5.3-codex and 5.3-instant

[cedar citrus]

gpt output

[next trout]

is that website a MVP guys guys guys

let's vibecode this XD

and route all the models to a cheap model

[cedar citrus]

okay bro i aint even arguing

i just want to discuss

[next trout]

that doesn't prove anything though..

[cedar citrus]

all you guys keep doing is clowning on opus and im not even trying to clown gpt

im completely chill

[surreal zephyr]

thats not 5.3

[next trout]

ur dad got scammed

[cedar citrus]

i do respect gpt 5.5 and i prefer opus in my opinion

[surreal zephyr]

thats 5.3-instant and 5.3-codex

[cedar citrus]

its simple

[next trout]

tell him to refund

[surreal zephyr]

and you still are afraid to check

the proof

i shown u

[cedar citrus]

bro

serious

i am trying to not fight

why are you doing this

[surreal zephyr]

yeah hes 100% ragebaiting

[cedar citrus]

i just dont want to fight in this server anymore

[surreal zephyr]

i gave him proof

he doesnt want to check

[cedar citrus]

I DONT CARE ABOUT YOUR PROOF

[surreal zephyr]

ragebait.

https://giphy.com/gifs/why-IDGNYvFLkJKLK

[cedar citrus]

I LIKE GPT 5.5 AND IN MY OPINION I PREFER OPUS THATS IT

THATS LITERALLY IT

[next trout]

ask it how do u process walk using ActiveProcessLinks

[cedar citrus]

ITS SIMPLE SO LET ME IN PEACE

i remember

it was the great times

[surreal zephyr]

another delusional claude glazer refusing to actually compare both while spreading misinformation it is :/

[cedar citrus]

son

okay bro you just trying to make me mad on purpuse

[surreal zephyr]

he wont ask anything, he knows hes scammed

he is probably advertising the scam site

[cedar citrus]

im actually getting angry now

i dont even want to fight bro

please stop

i mean i dont really want to fight

[surreal zephyr]

he's glazing a scam site

lol

[next trout]

@cedar citrus prompt it this, u'll know (it's a complex question) if it's actually gpt5.5 or a weak model: how do you EPROCESS walk using ActiveProcessLinks and PsInitialSystemProcess?

[cedar citrus]

ampro keeps making me angry

[cedar citrus]

i will do it

im sure it will respond good

gpt 5.5 is a good model

[surreal zephyr]

i sent him two prompts that actually check exactly what model it is

and he is scared to do them

lol

[next trout]

what were they?

[surreal zephyr]

two math tasks

5.5 does it in 3 minutes per task

opus 4.7 takes 30 mins and fails

[cedar citrus]

in math 5.5 is better

[surreal zephyr]

gemini fails too, after 12 mins

[cedar citrus]

hey

im dming the results

[surreal zephyr]

he wont even run it because he knows he doesnt have real 5.5 in that scam site

[cedar citrus]

also it was a good response

[cedar citrus]

i did run and im dming him the results

sorry her

im not english native

[next trout]

send the screenshot here

[plucky fractal]

literally this

[silent tree]

whats happening

[cedar citrus]

well friend

i have no idea

im trying to be nice and respect that gpt 5.5 is a good model but in my opinion i prefer opus but

[silent tree]

whats happening and rate the pfp @light sleet gave it to me

uhh

[plucky fractal]

that's not a watermelon bro

[silent tree]

IDGAF ITS MY NAME

[next trout]

is that sam altman

[silent tree]

YES

[plucky fractal]

bro why u jumscare me like that

use spoilers or something

[silent tree]

idc openai better than anthropic

[silent tree]

claude mythos isnt even real

lol

"myth" os

[next trout]

ok but why is he mogging us

[silent tree]

Idk honestly

[surreal zephyr]

W

[silent tree]

@light sleet generated these lol

[silent tree]

I js got u in my pfp

[silent tree]

thats not my sam 😡

thats a deepfake

fake fake

fake

[surreal zephyr]

real

[silent tree]

sam has no double chin

[surreal zephyr]

can confirm

i was there

[silent tree]

so fake

[surreal zephyr]

i was the phone

sam has trimplechin

[silent tree]

ok ill just admit it...

I'm the real sam altman

ampro isnt

[silent tree]

@light sleet generated this too

[next trout]

thomas shelbek

[light sleet]

I knew you were the Bay Harbor Sam, Altman.

[next trout]

sam is going to give us all free gpt spud

ok

he loves us

[light sleet]

rip anthropic

[minor pulsar]

I AM INCREDIBLY LUCKY. (IT WAS ALL IN A ROW). THX ARENA

[silent tree]

Dw I got u

[next trout]

needs to take off his glasses and give us a wink ok

[silent tree]

gpt always ahead of claude 😔

[silent tree]

dude im sam and im not giving anyone a wink

I can give u spud

[minor pulsar]

Claude opus 4.7 think...wow....Very...

[silent tree]

gpt 5.5 still mogged it

sorry

I dont care what u anthropics say

im always staying with openai

cuz they never giving us up

[light sleet]

truest words I heard today

[next trout]

ok gib

or scam altman

[cedar citrus]

cant stay in this server full of gpt users i do respect but doesnt respect us back 😔

[minor pulsar]

gpt 5.5 ..

in coding opus 4.7 is much stronger... This is important for me

[cedar citrus]

i agree with you

but they

will say alot of bad stuff

[next trout]

what is elon and sam fighting over

[silent tree]

or nvm heres ur winky wink

[silent tree]

elon trying to sue openai

cuz he wanted openai to be non profit ig

[cedar citrus]

please bro you dont want to fight with them they will say everything and not be chill

[next trout]

how is he going to win it xD

[minor pulsar]

Opus writing 5900 Rows in 1 message...WOW...It working at 100%

[silent tree]

why grok also hiding everything behind paywall too now 😭

[next trout]

he's left it for like 5 years

[silent tree]

Grok Imagine is supergrok

[silent tree]

jealoussssyyy

[next trout]

true

[silent tree]

And auto is supergrok now too

like WHAT

[next trout]

he's in the sneako chair that's why

[cedar citrus]

thanks god ampro isnt here he would annoy us alot

[silent tree]

is grok even usable anymore without paying

[silent tree]

@surreal zephyr

[next trout]

just use it on arena

[silent tree]

Prove it

[cedar citrus]

no bro

[silent tree]

yall have no proof

[silent tree]

I want proof

of opus being better than gpt 5.5

I need both results

[next trout]

i created a deobfuscator with gpt5.5 xd

[cedar citrus]

he will annoy me with his opinions

[silent tree]

opus 4.7 thinking and 5.5 xhigh

I need both results

lol

[surreal zephyr]

(and doesnt ocmpile)

[cedar citrus]

and i respect his opinionsd

OH NO

[surreal zephyr]

i gave him 2 benchmark prompts

he refuses to try them

hes ragebaiting

[cedar citrus]

@minor pulsar you need to leave before they annoy you too much

[light sleet]

You anthropics check this out by ampro https://goldime.netlify.app/

[cedar citrus]

@minor pulsar listen to me get out i like your opinion but they dont

[silent tree]

I saw it

far better than Opus 4.7

[surreal zephyr]

yeah 5.5 + skill = peak

[cedar citrus]

@minor pulsar i cant protect you for longer

[silent tree]

Lmao

[silent tree]

True

[silent tree]

HES NOT EVEN TYPING BRO

this isnt a movie 😔 💔

[surreal zephyr]

see this bro

[cedar citrus]

yeah unless i get muted for one hour

[surreal zephyr]

💀

and he sent a scam website

[minor pulsar]

why leave? This is their opinion. For me, an opus is better, for him an open ai. He just said this

[surreal zephyr]

😭

[cedar citrus]

i didnt send a scam website i literally use it

i just mentioned it

[silent tree]

😭

[silent tree]

see he knows ball

[cedar citrus]

this is literally what i been saying to ampro for 20 MINUTES

[silent tree]

but ampro is right

[cedar citrus]

GPT 5.5 IS GREAT OPUS 4.7 IS GREAT OKAY

[silent tree]

gpt is better than opus

[surreal zephyr]

no opus is bad

[cedar citrus]

I RESPECT YOUR OPINION

[silent tree]

gpt 5.5 is greater*

[minor pulsar]

Why am I being kicked out? I just praised the arena...

[surreal zephyr]

and you refuse to look at proof

[cedar citrus]

I RESPECT YOUR OPINION

ITS SIMPLE

[silent tree]

no ones kicking u

he asked u to stay out of the conversation

[next trout]

xD

[cedar citrus]

okay bro thats the problem

[surreal zephyr]

im losing my braincells here

😭

[next trout]

u guys must have been on the island

[minor pulsar]

That's it, I'm silent... Sorry

[cedar citrus]

dont be silent

your my ally

your a friend

you deserve to be at the community

we must respect each others opinions

[silent tree]

same guy who asked him to be silent and leave

[cedar citrus]

this isnt a anthropic vs open ai fighting arena

[silent tree]

elite ball knowledge

[next trout]

ur going to be sent to the island

[minor pulsar]

Thanks bro

[cedar citrus]

BECAUSE YOU GUYS WILL BE FIGHTING WITH HIM

IM TRYING TO KEEP PEACE HERE OKAY

[silent tree]

make one decision

[cedar citrus]

JUST DONT FIGHT

[silent tree]

tell him to be silent or not silent

[cedar citrus]

IM ASKING JUST THAT

[silent tree]

so confusing

[cedar citrus]

DONT BE SILENT AND YOU GUYS STOP FIGHTING OVER OPEN AI AND ANTHROPIC

[silent tree]

bro

[silent tree]

wasnt that u

whos you guys

[cedar citrus]

everyone

[silent tree]

me and ampro fighting openai vs anthropic?

what

but we both

are openai

ppl

[next trout]

https://tenor.com/view/cat-smile-happy-happy-cat-smile-cat-gif-18259408098481512759

[silent tree]

+1 ragebait

[surreal zephyr]

lol

[silent tree]

bros watching a nice movie

[cedar citrus]

me trying to respect your opinions and you guys filling the chat with openai openai i get it okay its good model but i like anthropic thats it

[silent tree]

so whats your problem with it

[cedar citrus]

I DONT HAVE

[silent tree]

weren't we on respecting opinions 💀

let us say what we want

[cedar citrus]

AMPRO KEEPS IRRITANTING ME

[silent tree]

u dont control us

do you?

[surreal zephyr]

• uses&advertises scam site
• claude glazer
  yeah sure

[cedar citrus]

IM NOT ADVERTIZING A SCAM WEBSITE

[surreal zephyr]

discord is 13+

[cedar citrus]

I AM NPT

[cedar citrus]

I AM 13

BRO

[silent tree]

but you are advertising

[surreal zephyr]

you did

[cedar citrus]

SON

[silent tree]

advertising is

[cedar citrus]

I AM NOT

[silent tree]

against

the rules

Of arena

[cedar citrus]

IM LITERALLY NOT

[surreal zephyr]

@silent tree thats why

[silent tree]

pineapple said it himself

[cedar citrus]

13+

[silent tree]

Lol

[next trout]

u guys are so evil XD

[cedar citrus]

its 13 years and older

[surreal zephyr]

he's ragebaiting

[cedar citrus]

??

[surreal zephyr]

lol

[silent tree]

he kinda went against arena rules

[cedar citrus]

IM NOT okay bro im leaving

[silent tree]

advertising a website is still against the rules here

[cedar citrus]

i just mentioned i use the website thats it

[surreal zephyr]

especially that it was obvious scam

[cedar citrus]

i didnt say ohhh guys use this its the best

[silent tree]

ok mention it somewhere else

[cedar citrus]

i just said that i use this and thats it

[silent tree]

all webs offering free frontiers models are a scam

[surreal zephyr]

he thinks he gets unlimited opus 4.7 and gpt 5.5 for 20$ @silent tree

[cedar citrus]

SAYING THAT I USE IT DOESNT MEAN IM ADVERTISING IT

[cedar citrus]

I DIDNT SAY THAT

[silent tree]

💀

[surreal zephyr]

and he thinks its legit

😭

[silent tree]

💀

[cedar citrus]

I LITERALLY DIDNT SAY THAT

[surreal zephyr]

u sent the link

to the scamsite

[cedar citrus]

I ONLY SAID THS

thats it

im leaving this

goodbye

[silent tree]

he did? oh its against the rules then

pineapple said it himself lmao

[surreal zephyr]

oh lol

[silent tree]

@median smelt mentioned

[surreal zephyr]

a mod deleted it so i guess he isnt getting banned for that scam link

[light sleet]

LMAO 💀

[surreal zephyr]

craz

[silent tree]

might be spicy fried chicken

so uhh

I guess gpt wins again

[surreal zephyr]

bro it says unlimited opus 4.7 and 5.5 for 25 / mo

holy scam

😭

[silent tree]

💀

obvious scam

[surreal zephyr]

yeah

he used this

and says 5.5 is dumb

[silent tree]

so he basically even advertised a subscription

[surreal zephyr]

and wants to be taken seriously

bro

[silent tree]

LOL

I'd rather use official chatgpt

then pay for some scam

[surreal zephyr]

who doesnt

[silent tree]

thats obviously not real

[surreal zephyr]

this best

[silent tree]

anthropic user

[next trout]

https://imgur.com/p4UgNts

[silent tree]

these the people we dealing with btw

[light sleet]

This was a damn nice movie

[surreal zephyr]

https://cdn.discordapp.com/attachments/1018470269326729286/1470906684347253038/togif.gif

[next trout]

post more of sam altman mogging dario

[silent tree]

u got a crush on him?

@surreal zephyr ...

[surreal zephyr]

honestly sam is evil

tibo is better

tibo is the king

[silent tree]

yes

what do u think about elon suing

[surreal zephyr]

sam wasnt even active before the sue

[silent tree]

openai

[surreal zephyr]

elon's right

imo

openai scammed elon

[silent tree]

sam doesnt talk in the discord 😔

[surreal zephyr]

and elon wants openai to donate to open source, instead of himself

[surreal zephyr]

he talks on x

(or rather his pr team)

[silent tree]

I wanna talk to sam personally

and tell him

to collab with x ai

[surreal zephyr]

https://giphy.com/gifs/punkedrocker-donut-glaze-the-level-of-glazing-rn-yTGU5g1oGJqtZDmJfx

[silent tree]

but he wont

[next trout]

real agarthan expert

[surreal zephyr]

literally this

[silent tree]

what if xai and openai collabed

[surreal zephyr]

grok's thruth seeking + openai's skill

easy win

[silent tree]

fr

and groks search

[surreal zephyr]

and grok has a lot of compute

and openai has efficient models

[silent tree]

yep

together they make the biggest model

Spudrok

[surreal zephyr]

groktato

[silent tree]

Gropud

[next trout]

name sounds regarded

[silent tree]

💀

[surreal zephyr]

💀

[silent tree]

I just have that feeling

of claude being discontinued

in the next few years it definitely will

anthropic cant keep up with openai

[next trout]

the juice is going to take away all the AI from us

[silent tree]

anthropic is so far away from openai btw

5.5 xhigh mogs

@surreal zephyr imagine the day elon and sam posts this

[surreal zephyr]

calude is dying

claude has no compute

they even paused account creation for a day

[silent tree]

you didnt have to ask AI to write that

😭

im not reading allat

still AI

Gemini

[next trout]

it's AI slop

XD

[surreal zephyr]

openai PROFIT (not revenue) in 2026 is 25 BILLION

openai mogs again

[silent tree]

Fr

[next trout]

it has the government.

[plucky fractal]

https://x.com/i/status/2050259511068201403

Tannor Manson (@Futurenvesting)

Anthropic is now showing off $44 BILLION in annual recurring revenue.

This is up $14 billion (+46.6%) since last month!

BULLISH for AI Infrastructure $NVDA $AMD

X · May 1, 2026, 5:02 PM

"claude is dying"

[next trout]

misanthropic

[plucky fractal]

block the trolls lmao.

[silent tree]

whats the matter

cant keep up

like how anthropic cant

LMAO

[next trout]

u guys do realize that dario would just take all the models away from u like he's doing with mythos

[surreal zephyr]

no they are literally profitable as hell

[silent tree]

watch them say "you worship sam altman"

[surreal zephyr]

gpt 5.5 doubled their api profit

[silent tree]

cuz they obviously will do

[surreal zephyr]

because it costs 2x on api while compute is cheaper

[silent tree]

mythos isnt even real

dont mention mythos

the names right too (Myth)

alright stop posting jewish stuff

[surreal zephyr]

wth was that

[silent tree]

<@&1349916362595635286>

[next trout]

<@&1349916362595635286>

[silent tree]

sorry bro

the big e is banned here

[next trout]

it's triple t

[silent tree]

💀

[surreal zephyr]

this u

[silent tree]

the big e is there

soo

[next trout]

fax

[surreal zephyr]

holy name

[next trout]

no big e

[silent tree]

such a weirdo

he likes to post big e jokes

[next trout]

stop agarthaposting

[surreal zephyr]

why claude users are the most, uhh, "unique"

[silent tree]

he js has no words

thats why