general

[silent tree]

I gave to myself tho

omg I got champion

yay

[graceful sonnet]

I think that nano banana models and gpt-image-2 models can bring back, but use credit limits, and refreshes every 12 hours, and everyone gets 100 credit limits starting, and donate for more credits

[deep adder]

@silent tree why did u leave 😿

[light sleet]

Welcome to the club

[silent tree]

lets not talk about that

[deep adder]

welcome to the Claude Opus fanclub

[silent tree]

GPT Club

[autumn stream]

How to delete the conversation in the agent mode? no having dot to archived

[silent tree]

https://static.klipy.com/ii/39f2394ae36df6e199be9eb7c9fa1012/a8/0d/rBcSATBg.gif

[deep adder]

Apple Intelligence is releasing so soon

gonna be sick

[compact flame]

I'm back guys

I heard situation got handled

[thorny current]

Prob gonna coast 1 API/kidney

[deep adder]

will have a 1.2T param gemini model

[deep adder]

free

[thorny current]

Really?

[deep adder]

they prob have 3.5 Flash

yeah they bought the weights

[thorny current]

Oh

Wow

[deep adder]

3.5 flash is prob 1.2T total params

[thorny current]

Yeah

[deep adder]

thats the best model they could use for Apple Intelligence ngl

[thorny current]

V4 pro thinking of deepseek is 1.6T

[deep adder]

not as good though

also not multimodal

[thorny current]

Opus 4.7 was over a T too i guess

[thorny current]

Yup

Good for thinking maths

[deep adder]

https://genai.apple.com/

[thorny current]

Didnt loaded
Vpn didnt worked either

[north marsh]

what did i miss can anyone give me a quick recap?

[limpid pumice]

Is the bug fixed now???

[echo aurora]

The team has made some progress, but it doesn't appear to be fully fixed yet. cc @limpid pumice

[unreal hatch]

@echo aurora when will be available most capable (aka - Opus 4.7, ChatGpt-5.5...) models in arena in Direct Chat?

[proud bobcat]

Qwen 3.7 seems goated

[strange tusk]

i need to edit people to make them like kiss together but i need the faces to stay the same which model is the best for this

[limpid pumice]

Woohoo I have ascended😇

And sleep deprived

Bye ya'll I'm gonna crash

[loud herald]

It’s crazy because I wouldn’t have picked qwen to be the first to actually rival opus 4.6 in front end

[light sleet]

Because it is

[loud herald]

Indeed

[echo aurora]

Yeah #4 is super good

[north marsh]

wait api is finally added?

[silent tree]

Pineapple can you check dm

[loud herald]

#1 DS & Qwen Glazer. Rebrand

[echo aurora]

WIll do, really busy morning but willt ry to get to everyone when I can

[silent tree]

Kk

[coral birch]

Hey, in the past week has overall generation speed been getting faster for you guys?

[loud herald]

It seems normal to me

[coral birch]

It seem to be faster for me, I don't have to refresh the page even

[loud herald]

And I still have no idea what this bug is lol

The site was fine when I used it

[frosty lava]

so which is the best model for frontend only frontend matter

[proud bobcat]

Evolution time

[coral birch]

same 😭 I have been able to access Arena for the past month with no issues at all (except the common one,but well, a refresh or a new chat solve it all)

[loud herald]

Lmaoo

[deep adder]

did gemini spark release yet?

[proud bobcat]

Peak

[loud herald]

Still opus 4.7 but not by a wide margin

[north marsh]

is api added to arena?

[frosty lava]

which one is the top 2

[loud herald]

I haven’t seen. What’s its stealth name?

[deep adder]

its a product

[loud herald]

Qwen 3.7 max

[loud herald]

Ahh

[frosty lava]

Alright thanks

[loud herald]

👍

[pine meadow]

https://019e6524-978d-7b8f-bfe9-6a48ee31246d.arena.site/

[deep adder]

micron up 20%

[loud herald]

That’s actually really cool LOL

What model did you use?

[deep adder]

china is so behind lol

Huawei said theyd make 1.4nm chips in 2031 LOL

intel had since 2024

🤣 🤣 🤣 🤣 🤣 🤣 🤣

[loud herald]

China is also making these on worse equipment than any US company

There’s a reason China has blocked Nvidia chips now

Now that they’re actually able to buy

[deep adder]

that means nothing 🤣

admitted to being behind by 7+ years

genuinely hilarious IMO

[loud herald]

It’s because they believe their hardware is better or a better over time

[loud herald]

Sure, they are starting from scratch ofc they will be behind years

[deep adder]

will never beat the US in AI

its like a joke

[loud herald]

They will

[deep adder]

lol

[loud herald]

Maybe not in full intelligence

[deep adder]

they still train on nvidia btw

and nvidia chips wont be sold

[loud herald]

But US companies will go bankrupt before any Chinese company

[slow lotus]

'Something went wrong' error

[deep adder]

uhh

[loud herald]

Refresh

[slow lotus]

tysm Lmarena

[pine meadow]

Max

[loud herald]

Deepseek v4 was trained on hauwei chips

[deep adder]

no it wasnt

[loud herald]

Ah

[loud herald]

It was literally said that they were

[deep adder]

do you believe tha

that

[loud herald]

It was said before it released but it was Nvidia chips nvm

Also, Nvidia won’t be the way for much longer either way

I think AMD will be

[slow lotus]

worked

[loud herald]

Nice

[deep adder]

its not like nvidia "wont be the way"

[slow lotus]

btw daily limit problem is fixed?

[deep adder]

there will just be more competition from AMD and intel

[loud herald]

Nope

[slow lotus]

is it still 15 or 9?

ohh no

[loud herald]

Yea 100% but I think AMD will be better to train on

[deep adder]

why

nvda is so entrenched

[loud herald]

A 1B parameter model surfaced recently that was trained on AMD and was the best in the 1B and 2B categories

[slow lotus]

what is the daily limit amount

[pine meadow]

https://019e647f-0cc2-706e-9de4-7ca2062ed62d.arena.site/

[loud herald]

You could say datasets but I think hardware has a big reason there

[loud herald]

I have no clue honestly

[slow lotus]

I even heard its 100

100 or 15 or even 9

[loud herald]

I just don’t use arena enough to know

[slow lotus]

oh k

[pine meadow]

Great Max model

[north marsh]

will ever arena be a subscription based platform

[pine meadow]

I also used it to create a PS1 emulator.

[pine meadow]

noooooo

It's an alternative for poor people like me

[north marsh]

or ppl that cant pay for ai model(have the money but cant)

[pine meadow]

It's an alternative for poor people like me

[loud herald]

Absolutely not. Their entire brand is on being free

[pine meadow]

try my ps1 system em

[loud herald]

So tell me if I’m wrong but I think the Roaring_waters 1 and 2 is Deepseek 4.1

Stealth name would make sense lol

But also is insane at front end

I had already picked it over opus 4.6

[deep adder]

GLM i think

[loud herald]

I know it’s an earlier generation but it’s not much of a downgrade from 4.7

[loud herald]

That would really make sense too

[deep adder]

deepseek is only focused on multimodality rn im pretty sure

thats the main thing v4 pro is missing

[loud herald]

Ahh

[frosty lava]

i believe that at some point a qwen or glm model will be better than all opus model at frontend

would be cool

only frontend

[deep adder]

nah

because they all distill from opus

[loud herald]

Chinese models are unpredictable so I really couldn’t guess which one would be

[frosty lava]

they train too from other thing it cant be 100% distillation

but yeah

[deep adder]

i mean opus is the trainer model

[frosty lava]

look right now honestly its really really close

in frontend

[loud herald]

You can distill from already made datasets

[deep adder]

yh but opus is much more agentic

qwen is quite agentic though so

[loud herald]

Absolutely

[frosty lava]

yeah but was talking about frontend

[deep adder]

its always been "close"

nah frontend isnt close

[frosty lava]

its closer than it used to be

[pine meadow]

try my ps1 system em

[deep adder]

opus is the clear best lol

[loud herald]

The thing that opus has always had that’s different from everyone is its personality

[deep adder]

also like

theres one thing

[frosty lava]

honestly not by much at frontend

[loud herald]

Anthropic tries to make it sound as human as possible l

[deep adder]

zero shot frontend vs editing frontend

[loud herald]

Alright

[deep adder]

its very different when actually using these tools in a codebase

[loud herald]

True that

Always good to check your own work with these models and not go off of someone else’s opinion

[frosty lava]

x ai will soon release a new big model too then in few month a 5t one

now its 1.5 t

that will be release

[loud herald]

Gotta hope it will be good. They have been improving grok for a while now

Very quickly too

[deep adder]

it wont be good

[frosty lava]

let's see

[deep adder]

grok is so far behind

they dont have anything

its now just anthropic vs openai

[frosty lava]

right now its honestly a small model that is being used

[deep adder]

i dont have google out but they need to get their head in the game rn

[frosty lava]

nothing you can compare to a 5t one

[deep adder]

they arent traning a 5T one rn

[frosty lava]

they are

they announced it

[deep adder]

it probably broke

the training run

[frosty lava]

but at first they will release the 1.5 t

very soon

[deep adder]

also xai doesnt have good enough data

elon even said evals look "good" not the best

grok 4 was the best in a lot metrics

now they lost everything

[frosty lava]

I mean it cost nothing to just wait and see, all i say is that the next model will be much larger

and its a fact

[deep adder]

btw

grok 3 was 3T

🤣

so yeah

@frosty lava

[frosty lava]

i didn't know this

but then why they say their model right now is smaller ?

even smaller than 1.5t

[deep adder]

efficiency

[toxic verge]

Xai will buy cursor

[frosty lava]

yeah so they trained a completely new one from scratch ?

[deep adder]

true

[toxic verge]

https://www.forbes.com/sites/sandycarter/2026/04/23/spacex-bets-60-billion-on-cursor-ai-the-real-winner-is-a-surprise/

3rd best coding model right now

[deep adder]

nah it isnt the 3rd best lmao

look at usage

[toxic verge]

It just came out

[deep adder]

its uhh benchmaxxed?

also no one uses cursor... it's a terrible product

[whole sundial]

<@&1349916362595635286>

[fair hedge]

i think rate limit is now removed

i checked and i don't find any thing related to RateLimit in console

[echo aurora]

What makes you say that?

[fair hedge]

@echo aurora now you tell the real truth

[echo aurora]

?

[fair hedge]

limit removed or still there

as per testing only i don't want to waste my limit so i just check logs

[thin grove]

Hi

[deep adder]

@surreal zephyr does codex caching not work at all?

stop pinging the guy

[blazing hill]

[echo aurora]

Limits were not removed

[fair hedge]

just want to know ans

[deep adder]

obv they didnt remove rate limits

[clever glade]

secret gemini 3.1 pro imagegen technique

[south beacon]

DeepseekV4 pro is the best AI... because its context window is absurdly cohernet and seems to go on forever.

no context rot

[thin grove]

No it aint

its 1 million context

[dapper quest]

please i beg you to put gpt 2 image and gemini again in direct mode

[sleek phoenix]

why is the image stretched lol

[thin grove]

costs to much

[short sluice]

why do i feel like opus is coming back next month

[thin grove]

cuz it is since credit system

[dapper quest]

life was so good when we had it

[limber hound]

doesn't v4 pro use Engram?

aww

[strange tusk]

is there like a option to buy premium to fix this?

[thin grove]

nope

[pine meadow]

nope

[strange tusk]

i only gave it 1 prompt and already limit

[limber hound]

do you use like 8k context model of something

😭

[strange tusk]

i used sonet 4.6 to make me a game

[pine meadow]

Use battle mode

[clever glade]

use qwen 3.7 max

[strange tusk]

i need a game with 3d models sonet was good for that haha

[limber hound]

-# don't make games in one prompt on arena. or at least use agent mode

[pine meadow]

Claude Sonnet is under a lot of pressure; try using another model.

[strange tusk]

i think agent mode is even gonna hit the limit faster haha

[south beacon]

You are right, its 1M. But it still feels like 2M and never gets context rot or slowdown.

Just remember. When they reach AGI, you're not invited to the party.

[pine meadow]

Everything will be paid, and that's frustrating for poor people like me who rely on free models.

America will be the first to reach it, but I think China has the ability to reach it after America.

[modern wedge]

YEAAAAAAAAAAH

[thin grove]

41 tool calls

[inland pilot]

Guys, what is the best model in the field of photo generation in Lmarena right now?

[thin grove]

gpt 2

[inland pilot]

How to choose it?

[thin grove]

you cnat in direct chat

only in battle

[cursive creek]

lesgooooo

[inland pilot]

how to summon him then\

[thin grove]

You need to have luck

[worn jetty]

@echo aurora
Thanks bro ❤️!
Take care!!!

[thin grove]

what

[echo aurora]

?stickstop

Okay, stickymessage gone!

[worn jetty]

Nothing arena just got fixed!

[inland pilot]

When will Gemini 3 Pro Image Preview 2K be returned?😭😭

[south beacon]

But it doesnt add up. If AGI is paid, then fine, but if it refuses the prompt "go on the internet and make me X amount of money " then why pay for it. If it does do that then everyone can subscribe for a day, make money and it pays for itself perpetually.

[thin grove]

when will we get the credit system

[sleek urchin]

THANK YOU

[echo aurora]

We don't have an estimated timeframe for that, or if it'll happen either. We'll have to wait for the new usage to see.

[thin grove]

im excited

[echo aurora]

Don't have an ETA

[south beacon]

AGI becomes the UBI.

[thin grove]

Alright, is it already on canaryarena.ai or nah

[feral elm]

uhm

[south beacon]

i just solved UBI

[feral elm]

I got bamboozled by gpt
or the UI

[silent tree]

So btw (following the DM) I can make it on my own?

[feral elm]

either way
bamboozled

[fair hedge]

my arena Account got deleted by my mistake can you please help

[echo aurora]

Nope

[worn jetty]

@echo aurora
Can u do something to this?
I mean it's my main email I used to chat?..
It's giving same...
Or ur fixing something!!!

[echo aurora]

Doesn't look like you're logged in.

[worn jetty]

I am logged in

[thin grove]

<no

[worn jetty]

Should I send u screen record

[feral elm]

hey pineapple
I never understood if arena.ai pays for inference of if it's provided/financed thanks/by the conversations/battles

[worn jetty]

Pineapple?

[echo aurora]

Can you use the #1343291835845578853 channel?

[worn jetty]

Okay 👌

[thin grove]

Maybe soon we will get Arena CLI

[echo aurora]

Find a similar post and share your problem there.

[feral elm]

🔥

[fair hedge]

deleted account can be recover?

if yes please help me

[civic plaza]

If on the website, sadly not

[echo aurora]

Sorry to say if an account is deleted there isn't a feature that'll un-delete it.

[worn jetty]

When arena was affected by bug, I deleted this email but now if ur saying arena is now fixed...
After I login to by main email...
It show same as I can't login in..
(I mean same login page)

[feral elm]

wtf does one do to get it deleted 😭

[fair hedge]

but please try to add this feature in future so it can be un-deleted

[feral elm]

lol

[echo aurora]

I deleted this email
This is likely being caused by deleting your account.

[worn jetty]

Whaaaattt

[echo aurora]

It may happen, I can't say I'm aware of this idea being planned in the short term.

[worn jetty]

No, I mean I am using 2nd email, I even deleted it but I can still login in as fresh account!!!
Please don't say that u delete me?🥀

[vague solar]

whats the difference between qwen 3.7 max preview and qwen 3.7 max

[echo aurora]

Lets take this conversation to #1343291835845578853. I'd like to avoid troubleshooting in #general

[worn jetty]

Apologies brother

[echo aurora]

It's no problem! I am sorry you're getting a login bug.

[worn jetty]

I created a thread🧵, I wish 🙏 u will help a little!

[gloomy mason]

is that problem with blank answers after 5-6 prompts fixed?

[worn jetty]

Use qwen3.7 max
Preview is the initial model for testing!
Qwen 3.7 max is the final product!

[echo aurora]

Yes, we rolled out a fix about an hour ago

[gloomy mason]

hooray

[echo aurora]

after 5-6 prompts
Would note the bug we were seeing wasn't just after 5-6 prompts, we were seeing reports on the first turn.

[gloomy mason]

what was the problem? or it was a test of a new credit system?

[echo aurora]

Hey @median ember I'm going to delete that message as it has your email in it, we wouldn't want others to have access to that here. Would note that some temporary email addresses are not usable, this is why you're getting that error.

[median ember]

I registered with my email address a year ago. Now it won't let me log in because it's a temporary email address.

[echo aurora]

Yeah unfortuantely, that email is no longer going to work.

[median ember]

If I give you my current Gmail address, could you transfer my old account there?

[echo aurora]

I won't be going into the specifics, but it took us awhile to figure it out. The team is going to have a post modern to look at why this took so long to spot the fix.

[echo aurora]

Sorry to say we don't have a feature that'd be able to support this.

[median ember]

Thank you for everything. I'm currently logged in with Gmail and I'm very happy with the current situation. I'd like to thank Arena AI and your excellent support team.

[echo aurora]

Thank you, I really appreciate that.

[silent tree]

@echo aurora Advertising

[south beacon]

But it doesnt add up. If AGI is paid, then fine, but if it refuses the prompt "go on the internet and make me X amount of money " then why pay for it. If it does do that then everyone can subscribe for a day, make money and it pays for itself perpetually. AGI becomes the UBI. i just solved UBI

[feral elm]

brah
AGI ain't gonna work like that

you should be able to figure that out yourself
if you can't
ask Gemini or gpt or Claude to explain the fallacies that are going on in there

[pine nest]

Why are Gemini and Chat GPT no longer available for images?

[feral elm]

made a thread instead

[lunar glade]

Whenever using Qwen 2.0 Pro 2026 4 22 model, it always come with this error Something went wrong while generating the response. Please try again. Trace ID: :9485511e-d2ba-, only this model having trouble

[normal summit]

same with claude etc

[south beacon]

yea i already knew that really, i was being ironical. Because what i actually expect to happen is that it wont accept the prompt, and AGI wont deliver the value it demands in the subscription pricing.

Then people will take their money elsewhere, where its likely to generate more wealth, and the AI industry would collapse under its own weight.

[lunar glade]

I dont have trouble using others, for me right here, just the qwen

[south beacon]

because they are fedup with paying billions of dollars for random kids to generate images of pokeymon on thier enterprise grade GPUS.

[amber oxide]

Wait the message is gone?

[random prairie]

what?

[amber oxide]

💀

https://tenor.com/view/ah-shit-here-we-go-again-gta-sa-cj-gta-again-gif-20425950

[whole sundial]

works fine for me

[cedar tide]

When agent mode in battle mode ?

[normal summit]

mine stuck here
after that tracking message

[unkempt garnet]

I just wanna say thank you devs for fixing the site, no matter how much we complained lol

unless it's still not working

[whole sundial]

i just realized that the removed image gen models can still be used using max (removed models are never supposed to affect max per pineapple)

[whole sundial]

this is image 2 btw, image 1.5 would look similar but a bit worse while the image 1 models would look completely different

[sleek urchin]

Its not devs fault for creating a buggy bug, its the provider error

[unkempt garnet]

Ok

[strange tusk]

is it right that i cant upload php files

[wicked sage]

2025 🙏

[half veldt]

Does mai not support image edit?

[echo aurora]

Just Text-to-Image

[loud herald]

Man it’s crazy seeing the giant leap from nano banana and gpt image

GPT image is very good but I don’t think it’s that insanely better

[loud herald]

Unfortunate

[echo aurora]

Checking with the team on it though

[compact flame]

I'm surprised qwen managed to beat opus 4.6 non thinking

[loud herald]

Very nice, thanks!

[compact flame]

Qwen got a high potential

In the future

[loud herald]

That’s what I was saying, out of all of the Chinese models I wouldn’t have guessed qwen

[pine meadow]

I think Alibaba has abandoned its main business of running its online store.😂

[pine meadow]

In my opinion, it can achieve even more than this and surpass Opus 4.6 Thinking Mode and GPT 5.4 Pro.

[wicked sapphire]

what do u think is different about it

[pine meadow]

who

[pseudo hemlock]

Not again

[echo aurora]

Got em

[clear vault]

Why I can't make unlimited videos

[polar niche]

Hi

[echo aurora]

We have reasonable rate limits in place for sustainability purposes. Offering unlimited free AI video generation isn't something we can feasibly support.

[shrewd citrus]

I love this diagram but I would love to see another line showing the average across maybe the top 5 models?

since the diagram basically shows all the info you need in such a simple way

[pine meadow]

https://019e65ea-f4b0-7e90-83b3-38ec999f4f5e.arena.site/

[shrewd citrus]

so having another line which shows the average performance of the top 5 or 3 models would be a great comparison

[echo aurora]

This is a great idea! I'll share with the team.

[slow lotus]

Is the bug fixed completely?

[echo aurora]

Yes

[slow lotus]

yayy

btw

[slow lotus]

what is the daily limit amount

for text

[limber hound]

1800 currently

[slow lotus]

Oh

My God

I can send 1800 replies to AI daily?

[limber hound]

🤷‍♂️

[echo aurora]

We don't have this limit currently displayed. Part of the new usage system will be to include this, so more information to come!

[limber hound]

i think you'll hit captcha loop sooner than send 1800 requests

[echo aurora]

https://klipy.com/gifs/will-ferrell-elf-17

[slow lotus]

Oooh allright

[slow lotus]

Btw why rechapta repeats it too much

[limber hound]

-# noooo

[slow lotus]

especially gemini

👏 to LmArena devs AAAA

Wth did u send just now

Yo guys which model do u recommend for writing and roleplaying

[crystal mica]

claude sonnet 4.6 dont work

[echo aurora]

Why not?

[polar niche]

When's the cli coming?

[crystal mica]

just error

[slow lotus]

.

[polar niche]

Arena CLI

[echo aurora]

If you can share the Trace ID to #1417174113092374689 that'd be best

[echo aurora]

Sorry to say no ETA I'd be able to share.

[polar niche]

But through the API you could do the same?

[slow lotus]

Which model do u recommend for writing, launguage and roleplaying

I tried gemini but its ehh after long sessions

[echo aurora]

Max

[slow lotus]

Hmmm

[hollow ivy]

-# bump

[echo aurora]

Would also encourage you to check out our leaderboards with categories - https://arena.ai/leaderboard/text/creative-writing

[ruby cargo]

Every message in Max AI chat triggers a CAPTCHA. It never completes — after ~60 seconds it stops responding and shows a timeout. Console shows 403 errors. Please help.

[slow lotus]

I saw that but idts the models in leaderboard are not too good

And Max sometimes sends sonnet 4.6 but its horrible

[south beacon]

Anyone noticed AI models releaces have become like cellphone releases now, incremental not meaningful.

[hollow crypt]

And like phones in many ways they regress from previous version

[echo aurora]

We have been seeing some users hit this kind of captcha loop that was difficult to get out of over the weekend. Our team has been looking into this and are going to make some adjustments to this system. Overall, if you’re seeing frequent captchas, it typically means the system is detecting activity that may appear inauthentic. I’m not able to go into the specific signals used, but certain usage patterns can trigger these checks. If there is anything in your usage that may be contributing to this, it may help to adjust that behavior and see if things improve over time.

[ruby cargo]

Thanks for the update. Two quick questions:

Any estimate on when the team adjustments will be live?

What specific 'usage patterns' should I avoid? I'm just using the chat normally.
I'll wait for the fix. Thanks.

[flint raft]

I can't with these CAPTCHA's

[echo aurora]

I don't have an estimate to share sorry to say.

What specific 'usage patterns' should I avoid?
Would note that I mentioned:
I’m not able to go into the specific signals used

I'd encourage you to use #1499964195029123082 and share there

[loud herald]

Wait so it’s actually a thing coming?

[ruby cargo]

Understood. Let's skip the "usage pattern" discussion since you can't share the signals.

Let me refocus on the technical issue:

The CAPTCHA fails with a reCAPTCHA Timeout error in the console.

The server returns 403 errors on /nextjs-api/stream/post-to-evaluation/.

This happens consistently in Max AI chat.

This isn't about "behavior" — the integration with reCAPTCHA seems broken on your end (token validation failing or timing out).

Can you please:

Confirm if your team can reproduce this in Max AI chat specifically?

Or suggest a real workaround (e.g., disabling CAPTCHA for trusted accounts temporarily)?

I'm happy to provide a HAR file or console logs if your engineers need them. Thanks.

[echo aurora]

We make frequent changes to this system overtime, but this weekend specifically we were seeing some odd behavior that we're going to correct.

[feral elm]

who ping

:(

[echo aurora]

I'm 99.99% certain this issue is already on our radar, but I'll check with the team and followup with you if that's helpful.

[loud herald]

Sorry I’m a little confused here. What I was asking is if an “Arena CLI” was going to be a thing

[feral elm]

so something smarter better? another economic cycle? you don't know how it would go but you know the drill

[flint raft]

same thing is happening with me to reCAPTCHA rn, for the life of me it won't let me get it right

[ruby cargo]

Yes, please follow up. I'd appreciate knowing if it's confirmed as a bug and when a fix might be expected. Thanks for checking with the team.

[echo aurora]

Oh sorry I thought you were talking about captcha. Regarding CLI/API I won't be able to share details about what is/may be upcoming.

[loud herald]

Ahh I see

[ruby cargo]

Update: after the CAPTCHA times out, the chat dies completely. Error: Something went wrong while generating the response. Please try again. + Trace ID (e.g., f2710b2a-8bbe...). The only fix is to log out and log back in.

This is not just a CAPTCHA loop — it's a session-killing bug. Please escalate this specifically. Happy to provide HAR or Trace IDs.

[flint raft]

same thing is happening to me rn

[feral elm]

let the folks get to that ticket 😭

[shut marten]

GNG I CANT STILL MAKE A IMAGE

@echo aurora

8HOURS WAITING

[slow lotus]

Chapta bug

[shut marten]

fr

[slow lotus]

It kills my chat

How we can pass it

Or I hope we can

[echo aurora]

I'd point you to this message: #general message

[echo aurora]

I'll be sure to get to your Trace when I can

[shut marten]

is the trace the random letters at the top if so ilsend rn

[echo aurora]

Yeah it says "Trace ID: "

When you get a bug

[shut marten]

i dont have a bug

[shut marten]

The AI service is busy. Please try again in a moment.

been happening for 5hours

then i had another

just dosent respond

[echo aurora]

Okay you're likely using the site too often and getting some kind of bug. I'd wait couple of hours and try again. Keep me updated then what happens.

[shut marten]

ive waiting all day

for 1 image

and it just wont do

it

and your saying to wait more

[wicked talon]

Hi

[echo aurora]

How often have you tried to prompt during that time?

[shut marten]

ive told it to make a image then it didint work

so i waited 30m

i said

"so"?

5 more times

and then just gave up

and tried hours later

[pseudo hemlock]

I was gonna say this too

I'm literally looking at the image edit leaderboard looking for this xd

[echo aurora]

Okay yeah same point still stands. Since we had the large outage over the weekend that we're recovering from, I wouldn't be surprised if there is some odd behavior left over from that. Let's give it a couple of hours, and if no luck create a #1343291835845578853 to share more information and we'll take it from there. I don't want to be troubleshooting in #general.

[wicked talon]

Qwen 3.7 max better than Claude 4.6 wtffff

[icy yew]

Is the bug fixed

[lean sphinx]

Do any of y'all know what's the limit for the agent?

as in quota limit

[rare swallow]

Dear god

[echo aurora]

We don't have the usage limits displayed, yet. This is something we plan to change shortly.

[north marsh]

sorry what was the contest?

[timber frigate]

Guys, what`s going on with nano banana model?

[lean sphinx]

Oh okay, appreciate you

[north marsh]

why sometimes it starts to genarate indefently?

[pine meadow]

@echo aurora What is the daily limit for creating videos?

[echo aurora]

Would encourage you all to use #ask-here for these questions. cc @timber frigate @north marsh @pine meadow

[timber frigate]

okay, thx

[shut marten]

still not working

[echo aurora]

Yeah like I said we should give it a couple hours. I wouldn't be surprised if repeated attempts is making the problem worse.

[shut marten]

Something went wrong. Please try again.

[neon timber]

Has Arena lost the Gemini and more advanced GPT chat options?

[shut marten]

oh what

i refresh and it worked

[neon timber]

For image creators?

[pine meadow]

Get out bro

[pliant kindle]

@echo aurora hey do you know if theres a way to force stop a generation via console or smth like that ?

[echo aurora]

Not all modes have a Stop Button, but you can use the Stop Button.

[neon timber]

Has Arena lost the image creation options with Gemino and GPT chat?

[pliant kindle]

yea ik i was asking that cus i have the infinite generation glitch

i alr tried the troubleshooting

so i wondered if that was possible

[echo aurora]

Would encourage you all to ask in #ask-here

[pliant kindle]

alr did lol

[wicked sapphire]

i have a query about limit , can u help 😅

[echo aurora]

I'd ask in #ask-here first

[mint vigil]

Bro just put grok imagine pro again 😭😭

[pseudo hemlock]

temporarily the most expensive models aren't being offered until a more economical way to offer them is found

[whole sundial]

they can't because xAI killed off the model, it's been replaced by grok imagine image quality

[lean sphinx]

Got this on agent mode, does that mean I hit the quota?

Doesn't seem like it, I can start another agent chat

[wind ember]

whats agent mode even about?

i honestly dont get it

[stray aspen]

Whats mai image lol

[thorn hornet]

Something agenty

[echo aurora]

Spot on. But if you're looking for more information I'd check out https://help.arena.ai/articles/1811908126-arena-experiments-agent-mode @wind ember

[wind ember]

thanks

https://help.arena.ai/articles/7138211024-arena-experiments-code-arena-fullstack

interesting

[surreal zephyr]

Honestly this sounds like written by someone who has no idea how llms work

[wind ember]

@echo aurora will this be enabled later

[wind ember]

why

[surreal zephyr]

The point of chat switching is context management 😭

[wind ember]

yea its badly written ig

[fallen verge]

Has it finally been fixed?

[echo aurora]

Yes, this has been fixed.

[dim basin]

guys anyone know how to get claude pro/max or claude code for free

[sharp mirage]

Hiiiiii

[wicked sapphire]

Oh lord i am so scared of "There is a daily usage limit managed via a credit balance that resets every 24 hours. When you hit it, you’ll see a message and can resume after the reset." this

what if this is too low

I only use it for texts

[wicked sapphire]

bye

[sharp mirage]

Bye

Omg finally no sticky bot

[wicked sapphire]

its been fixed

yay

[sharp mirage]

Yaya

[wicked sapphire]

@echo aurora

[fallen verge]

Thank God

[outer flicker]

Yes boomshakalka. Thank you @echo aurora

[terse swan]

Stickybot

Hmm

No stickybot

I assume it's fixed

[wicked sapphire]

yess

[fiery quest]

LESGOO

[wicked sapphire]

not really

[wicked sapphire]

theres usage limit now

[fiery quest]

wat

[wicked sapphire]

yes

welcome to hell

i asked servers ai about it

and it said

[fiery quest]

its gone now

holy

[wicked sapphire]

"Regarding text generation on our platform:
There is a daily usage limit managed via a credit balance that resets every 24 hours. When you hit it, you’ll see a message and can resume after the reset."

i am scared asf

[red shale]

🥲 me too

[fast dew]

Jit ur lucky to be using it for free at all

[wicked sapphire]

hows it lucky if u get hit with a limit right after like 15 conversations

[fast dew]

Not even 15

[wicked sapphire]

plus it resets are a day

after a day*

not like an hour or 6

[fast dew]

Then wait lol

[wicked sapphire]

Thats the point

it gets boring and inconvinient

[fast dew]

Mate wdym it gets boring

[wicked sapphire]

waiting

it pisses off

[fast dew]

How boring does ur life have to be that the only interesting thing in ur day is to use an AI model

Js find something else to do

[wicked sapphire]

Some people work

mate

[wicked sapphire]

Thats the reason this site is working at all

Ur entire point is "if ur broke just buy a house"

[fast dew]

The daily usage isnt even 15 lol

[wicked sapphire]

what it is then

[echo aurora]

You'll have to wait for the timer to reset. This is a part of the new usage system we're playing.

[fast dew]

Idk but not 15

[wicked sapphire]

ik , and the timer is 24 hours

[fast dew]

Js wait bruh

[wicked sapphire]

i was talking about the limit

[echo aurora]

IIRC we're going to do a reset at a specific time of day kind of thing, that applies to everyone.

[fiery quest]

new usage system is added now?

[wicked sapphire]

if it doesnt bothers u , its not ur concern , plus here are more members who agreed with me .

Just yesterday]=

[echo aurora]

No, it's not fully releasesd but we're going to be adding pieces of it here and there.

[wicked sapphire]

its being worked om

on

[fast dew]

Why cant u js wait though

[wicked sapphire]

Thanks pineapple , the reset time doesnt concerns me , but the limit

[wicked sapphire]

Cause i have to work on a project

it requires daily interaction with ai

i am a freelancer

[echo aurora]

Understood. We plan for the limit to be as generous as we can feasibly offer.

[fast dew]

How many requests do u need a day

[wicked sapphire]

please do so , it would be a great help really

I hope it doesnt becomes claude

[wicked sapphire]

Tbh they depend on the project i am working

Around 35-60 ig maybe

[fast dew]

Sooo like on average

Bruh

[wicked sapphire]

Not sure

I said i work with multiple clients

Idk why what i use it for concerns u more than the actual problem

[robust burrow]

well, it the down time was a ploy to make me learn to use agent mode and love it...it worked...damn you 😉

[wicked sapphire]

If the limit goes too low i'd have to look for an alternative

[fast dew]

Cause ur fussing over crap thats not even that big of a problem

[wicked sapphire]

Not for u sure

lol

[fast dew]

Yh cuz im patient

[wicked sapphire]

u are ik

cause ur work isnt imp enough

mine is

[fast dew]

Mate what do u even do lol

[wicked sapphire]

!!!

plus some personal stuff

[echo aurora]

Lol well as unfortunate the the outage was, I am happy that you found a silver lining

[fast dew]

https://tenor.com/view/stan-twt-stantwitter-stan-twitter-treyreloaded-relicario-gifs-gif-11416191407615241229

[wicked sapphire]

Great logical comeback

This gif has been my reaction during this entire convo

[fast dew]

Gng js buy a subscription if u cant wait lol

[wicked sapphire]

gpt 5.1 has been shut down

it doesnt even comes with subscriptions

now

lol

[fast dew]

??

[wicked sapphire]

Gpt 5.1 is not accessible

even with subscription

[fast dew]

What abt 5.1

[wicked sapphire]

i use it with arena

[fast dew]

It's a trash model but aight

[wicked sapphire]

This is so pointless

[wicked sapphire]

Not for me

it has not enough Censorship

[fast dew]

Lol u tryna project or smth

[wicked sapphire]

Right there

[dusky ravine]

the cat sat on the mat

[fast dew]

Ooh ohh i'm so scared Arena is gonna make usage limits low

[wicked sapphire]

Saying anything to u is pointless

cause u have surely hit ur limit

i'll try after 24 hours

[fast dew]

❓

[storm dust]

the mat sat on the rat?

[loud herald]

It absolutely is

Guys putting his resume in

[robust jacinth]

Ty

I was warn pls tell me where I need to put my resume

[echo sinew]

https://jobs.ashbyhq.com/arena.

[robust jacinth]

Ty

[primal helm]

Huh

Why

[echo aurora]

To get usage limits more in-line with what we're planning.

[west lodge]

god damn people need to secure their gateways

i was messing with queries on fofa some idiot left some azure router open with no api key and gpt 5.5 just there

goodbye wallet

[light sleet]

yo pineapple it's eid for me 🎉

[echo aurora]

eid?

[quaint dirge]

Lm arena fixed?

[covert totem]

Chat with claude-opus-4-7 — type 'exit' to quit

You: hi

Assistant: Hi there! How can I help you today?

You: what model are you

Assistant: I'm Claude, made by Anthropic. I don't have information about which specific version of Claude I am. Is there something I can help you with?

You:

Lowk miss

Peak ngl

[echo sinew]

Please, see this message from Pineapple: #1417174113092374689 message

[quaint dirge]

W man

[echo aurora]

Hey everyone - heads up that we're going to be removing the ability to upload files or media to the channels and prevent link embeds during certain hours. This is a temporary measure. Thank you all for the understanding.

[whole sundial]

is this because of what happened last night with the nsfw posts?

[echo aurora]

Yeah, we're looking into more perminant measure, and want more assurance the auto-moderation systems are working properly, before enabling this again.

[whole sundial]

are you going to make a proper announcement? i think some people aren't going to see this and will try to upload an image wondering why they can't

[tropic whale]

Bad 😞 agent

[quaint dirge]

Max ai is my favourite soo smart

[gray oyster]

Never tried max

[echo aurora]

We’re going to encourage people to ask in #ask-here more, there should be an automated message that gives a response for why

[primal helm]

This is unacceptable

Literally just add restrictions to only tenor.com

[echo aurora]

It’s only temporary

[dim basin]

if anyone is like good with veo 3 or just good with prompting images in general, i have a task that im trying to do with veo 3.1 lite and if anyone could help please respond to this message or send me a dm 🔥

[fallow anchor]

is the agent mode failing to create a chat session for anyone else

[toxic verge]

What do u need

[light sleet]

Its an Islamic tradition

like christmas

but Eid is like a holiday for us

[ocean venture]

eyyy happy eid adha bro banana (from Indonesia)

[light sleet]

Thanks!!

Me and @silent tree prayed at same spot xd

[split topaz]

Hey I have an idea

A mod between battle mode and side by side, basically we would be choosing a model and the other model would be anonymous and revealed upon vote, it would especially be useful for comparing some good performing Chinese models to Western frontier. Or just seeing the performance of any chosen model against a random model.

I would call it Half-Cognito.

Happy Eid-al-Adha to all Muslims btw.

[dim basin]

on second thoughts im figuring it out right now with AI but if i have any questions when I am testing it out ill be sure to send you a dm

[fiery gull]

Hi guys 🙂

What you thinking about qwen 3.7?

I'm using now, this suck don't ative the thinking in my phone app

And do some issues 🥀

[split topaz]

Well I heard some nice things about it, like a Tetris post. Qwen 3.7 and the price cuts in multiple Chinese models(so running these models becomes a little bit more affordable for Arena) is actually a reason for my Half-Cognito idea. On the app interface it performed at least better than prior versions.

[fiery gull]

Waiting my small model, but I think just in 4.0 version I'll see a <10b model

[nocturne hamlet]

Is the arena Ai issue resolved??

????

[split topaz]

Oh check out the new Chinese model in AA

MiniCPM5-1B

Insane model for size

%5 on HLE

[split topaz]

And surpasses multiple versions of GPT-4o on AA intelligence index. 1B parameters.

[deep adder]

Where u from

[whole sundial]

look at aa-omniscience lol, refused 99% and got 0% correct

this model is actually very dumb and they are using refusals to cover that up

[whole sundial]

even llama 3.2 1b managed 6% correct

[light sleet]

We both from UAE 🇦🇪

[wraith bane]

ive been using gemini so ive been curious abt this rn

are u able to send more than 15 prompts

[solid stream]

so no more Gemini pro 2 in Arena image?

[whole sundial]

it's been removed from direct and side by side mode, yes

it may come back later with the new usage system, but that is unknown

[nocturne hamlet]

I haven't tried it yet 😭

[wraith bane]

ill test it out

[nocturne hamlet]

Oh ok 👌🏻

[wraith bane]

ive heard it was resolved for most ppl so if i still got it then that should be important info for the devs maybe

at least regarding that its still around for certain users

[nocturne hamlet]

Hmm

[toxic verge]

You know how they train models right?

Data Annotation

[dense sphinx]

I won't complain ever.

No matter if there is captcha many times.

I realize that now.

[toxic verge]

Damn we can’t upload nothing any more for a while?

[finite lion]

@echo aurora , Kindly consider giving option to rename chats!

[buoyant walrus]

hi, has the token limit been reduced?

I send very few messages but still get limited tokens.

@echo aurora

[burnt sinew]

anyone here want free opus usage on api my friend has been making a platform with that

[sharp mirage]

Me : )

Why i can't send a photo to chat here ?

[sharp mirage]

Eid mobrak for you : )

And also me 🙂

[echo sinew]

The ability to upload files, media or link embeds to the channels was removed for certain hours. I encourage you to ask in https://discord.com/channels/1340554757349179412/1466486650170245435, there should be an automated message that gives a response for why.

[sharp mirage]

Yeah i just read it

Thx

[pseudo hemlock]

Sure

@echo sinew Is there a reason why when new models come out, for example 3.5 Flash, it gets added to the text arena but not the search arena?

Is it about numbers of chats in each category?

[echo sinew]

Sorry to say that the Moderation team only handles issues related with the https://discord.com/channels/1340554757349179412/1343285970375540839 (for now), so unfortunately I don't have an answer to your question.

[pseudo hemlock]

o ok

ty for the response

u da goat

[sharp mirage]

Wsg pain guy

[whole sundial]

yeah i'd suggest asking @echo aurora

[pseudo hemlock]

hi hi

PINEAPPLE

[echo sinew]

Hopefully we can give a hand to Pineapple in other matters in the near future

[pseudo hemlock]

well we always appreciate you

[echo sinew]

Thank you

[pseudo hemlock]

hi kiri

[whole sundial]

although with what i know, they don't add models until a 2,000 vote threshold is met so that's probably why it may be added to one arena but not another arena where the model was also evaluated, it also could've just not appeared in that arena before release even if the model is capable of vision/coding/web search

[pseudo hemlock]

that's what i was thinking

its just numbers

well

[sharp mirage]

Happy eid for you yall

[pseudo hemlock]

not that they don't add it to the leaderboard because it doesnt have enough votes but they dont even add it to battle because not enough people use search for it to realistically get to even a few hundred votes

[pseudo hemlock]

Khair Mubarak

[sharp mirage]

Why i got nothing to say ^_^

[radiant canyon]

hello

[austere bison]

Sooo

Still no updates?

[austere bison]

W

[twin lava]

Hello people of the internet

[lofty crane]

Hello there our internet brotha

[pseudo hemlock]

hi

[twin lava]

i wanted to post some image battle result but i cant upload images ? i had anew model : phantom_brush

The new Mai model also is quit good ! almost as good as gpt 2 !

yeah i cant post images ... any help ?

[echo sinew]

The ability to upload files, media or link embeds to the channels was removed for certain hours. I encourage you to ask in ⁠⁠https://discord.com/channels/1340554757349179412/1466486650170245435, there should be an automated message that gives a response for why.

[twin lava]

oh it's nighttime for server admin ? it's morning in europe !!! ☕

[echo sinew]

Yes, our friend Pineapple is sleeping.

[twin lava]

dang europeans always left behind 😶‍🌫️ 😆

[nocturne hamlet]

Hey dose

Any body

Had made an ai..

[twin lava]

yeah my granma did ...

[flint raft]

the CAPTCHA and everything has been fixed for me for a few hours now, just wanna say thank you to the team!

[pseudo hemlock]

wake up pineapple

its only 10 pm

[sharp mirage]

Bro its 8:15 Am

Like its 12pm for him

[wild isle]

its 27am for me

[pseudo hemlock]

No it isn’t

U wrong

It’s 1:21 AM

[old pagoda]

guys im not able to access nano banana pro 2k and gpt image 2 medium model in side by side image tab , what might be the problem did they removed both the models ?

but also in my past convos i can normally access them the problem is only with new convo

[pseudo hemlock]

They’re temporarily removed

They’re looking for a more economical way to offer them long term

[old pagoda]

🥲 when will they be back to use , also any other way to access i have gemini pro but whenver i use it to make images related to politics or lets say with leaders and all it never gave me image

[vale vortex]

Heyoo can you guys send more than 15 prompts now?

[burnt sinew]

dm

[north marsh]

can some one in the team cancel a response for me? i really need to continue that chat

[short sluice]

files disabled

i was about to share a bug

wait do i send it as a link

https://ibb.co/R4vD5cdk

ok so heres my problem

i think it happened due to me almost flunking the captcha

[north marsh]

its been 12 hours for me and its still genarating

[short sluice]

im getting too many battles in direct

i dont usually get that many

[north marsh]

i think arena is having a hard time

[spiral seal]

Hi everyone, I'm glad to be in this community.

[short sluice]

ok

[potent snow]

guys, would you choose galaxy s26 or pixel 10 pro? Need to upgrade phone

[short sluice]

im biased

i think galazy s26 is better

i have a26

[azure cove]

is qwen 3.7 max flash good or thinking

[twin lava]

you can use your google api to gen for free in lobehub ... you can use other free api and mix them there 😉

go in settings -> providers

dude you can read what i said ? stop spamming ! if i tell you that is that i am already on there ...

i tell you you can use api for free no credits you tell please use my link for the credits ...

[vivid quarry]

Hello everyone, does everyone have a Agent Mode? the fact is that the AI service is busy. Please try again.

[twin lava]

no not really its highly limited without api trick ...

[shrewd dome]

how can i use ai ?

[golden ocean]

https://media.discordapp.net/attachments/1196470968328134696/1255185126996967524/vcollide_wireframe.gif

https://cdn.discordapp.com/attachments/389626260562771988/1383950227911737458/1746032488474498.gif

💔

https://tenor.com/view/embed-test-gif-23353939

[distant glen]

/

[pine meadow]

Aide mobark

[blazing hill]

day 4 without Gemini Nano-banana-pro 2k

[naive spruce]

Are you really that broke? Just buy pro. Or if you dont wanna spend money, google flow is exist without credits. You can generate with nano 2 and pro

[supple sphinx]

It's really a shame they removed Gemini's models; they were the ones I needed.

[pine meadow]

yes flow its free unlmited

[pine meadow]

There is no professional version in Arena

[blazing hill]

It's not about money, I would pay 4k if they put 4k on the arena with money, it's about restriction, Gemini and Google flow are so restricted that they can't create a simple image, and I'm not referring to rorn or what, I mean you can't take a picture with celebrities, let's say with Michel Jackson you can't on Gemini or Google flow because it says you can't do it with public figures, which ruins all the fun

[blazing hill]

True

[pine meadow]

I want a little of your money

[deft spruce]

well that message is gone i think it's fixed

what was the problem?

[blazing hill]

you are right they solved the problem

[deft spruce]

yeah so what was the cause?

[deft spruce]

...ok i got it

[echo dome]

server got fixed so i hope childrens won't ruin it again

(over simple ai model bro)

who is this guy

[tranquil gust]

Who am I doesn't mater.

[buoyant fulcrum]

Why is the website down again?😂

[wicked sapphire]

Aw Hell ywah , thanks for this

@ember bough

[steady sequoia]

Hey! Does anyone know when Nano Banano Pro 2K is coming back to Arena? It only really worked well there... 🥲 🤌

[supple sphinx]

Yes, I can't find it either. Please bring back the Nano Banana v3.1 and v3 (2K) models. Even better, bring back the v3 (1K) model if possible.

[azure cove]

glm 5.1 or claude or qwen for building projects

[wicked sapphire]

Ig it will be

[wicked sapphire]

People really have been rooting for qwen here for its potential

[azure cove]

i see

[toxic verge]

I told u

I knew who u were

[compact flame]

I doubt that was the guy who was responsible for the incident

Could be somebody different

[toxic verge]

It is

Nawh

[compact flame]

Okay ig

Yeah uh anyways let's move on to ai topics now

We don't need to remember what happened back then

Today's about providers of free ds on openrouter

Their well free deepseek is having an emotional rollercoaster

You guys can look yourself what is happening to it ig

I forgot we can't post images anymore

[toxic verge]

Yeah hard to forget when we still paying the price lol

In one way from a symbolic point of view this was an L

He posted it but we get restricted.

🤷‍♂️

[calm spade]

It's been almost a week, why hasn't the Gemini 3 Pro image preview 2K been restored yet?

[compact flame]

Probably too expensive to keep it public rn

[worn palm]

did they fix the issue

(I dont see the stickybot msg)

[bright shard]

@echo aurora The Gemini 3.5 Flash model is INCREDIBLE for programming

[remote marsh]

Hmmm 🤔

I’s gem open source

[calm spade]

I'm temporarily using the old dialogue model.

[remote marsh]

Be nice I use hugface to get them haven’t had a chance I have over 70 models

[polar niche]

AI is gonna eat the world

[worn palm]

:WAAAHH: No its not

[remote marsh]

It is

We have a new data center being built for AI in Jersey

[worn palm]

Ok then it is

[polar niche]

Of course, the overlords want AI everywhere

[worn palm]

Yeah they do

I mean chatGPT's going to go bankrupt in a year (According to some news)

[polar niche]

Yeah we need AGI now

[worn palm]

whats AGI?

[polar niche]

💀

[worn palm]

Oh Artificial General intelligence💀😭

[wicked sapphire]

Aw hell i already made an acc , can i put ur referral now ?

send me the link

also is there no free way to get ur credits up .

[wicked sapphire]

Ig i did that . Thanks

[wicked sapphire]

idts they would tbh , they would be backed by a lot of funds for the future promises they can make cause of the large userbase they have

but ofc they can screw things up too

[pine meadow]

@echo aurora Bro, when I use a chat token, I can't send messages later or after a few days, which causes all my projects to go to waste, and I can't upload HTML files for the chat.

[inland quest]

3.5 Flash is for some reason better then 3.1 Pro for coding?

[obtuse smelt]

try gemini 3.5 flash image generation in side by side, is good results

[remote marsh]

@echo aurora Where to apply for staff

Never mind I got it thanks

[north marsh]

cna some one help me the ai is just genarating for ever how can i just stop it?

[mighty reef]

Hello guys how can I send a .zip file ??

or a .txt

I can't ??

[remote marsh]

Zips files aren’t a feature as what I know of. I’m sure you. An upload a link and it can grab files

[mighty reef]

Yh I know do you think with a github repo it will be goo d?

[remote marsh]

You can I use that in my project. Like idk if understand it but I use it like I build what need here or if I need to fix something or whatever it be it has a api GitHub that allows me to upload my files deployed my files and work anywhere as long that gui program or tool or whatever you make website must have that and once you upload it be your project and you can edit in any ai website or desktop apps like vsc virtual studio code

[remote marsh]

So you must make sure it saves before you leave otherwise might not upload it’s own you have to edit what you want and then upload to GitHub and you be fine if makes sense

[mighty reef]

Okkk

TRyy

[remote marsh]

Yea if not I can make one for you and you can use that

[mighty reef]

No it's okay I can do it myself

[remote marsh]

But try

Yea I was goin to say so you can learn how to make things

But if fail or need help ask you be happy then you was before lol 😂

[surreal zephyr]

GPT 6.7-Mango

[timid spruce]

wait is the problem fixed ?

i mean the message here vanished

[wicked sapphire]

Yes

[frosty lava]

Honestly is backend a real problem anymore ? there is still thing they can't properly do ? or its only for very very complex task

[blazing hill]

[frosty lava]

i feel like the big problem right now is only about taste / frontend

[robust terrace]

#general

[sleek urchin]

Is there a source code?

[north marsh]

Yes but I didnt upload it anywhere

[sleek urchin]

Can you upload on github please :> 🙏

[north marsh]

let's see what can i do

some how i forgot to mention @echo aurora

why my message was deleted?

[clever apex]

Hello, why have the 3.5 flash and 3.1 pro models for creating photos been removed? Will they be added back?

[marsh pendant]

Has the Gemini Nano Banana been removed from the website?

[clever apex]

• gemini 3 pro image preview 2k
• gemini-3.1-flash-image-preview
• gpt image 1.5 high fidelity
• gpt image 2 medium. these are the only 4 models i used in arena.ai and they are removing them? why and when will they release them again?

[clever apex]

It has been removed. This disappoints me and I use it all the time.

[open wind]

What is the probability of eating R4T...

[thin grove]

Guys i made my own weather api like OPENWEATHERMAP

[north marsh]

I know what a Remote Access Tool is but I don't get "Eating it"

[toxic verge]

Gemini 3,5 is trash

They made Gemini impossible to use or get work done

Without upgrading to higher tiers which arnt worth it cause you can just pay as much for gpt or Claude

[deep adder]

@surreal zephyr do u know exactly what 5.5 pro is

is it best of 16?

[sleek urchin]

Github?

[north marsh]

not yet...

My github desktop is a little buggy u have to wait

[toxic verge]

Gemini 3.5 pro will eat up all your daily weekly usage with like 2 prompts lol

[clever apex]

Yes, but for what I do, he's the best.

[thin grove]

Yo Aero

[north marsh]

did you even test the build or are you just waiting for me to upload the source code?

[thin grove]

Prolly a RAT if no source code provided

[sleek urchin]

I am waiting cus I dont wanna get virus

[thin grove]

Aero

[sleek urchin]

?

[thin grove]

Aero

I got a new

way

[north marsh]

idk use totalvirus

[thin grove]

glm 4.1 for free

[thin grove]

Nah

[thin grove]

5.1*

[sleek urchin]

GIME TUT IN DMS PLEASE

[thin grove]

Aight

[light sleet]

e

[toxic verge]

U.S. law enforcement agencies have taken notice of the growing backlash to AI and in response are shifting their sights to what they are calling “anti-technology extremists.”

This shift is documented in over 1,000 pages of unpublished reports from the Department of Homeland Security (DHS), the Federal Bureau of Investigations (FBI), and fusion centers across the country obtained by Wired. Fusion centers are intelligence-sharing hubs where federal, state, and local law enforcement agencies work together to collect, analyze, and share information.

Nearly half of US data centers planned for 2026 are facing delays or cancellation

[clever apex]

but will Gemini 3.1 pro and 3.5 flash be added again?

[toxic verge]

At least 25 data center projects around the country were canceled last year due to opposition from the local communities, according to research from intelligence platform Heatmap Pro.

People are finally starting to wake up 😅

This is lame we can post images lol ….

[compact marsh]

can I give you something for free in dms?

[thin grove]

What

[rigid copper]

they have removed it, and the only way is to use google flow i guess, that's what i'm been doing it in response to this