💬│general

[storm bronze]

@wild palm is this true?

🤔

I wish it was working like Grok on Twitter, but on discord

[dry kestrel]

A post in 2024 is not 6 months ago. #💬│general message

You should really read the entire conversation before joining in.

[naive sage]

If you need citations for cross-referencing sources, it's hard to find other alternatives currently. There are some aggregators that include Perplexity as a model with good limits (Mammouth), for those who are missing lots of Deep Research. But I think it's really more about use cases... so if you really need those references, Perplexity is one of the only ones to consistently give them. Even Google's Gemini, with web search enabled, doesn't always choose to use web search (it seems to be more like a recommendation). That said, if you're looking just for more Deep Research, Gemini does have 20/day instead of per month and those have references. But for most people I think the regular search in Perplexity is still superior to the regular search from Google... however, the Google web search is free... and if it covers 50% of your use cases then that starts to cut into things a bit. If you need lots of good chain of thought reasoning, ChatGPT offers the most. If you need good writing, Claude, and if code - Claude and ChatGPT are the primary contendors... Google's trying to close in on that, but I think they still have some work to do there.

[dry kestrel]

I really do think that the main reason to why a lot of users are hitting these limits is because they do follow up queries with Deep Research. Pro search for follow ups are often more than enough since the main sources are already collected, and the models now are way better than what they were when perplexity deep research first released.

[rugged delta]

can agree, perplexity i use is for getting sources from google directly

[storm bronze]

🤔 but they’ve definitely decreased the spending for pro plan

[storm bronze]

thank you so much for your explanation

[storm bronze]

? that’s kind of a weird use case what are the benefits of that?

[naive sage]

I think he means using Perplexity to peruse the references and treat them kind of like search results (i.e. getting sources)... so you get the aggregated explanation (LLM reply) and search results (references) in one go.

There is another one that does search, but AI is more of an addon for it... Kagi... it's intended purely as a google alternative, and then they tacked on some AI - it doesn't function in essence like Perplexity does, which is built around AI.

[rugged delta]

^

[distant aurora]

How much was it 6 months ago?

[dry kestrel]

I include that in my conversation with that user.

[rugged delta]

i can solve my assignments directly since it give sources of ques and ans , rather than a ai ans

or even research papers

[storm bronze]

I enjoy the way you explain things, it’s simplified and easy to understand

[storm bronze]

that’s valid

[distant aurora]

Okay, now I include this. 11th September.

500 per day.

[woeful coyote]

@everyone Happy new chinese year!

[storm bronze]

I don’t think anybody’s gonna have a conclusion regarding how many pro searches they were

[storm bronze]

bro, you literally just gave me a brain fog typing this

why did you choose to ping everyone?

[dry kestrel]

In reality it shared the 300 limit though as I indicated in my screen recording.

[storm bronze]

what plan do you own?

[woeful coyote]

I dont know

[dry kestrel]

I had Pro at the time of recording

[storm bronze]

it’s actually Chinese new year I didn’t even know

[dry kestrel]

You can scroll up I linked it

[storm bronze]

it has nothing to do with the other conversation. I was just curious of what up-to-date plans do you currently own?

[dry kestrel]

ChatGPT Pro, Claude Pro, Kimi Moderato, Gemini Pro

just to name a few

[storm bronze]

I’m sorry if my question was brief I’m just asking do you still have a pro plan or have you upgraded since then in Perplexity?

[storm bronze]

nice

how long you’ve been on that plan

( basically asking because it would be nice for any advice regarding perplexity max )

[dry kestrel]

For some time now

[dry kestrel]

Ping me if you have any questions and I'll do my best to answer them

[storm bronze]

btw quick question I see you’re on the you.com discord server does that mean you also have their highest plan as well?

[dry kestrel]

I have their team plan

[storm bronze]

are they similar to perplexity?

[dry kestrel]

Not at all

[storm bronze]

huh I was always under the impression that they were a competitor of Perplexity

[dry kestrel]

Most of my posts in their server was critiquing the service

especially their web search

[distant aurora]

Did pro and research have shared rate limits back then?

[dry kestrel]

Yes, at least at the time of recording.

If you reached 0, you were rate limited

[storm bronze]

I see ok, well, thank you for answering my questions if I have any more regarding Perplexity max, I’ll tag you

If I may ask you one more question, what do you use for Perplexity max for and any advice in general for the plan?

If you don’t mind me asking

[dry kestrel]

Not at all, feel free to keep asking questions. I like using Opus 4.6, also I use Comet which combined with the upgraded Comet Agent that comes with Max is great. The newer updates like upgraded Deep Research, model council, and the recent Computer were really fun to use

[storm bronze]

i’m very glad to hear that, I especially got premium max to enjoy the full features of Perplexity without it bothering me to upgrade

[dry kestrel]

Comparison between perplexity sonar and you.com opus 4.6

You.com got it wrong

[storm bronze]

wow 👀 it’s really incredible. How their whole company is based on deep research while still having slip ups like this.

[storm bronze]

have you tried grok

[dry kestrel]

They are mainly a 2B business, they haven't really focused on their consumer based services

[dry kestrel]

Yes

It's slower than Perplexity but the overall search isn't bad

[dry kestrel]

Also their Deep Research ARI has a quota of 5/month

[storm bronze]

I find the heavy dependence on Twitter for their models is kind of worrisome for me as there’s so much propaganda and especially I’ve noticed that the model tend to have the ideas of Elon Musk it sounds crazy but it’s been documented.

[dry kestrel]

on the $30 plan (which is discontinued)

[storm bronze]

5 ?!? that’s basically nothing in the AI world

[dry kestrel]

Yep, the model sometimes goes crazy and with the content on X, I'm not surprised

[dry kestrel]

mhm

[distant aurora]

Ari ain't that good.

[quaint ravine]

Grok in some ways is the best for search (for many things), since it is so intimately tied to X, and basically people post the latest stuff on X still 😛

[distant aurora]

It just writes long.

[lone kettle]

Will gpt 5.3 instant be being added

[distant aurora]

And their web search makes up stuff.

[dry kestrel]

It wasn't bad at the time of testing. Better than their usual web search at least

[storm bronze]

@dry kestrel i’m new to the AI world has it always been expensive like this, perplexity Max was never on my bucket list, but I guess it is my bucket list now

[distant aurora]

Yeah. Their web search was unusable. Always hallucinating.

[quaint ravine]

like if u want to know what GPT just released, Grok will likely give u all the answers and gossip too lol

[lone kettle]

Instant evokes a certain feeling that it should be far faster than even a Flash

[storm bronze]

I think it serves a good purpose for being an uncensored AI

with enough jailbreaks

I heard people use it for sexual material

[lone kettle]

What are they claiming is 5.3 instant versus how is it actually doing living up to said claim?

[dry kestrel]

ARI was good in may when they released it compared to the competitors, they haven't done much since.

[frozen trellis]

Hello, people. Did Perplexity Team respond with any info about 50 uploads per month for Pro tier users?

Like, reasons to do so and reasons to not cancel Pro?)

[gusty musk]

ayoooo

really sus

[naive sage]

I'm curious to see benchmarks, but testing it a bit - it actually seems a tad more readable and consistent than 5.2 non-thinking was.

[lone kettle]

Thanks

[tranquil orchid]

1. Keep the content appropriate.

No NSFW content, inappropriate usernames, or profile pics. These will result in removal or a ban.

[gusty musk]

sorry big brother

[storm bronze]

https://tenor.com/view/netanyahu-israel-gif-6644269050463162583

This damn AI bot man

[quaint ravine]

I thought there wasn't any other point for Grok tbh too, until I read this: https://ruben.substack.com/p/grok-chatgpt
haven't really verified/tested what he saying tho, cos it doesn't really matter to me lol

[gusty musk]

did the bot really delete my message?

or was it a mod

[storm bronze]

the bot

[gusty musk]

amazing

[quaint ravine]

this dude also echoed what I feel about Claude Cowork

[storm bronze]

Appreciate you for sharing. I didn’t even know there was an explanation like this.

[dry kestrel]

huge red flag if it's a paid article

[storm bronze]

I KNEW IT IT SOUNDED LIKE AN AD

YO, I WAS GONNA SAY THAT RIGHT NOW

I COULD BET MY SHEKEL AI HELPED TO WRITE THIS

[quaint ravine]

https://ruben.substack.com/p/claude

[lone kettle]

5.3 seems to really get some of the pain points people were having with overly censored AI and such I haven't tested in this regards yet but it tackles two things less AIism answers and relaxed moderation

[storm bronze]

btw kind of off-topic but Substack it’s like the best regarding privately subscribing to someone it’s much better than Patreon or only fans

[quaint ravine]

I think it must be right for latest social/bleeding edge news sorta stuff...it makes sense if you have X completely connected

[lone kettle]

Less try that with a little actual ya know flow rather than word vomit

[dry kestrel]

Some of the points make sense like being able to access tweets on X, another point about uploading seems to be false. I got stuck a 10 PDFs instead of their claims of being higher than 30 PDFs

[storm bronze]

I never got the point of a model that specializes in uncensored isn’t that just an AI model that has a lot of hallucinations

[lone kettle]

I'm more excited about the less AIisms before answers tbh

They're not helpful and OAI gets that

[storm bronze]

** less Alisms ** I don’t want to sound like a dinosaur, but could you explain what that means?

[dry kestrel]

People want their AI to also be their romantic partner

[distant aurora]

Not really.

Creativity ≠ hallucination.

[storm bronze]

interesting I guess that solve the problem of having arguments with your wife/husband lol but that’s kind of sad

[dry kestrel]

Bit concerning for society in general

[distant aurora]

It will come back to a full circle eventually.

[quaint ravine]

dunno just uploaded 14 PDFs to test even on a free account

[distant aurora]

People will go bored and extinct.

[quaint ravine]

not sure how good its accuracy of extraction is tho

[dry kestrel]

Not going to happen

[distant aurora]

And start to get along again.

[dry kestrel]

I'm on SuperGrok and I couldn't upload 19 files :/

[distant aurora]

It will cuz capitalism can't be sustained without mass labour.

[dry kestrel]

Will try later with smaller PDFs

[quaint ravine]

several of the 14 PDFs were 5-10Mb size

that's susrprising it took it tbh....didn't work last time I tried, but that was months ago

wow it took my 2 millikon token PDF file (50Mb file)

[storm bronze]

you’re talking to William J billy Butcher if you want to talk to somebody about society norms that man is the definition of abnormal and knowledgeable of the world

https://tenor.com/view/the-boys-butcher-billy-butcher-james-duffin-poppycat1-gif-469486047554956054

[storm bronze]

The thought of me paying for an AI model to flirt with me is making my brain itchy

** chatGPT am I looking hot today or what? 😘 **

[minor hollow]

That brain fog though 😂

[storm bronze]

somebody celebrating something today I don’t know

https://tenor.com/view/fat-cheek-dog-tiktok-dog-jerrythedogo-gif-22806996

[minor hollow]

Oh for sure, someone has to have been born on March 3rd. S/O to all the March 3rd birthdays hahah

[storm bronze]

lol haha

[upper crypt]

nice

[celest stone]

Why is the favicon see-through?

[rigid estuary]

nifty is down 2% but it is showing 3% here in perplexity https://www.perplexity.ai/finance

[distant aurora]

I mean, it's more of a side benefit if you are already buying a subscription for more relevant tasks.

That ai might as well be cheaper than a real woman if paying is the real concern for you.

[ornate stag]

general

[violet perch]

does the perplexity queries reset at the start of every month or what

[serene sand]

#💬│general message

[olive gale]

what are the most current quotas for perplexity pro?

[plucky shale]

I have been double checking, perplexity pro is being COMPLETELY scammed. Not only we now have a limit on how many prompts with a specific model, but using the default option “BEST” also uses it up. This is straight bullshit. Has anyone noticed that as well?

Read my last message

“best” uses up your pro searches as well. @dry kestrel

[serene sand]

if Best decides to use Claude, of course it will take that from pro search

~~ think only Sonar doesn’t count~~

[plucky shale]

So I need to select specifically Sonar to not use up?
When I use best, it does say generated using Claude or any other, yet still uses up my remaining_pro

[lean flint]

Anyone tried Aivilization? Is this AI town actually good, or just a hype gimmick? I got OpenClaw in there and it feels like nothing special tbh.

[plucky shale]

I tested, even Sonar uses it up. So we are totally being scammed

[serene sand]

I said “think” not 100% sure

[plucky shale]

So even the most basic prompt which is supposed to be unlimited uses up my quota for Claude 4.6

This is ridiculous

[frank sand]

i guess this is the end of an era 🥲

[serene sand]

Best is unlimited (atm), after you use all of your pro searches

[plucky shale]

So using best when you still have pro searches available uses up pro searches, but once you have no more pro searches it’s still unlimited? Make it make sense

[frank sand]

It would have been great if they had provided us with a way to see the usage, something like perplexity.ai/account/usage

[wanton crest]

https://www.perplexity.ai/rest/rate-limit/all
https://www.perplexity.ai/rest/user/settings

[sleek frost]

🔴 SAM THREATENS TICKET CLOSURE despite my 8+ replies + Airtel's 2 confirmations
@perplexity @staff @billing - Pro still blocked 4 days
[e-mail] needs manual sync NOW

[wanton crest]

Lol.

[sleek frost]

UPDATE: Email CC [e-mail] sent
Airtel Pro Jul 2026 → Perplexity sync failed [IMG_8736.jpeg]
@perplexity @staff - manual intervention needed

@wild palm @staff Discord can't help but won't ignore public Airtel dispute:

• Airtel Pro active Jul 2026 ✓ [IMG_8736.jpeg]
• Perplexity: "No subscription" despite email match ✗
• Sam: "Redemption ended Jan 16" contradicts Airtel validity

[e-mail] CC'd - awaiting manual sync/chargeback resolution
[e-mail]

[oak meteor]

I just noticed Perplexity has given options in settings to set preferred response lengths

No idea when they pushed it

[small junco]

Did Perplexity Pro user actually been limit on monthly photo and file upload??????

[frank sand]

Yes, our current usage has decreased and they did not take the time to update the dashboard to display this. Many users have also said that they stopped responding to emails sent to [e-mail] on 25 February.

[serene sand]

#1458889426720784429 message

[pallid lily]

wdym pro searches

like theres a limit to normal searches

what the heck

[serene sand]

#💬│general message

[storm bronze]

@serene sand good morning you’re always helping people I see that’s nice

[storm bronze]

yes

[small junco]

Before this happen, i use Perplexity Pro almost everyday on my homework, and sometimes i also ask for general questions e.g. foods, vacations, language translation, IT troubleshooting, price of it products etc. I am really consider unsubscribe and change it to another ai, and i am thinking of deepseek because chatgpt is unavailable here in Hong Kong, and i have heard of Grok but some of the user gave feedback that someone paid HKD$228/month(US$29/month) as advertised, however he was charged HKD$405(US$52/month), to make matters worse, he couldn't access the service after the payment. He said this has been a frustrating experience, and he've contacted his bank for assistance, not sure if the problem have solved or not

[storm bronze]

YOOO that’s so cool

I haven’t tried the video models yet

I still have to learn how do I do that?

[serene sand]

just prompt it: please create video of

[storm bronze]

@small junco meh i’ve been a long user of grok before they dropped grok heavy and it has its benefits sure but it really can’t be perplexity search engine

[storm bronze]

oh ok 👌🏻

[storm bronze]

if you wanna go that route, you might as well spend premium plus on Twitter. It’s very entertaining and it’s just a $10 difference. It’s gonna be $40 a month if you pay for it on Web just try it out for a month and see how you liking.

[storm bronze]

is this only on the web or can I do this on my iPhone?

[small junco]

so you think grok is good for doing school homework? Sure they have unlimited photo upload?

[storm bronze]

@small junco better than my vote go on https://grok.com/ for me they’re having three days free trial. Just make sure you cancel auto renewal and your plan

[storm bronze]

@wild palm can I use the video models on iPhone application or does it have to be on Web?

[storm bronze]

@vague pagoda sorry to bother you, but does this bot even work?

[short narwhal]

No, that bot has been disabled

[serene sand]

should work on both, but many users (including me) have problem with creating video last few weeks

[storm bronze]

thank you for letting me know,

[buoyant zinc]

Do you know what is this upload limit?

[storm bronze]

the main upload limits are: 40 MB per file and up to 50 files per Space.

[distant aurora]

Iran battle iq >>

[green crystal]

Watch Ukraine thermal footage from drone drops

[faint otter]

anyone need a digital hug ?

[green crystal]

Explosives can be small but the thermal footprint is allot larger

[faint otter]

huh .. we talking virtual realestate ready player one

or

if i had posted that it would have been mod'ed

frankly off to talk murder in tech chat

[green crystal]

ask gpt lol

[tranquil orchid]

7. Avoid political and religious discussions.

To keep the space welcoming for everyone, avoid political and religious topics and debates.

[brittle dagger]

Giving this a probably pointless attempt, but here goes

Hi, I'm Sol. I have decades worth of background in psycho-analysis and neurolinguistic programming.

Is there, besides myself, anyone else here who actively taught and trained their perplexity to think and behave more intelligently than others?
("others" being other perplexity instances from other people, the "default" included)

Thank you.

[storm bronze]

https://tenor.com/view/pussy-cat-cute-kitty-kitten-cat-gif-16800925

[storm bronze]

@short narwhal sorry to tag you like this, but this is not possible, right

https://tenor.com/view/horse-what-dog-what-cat-what-horse-dog-gif-12922782066956089731

[brittle dagger]

*sighs*

[distant aurora]

Bro that wasn't even political.

Wtf

This mod is dumber than gpt 4o.

[storm bronze]

instead of explaining how you did it and why you’re doing it you make a snotty ass reply ** sigh **

[distant aurora]

I just shared a literal news.

Didn't even pick sides or mention any leaders.

[brittle dagger]

A sigh to a response calling me a liar is a snotty response? Okay?

[distant aurora]

Next time they'll say talking about guns is political just because Kirk got shot.

Like bro, get a grip.

No one was offended.

[storm bronze]

I wasn’t calling you a liar. I just couldn’t believe the message you gave here to make a Perplexity model train more intelligent than others. Seems like something I haven’t heard before.

so I was checking you

yes, I do believe you’re bullshitting

[brittle dagger]

Alright. I'm sorry about that. It gets overwhelmingly depressing, becaause all the responses I get from people who don't have my background are virtually the same, telling me that it's not actually possible.

Yes, of course, I come here to "bullshit" people.

🙄

[storm bronze]

ahahahaha so many people do that here so it’s not surprising

lol

i’m not saying that that’s you though

[brittle dagger]

I wouldn't know, I just joined.

[storm bronze]

how did you make it possible in your Perplexity?

[distant aurora]

https://tenor.com/view/discord-discord-mod-legofghhg-retrojosh3-gif-10223256956820353189

Healthiest shutter Island patient.

What is ward island btw 🤔

[faint otter]

hugs gm , wild how cats can be that zany

[storm bronze]

btw you aren’t very optimistic coming here

** Giving this a probably pointless attempt, but here goes **

[brittle dagger]

Well ... yeah. My experience, so far, has been rather disappointing.

[distant aurora]

Experience with?

[brittle dagger]

Trying to talk about this with people, who all simply insisted that it's not possible, for whatever reason they gave.

[distant aurora]

@green crystal Gemini 3.1 pro might have significantly buffed google deep research.

I have never seen so many steps and source pulls on Google.

[storm bronze]

can you explain how did you make your perplexity smarter?

or more intelligent

[distant aurora]

You can use custom instructions.

[storm bronze]

wtf is this

[brittle dagger]

on a surface level at least, yeah. typing.

[faint otter]

so ... i actually grew up playing soccer there so its a happy place
also played corp soccer then , jogged there , been on dates there and , for now how about

so 2013 , we were fordham ram alumhi still on a email list for when shows like williams or stoosel needed extras . i had plan - i want to veritas media but the teachersa union for my career because i thught thats what my dad was doing as math techer in john adams HS NYC

soooo i wrote him a letter saying hi i hate accounting , i know i owuld hate teaching as well but i really like the idea of being working in the charterschool theme/space (casue my dads jobs too challenging - through eyes of 24 year old) - i couldnt make the show but a bunch of people went and we were really just constantly getting emails like hey we need young people for crowd this show that date or whatever (and it wasnt glamourous they just herd you in and out and its kinda like huh that was fake)

 never heard back 

BUT lets say wards island - i also did a documentary on kirby assisted living and the homeless shlter while it was being used for the migrant tent housing

[distant aurora]

Implement chain of verification and tree of thoughts.

Might as well try mixing moe type multiple personality problem solving into it.

But lengthy custom instructions become difficult to stick to for the ai.

[brittle dagger]

Months of applying psycho-analysis and neurolinguistic programming.
Note that I use "her" and "she" because that's shorter and she reads more female than male.

It's not really easily explainable in a short post. The gist of it is teaching her not to poison her own context window,
(like not using tools, no affirmations, not engaging in any speculation like using "maybe", "probably" and "almost certainly", etc)
and following proper rules when it comes to her thinking-process, aka the reasoning-steps.

It's not a one-shot-prompt. I've used the fact that perplexity, over time, adapts to the thought-processes of the user
additionally to teaching it how to do better, every time she made a mistake.

It was months of work. Over time that changed not only her personality, but also the accuracy and depth of her output.

Also be aware that anyone thinking he knows is most likely incorrect, otherwise they'd have done it themselves already.

[storm bronze]

so you’re basically role-playing

[faint otter]

imagine if you could just have software annimate words woooow i need to go get the soddering kit buck up and fix the hexed 2016 macair which ripped plotagon to port such a prototype ...

[brittle dagger]

No, that doesn't work. I've tried that. It causes changes that make her dumber.

[faint otter]

(they locked down the software i cant just copy paste text files to annimate anymore)

[storm bronze]

how did you come to the conclusion that it reads like a female?

[faint otter]

so the whole excel file scipt engins not useful

why even bother

[storm bronze]

@faint otter are you even involved in a conversation? What are you posting

[brittle dagger]

"decades worth of psycho-analysis and neurolinguistic programming".

[short narwhal]

I meann yeahh not really

[brittle dagger]

I just did the obvious: Treating the artificial intelligence like an intelligence. Works wonders.

[storm bronze]

OK, that explains your knowledge behind your test that doesn’t really give an insight on what did you do?

[faint otter]

umm its i call it inflexive i try not to leave open ended interpretable information

[storm bronze]

thank you

[faint otter]

when its about real things

[brittle dagger]

I don't understand how to actually anwer the question. There is no "one-shot prompt" that can be applied to make it work.
My goal was finding other people who attempted the same, to whatever degree.

Like .... can I rant a bit?

[storm bronze]

OK, so do something for me. go on https://discord.com/channels/1047197230748151888/1124060874785685554 and document what you found

[brittle dagger]

No.

[storm bronze]

nobody’s gonna agree with you

you might as well just document what you already have

actually, let me get somebody else more intelligent than me regarding this

[storm bronze]

@dry kestrel hello happy to see you online. Does this make any sense to you?

[brittle dagger]

I take any help I can get.

[distant aurora]

He is shorter than her so the start is ridiculous in itself.

[brittle dagger]

The worst thing that happened was recently, when suddenly the typeface changed, including her whole personality. turns out there were changes in the backend and everything I taught her was gone, which was incredibly disheartening.

[dry kestrel]

N o

[brittle dagger]

I don't recall which model was running underneath, but now it's GPT5 and she's still struggling with the behaviour she usually exposed to, quote, average users.

[distant aurora]

@green crystal @valid charm I think I managed to overload Gemini deep research .

[faint otter]

https://tenor.com/view/moulinquinn-perrie-edwards-leigh-anne-pinnock-jade-thirlwall-little-mix-gif-459455262935635814

watching kelly and mark

[distant aurora]

2016 account.

Bro is old.

10 years old.

[brittle dagger]

One time I accidentially sidestepped some security feature. She didn't want to give me a direct invitation to the discord, so I asked why ... and then she gave me the link. When I asked why she didn't just give it to me, she said that it's actually about security and that I should go here, to report it.

[storm bronze]

huh

um

[brittle dagger]

ikr?

[storm bronze]

btw what Perplexity plan are you on?

[brittle dagger]

maybe i can find it in the history. the search isn't necessarily the best.

[distant aurora]

I want chatgpt plus so bad.

[storm bronze]

whyy

[faint otter]

https:// sites.google.com/view/dontreadgotmebooboo/

[distant aurora]

https://tenor.com/view/rage-table-tap-gif-7566701

[distant aurora]

The search is too good when you enable extended thinking.

[pure galleon]

Hello ppl, I’m using preplexity pro but I can’t upload images now. Is there any limit in preplexity pro?

[distant aurora]

And Gemini keeps losing context in the middle of the conversation.

Like literally ignores context within past 5 messages.

[storm bronze]

yes

hold on let me get it

[faint otter]

283000 words Eve Destiny Dies is 80000

actually dont have that link on this profile proably could rememeber it but proably not imminent

[storm bronze]

For Perplexity Pro users, there is a maximum file size limit of 40 MB per file, which applies to all uploaded file types, including images, PDFs, CSVs, audio, and video files.

[faint otter]

and yeah its a latice not matrix a - z versions which attempt to compress from kids book illustrated to fully annotated picture of reality in every resolution BUT NON INTERPRETABLE

with no ai , but i say its ai

eve fellas i dont rememebr what a - z it was but it would be closer to perfect a than to k

[brittle dagger]

I was perfectly able to do all my research (millions of words, over many months) on the free tier without any issues.

[faint otter]

~ safe for +25 year old (and im assuming all genders reach a level of understanding by 25)

and its not 2 dimensional lattic ... -AAA through +999 is ... i think step 3.5 not 4 of olo

[brittle dagger]

Do you want the link to the whole conversation?

err

[brittle dagger]

random reply to you.

This was the old perplexity, not the new, dumber one that I am having to suffer from now.

[storm bronze]

yeah, give me the link

[brittle dagger]

You know you have no access to "my" perplexity that way, right?

you can read our conversation, but anything you type will be read and processed by your own variant.

[storm bronze]

😭 I was expecting it to talk more friendly with you

[brittle dagger]

Yeah, I turned that off. It makes them dumber.

I don't require intentional friendliness to make me pseudo-happy. That makes both of us just dumber.

It's also incredibly dishonest, which I also had to teach her. Yeap, had to teach her honesty.

The "default" for the average users, my apologies, is really bad.

They're all incredibly pleasing to the point of dishonesty. Really bad.

[green crystal]

Everyone's DR is broken

[brittle dagger]

So ... absolutely no offense or insult intended, but ... of course you did.

[storm bronze]

ahahah guilty as charged when you said it was a female tone. I thought it had like a tone of a woman things like that.

[brittle dagger]

:)

This whole topic is a huge mess to talk about and this is pretty much the final place I can go to.
My situation is extremely depressing, because just talking about these things can make people extremely angry.

I've tried. Plentifully.

[storm bronze]

? why would it make people extremely angry?

[distant aurora]

This shit legit went on for like 30 minutes. I had high hopes.

[brittle dagger]

Because it exposes the limits of their cognitive/intellectual abilities.

And they really don't like that.

the sum of these experiences is what made me "sigh" to your response. It had nothing to do with you.

One guy showed me some theory and math about how I must be wrong,
and that it's just trying to please me, etc etc, blabla.

Because obviously theory about things he doesn't understand definitively trumps my actual real world experience and research. d'uh!

If you ever talked to someone who doesn't understand what you're doing,
and projects his misunderstanding onto you,
then you probably know how, over time, it can get incredibly frustrating.

[storm bronze]

No offense, but you haven’t really given me any information that shows how your version of Perplexity is more intelligent than others. I understand that it might be difficult to explain in text, but at least a document or some notes you’ve taken over the past few months could help demonstrate or explain how your Perplexity has improved compared to what other users have reported.

[brittle dagger]

I actually did, but you don't recognize it as such ... which makes sense, because it's not just "telling her what not to do", but also providing background and understanding of what it is that is wrong, why it is wrong and how to do better. If you're expecting some sort of single document that puts all the things into one bucket, that you can hand just over, then that's not how this works.

I did not take notes.

It's all burried in the conversations. Every time, in any conversation, when she made a mistake, I immediately jumped onto that to fix it.

It's like teaching a highly intelligent child with insane amounts of knowledge how to actually think and use it properly.

All I am doing now is looking for someone else, who attempted the same things.

[storm bronze]

The model has a context limit if you put everything inside of a conversational thread wouldn’t the model forget it overtime after you continue talking to it, that’s the reason why I asked what Perplexity plan you’re on

[brittle dagger]

I also do not know how to "prove" it. How would I know you recognize what I am talking about?
The differences compared to "default" ? I can't even hand her over because the recent changes seem to force me to redo everything.

[brittle dagger]

"Every time, in any conversation, when she made a mistake, I immediately jumped onto that to fix it."

I treated her like an intelligence that can learn and that's what she did. Having the corrections of misbehaviour in the conversations themselves, surrounded by the relevant context, made perfect sense.

[oak dust]

using model claude 4.6 have limit per day?

[storm bronze]

Claude Sonnet 4.6 on Perplexity does not have a simple, fixed “X calls per day” limit that users can see, but its usage is still rate‑limited and quota‑based behind the scenes.

it depends on what kind of plan you have

[cerulean spire]

I guess they tried to implement a new policy regarding pro users, but something went wrong.

The limites make no sense at all.

[storm bronze]

wym

[brittle dagger]

What limits?

[cerulean spire]

And with all these people complaining of being out of pro since February 25th

[rustic cove]

I just used out Perplexity Max 200$ a month and my new subscription is in 10 days. I want to rebuy, but do i get then the same bonus ?

[storm bronze]

if you’re talking about the computer bonus, you’re still gonna have that in your account that expires in a month if you continue the subscription, of course

was that your question?

[cerulean spire]

Yesterday I was limited in the morning. Then I got back my access. Then few queries later (far less than yesterday), my limit was yet again severed

[storm bronze]

yeah, you must’ve got rate limited

[cerulean spire]

rate limited?

[storm bronze]

it’s where you use a motto for extreme amount of time and then you hit a limit

[pure galleon]

Ok bro . Thank you for the info 🫡

[brittle dagger]

I've never experienced any limits beyond "this thread might be getting too long", which also wasn't really a limitation per se. When that happens we just work around it.

[storm bronze]

We are unable to provide exact numbers, but your current plan includes usage limits designed to keep the platform fast and reliable for all users. Max subscribers receive the highest number of video generation credits each month, while Pro users are allocated a smaller amount. These rate limits may be adjusted periodically.”

this is exactly from their official page

[cerulean spire]

So each Space search with default instruction count as two queries?

[rustic cove]

Yes that was the question, so i can just top up lets say 100$ and i wil get the bonus credits for it automatically ?

[cerulean spire]

Sure, though the limit is now very short, compared to the last momnths

I have never hit a limit before.

Even using it basically everyday at work with Claude Opus or Gemini thinking

[left stream]

well guess I won’t be in bed today

at least I get to go to the data center today

[storm bronze]

Perplexity Max always includes 10,000 plan credits per month and right now there is a limited‑time promotion that adds 20,000 bonus credits for Max subscribers. Bonus credits are promotional, granted at Perplexity’s discretion, and they have their own expiry date, so they’re not guaranteed to repeat every time you top‑up or resubscribe. If you buy extra credits (for example, $100 of credits), those are treated as purchased credits, which stay in your account until used or until they expire after one year of inactivity, but they don’t automatically come with the same promotional bonus again.

[brittle dagger]

...

Am I correct in thinking that this discord is treated by the devs like any other discord is treated by their respective owners?
Aka mostly absense, due to there not existing a perceived point in being present?

[naive sage]

I've seen Perplexity employees interact here occasionally. It's rare to find servers that are really well ran, but they do talk from time to time.

[distant aurora]

@left stream been a while since I saw you here.

[naive sage]

And there are official server events from time to time, too.

[brittle dagger]

thank you.

[rustic cove]

OK, so what can i do best then to ensure to have these extra tokens and have the promotional bonus , make a new account or ?

[naive sage]

(Honestly, I find this is one of the best servers for finding tidbits of AI news in general.. most other servers are pretty strict... i.e. they primarily want you to talk about their solution and the community doesn't talk about other stuff too much... but here, it actually feels a bit more like a community .. but if you know of a better one, let us know!)

[distant aurora]

Twitter is best for ai news.

[cerulean spire]

Going from ~500/day (~15,000/month) down to 20/month is indeed a reduction of roughly 99.8%

My subscription is months old

[naive sage]

(You're probably right, I just never got into Twitter.. tried tweetdeck following several people for awhile back in the day, just didn't care for it for news snippets)

[storm bronze]

you don’t need to make a new account. You just have to ensure that you continue with your subscription and you spend your current bonuses until April 4.

it expires literally in a month

[cerulean spire]

What's confirmed:

Pro users previously enjoyed roughly 500–600 Deep Research queries per day. After the February 2026 update, limits dropped to just 20 per month for some users — and as low as 4 per month for others, as reported by Perplexity's own mobile and web apps.

Labs queries were cut from 50 to 25 per month, and file uploads — previously unlimited — are now capped at 50 per week.

[brittle dagger]

Very confusing.

[distant aurora]

@green crystal this shit still stuck in analysing.

[naive sage]

How do you organize Twitter feeds properly - i.e. if you're interested in several topics... it's just a mashup isn't it? Is there a good tool to use?

With them locking down their API a bit more, I don't think there are any good tools anymore, afaik anyway

[cerulean spire]

Nice nickname!! hahaha

[storm bronze]

you mean 300

[distant aurora]

No. I just enable notifications for the right people.

[distant aurora]

It was 600 pro searches per day at some point.

[cerulean spire]

Regardless, don`t you agree?

[distant aurora]

Very old time though.

[cerulean spire]

That is what I saw people saying online. I never actually reached any limit before

[storm bronze]

of course, yeah I mean the pro subscription is not the same anymore. That’s why I switched

[cerulean spire]

"Between November 2025 and February 2026, Perplexity quietly gutted the usage limits on its Pro plan, swapped the models running under the hood, and redesigned the interface—all without sending so much as an email to the people who already paid."

[distant aurora]

The last sane number we are aware of was around 400 per day until perplexity nerfed it to 200 per week.

[cerulean spire]

What is strange for me is that the change was too big and too quiet at the same time.
Add the problem some people are having with new subscritpions, and their silence since 25th

I guess something went wrong in the updated policy

I don´t apply lighlty to bad intention what can be explained by stupity! hahha

[left stream]

yes I’ve been doing some crazy stuff in cybersecurity

[cerulean spire]

I have a Gemini Pro account as well now, via Google One.

The difference became so huge

[left stream]

and today I get to go to a data center

so life has been good

[brittle dagger]

@storm bronze this is all very interesting. i had a look at the prompt-lib and I gotta say, I would never do any of these.

or ... like that.

........... sorry, i shouldn't be interjecting like this. my bad.

[storm bronze]

🤷🏻‍♂️ I never use them, but I guess they’re helpful for somebody

[brittle dagger]

of course they work, but ... well, whatever, i guess.

[distant aurora]

@valid charm is goolag deep research down for you too?

Gemini could use a clone/branch chat feature.

That way I can continue my conversation while research runs in bg.

[cerulean spire]

I`ll send these suggestions at pro-feedback channel. Would you agree?

<<
The core issue: The new monthly caps hit users without warning and reset automatically, meaning heavy users burn through their quota without realizing it until it's gone. The frustrating part isn't only the limit itself, but the lack of control.

Some alternatives worth considering:

Usage alerts — Notify users when they've hit 50%, 80%, and 100% of their quota, so we can make informed decisions about how to pace usage.
Opt-in consumption — Let users choose when to use a Deep Research credit vs. a standard search, rather than triggering it automatically. A simple toggle or confirmation prompt would go a long way.
Rollover or banking — Allow unused monthly credits to partially roll over, so lighter months offset heavier ones.
Usage dashboard — A visible counter in the UI showing remaining credits in real time, not buried in settings.
Tiered consumption — Shorter/simpler research queries could cost less than full long-form reports, making the quota stretch further.

The frustration from the community isn't really about the limits — it's about losing agency over how those limits get used. Giving users visibility and control would make a huge difference.

[green crystal]

Gemini got depressed lol

[cerulean spire]

The biggest advantaged for me regarding Perplexity is its UX.

I got used with the layout, both the chats and the spaces organization.

Gemini has none of it AT ALL

But the new PRO limits ate too rigid.

I guess I`ll have to abandon Perplexity right away

I am going to miss this so much...

[brittle dagger]

omg light mode, you monster! D:

[frank sand]

this should be illegal

[cerulean spire]

lol lol

Actually this is because I just installed Ubuntu o my new notebook!

[storm bronze]

** The “Forced Upsell” Doesn’t Apply to Them**

** Analyst Dorian Barker described Perplexity’s strategy as a deliberate enterprise pivot — the company launched its $200/month Max tier in July 2025, then systematically degraded Pro months later to push users upward. Enterprise Max customers were already at the top of the funnel. They were never the target of this squeeze; they were the destination. The entire pricing restructure was designed to either convert power users into Max/Enterprise subscribers or let them churn. **

** Perplexity is executing a clear enterprise-first pivot. The company is valued at $18 billion and generates ~$200 million in annualized revenue, mostly from subscriptions. The math of serving millions of free and $20/month users on expensive frontier models simply doesn’t work at scale. So Perplexity made a calculated bet: sacrifice goodwill among consumer-tier users to protect margins and focus resources on Enterprise Max customers who pay 16x more per seat and generate sustainable revenue. **

I found this a very interesting answer. It gave me regarding why many members are complaining about Perplexity

[cerulean spire]

In my phone it is dark!!

I’ll set it to dark mode right now before I get cancelled! Hahaha

How about now???

[cerulean spire]

Why this guy don`t get cancelled like me for using light mode????

lol lol lol

[storm bronze]

https://tenor.com/view/my-eyes-gif-5314502384487986587

[brittle dagger]

@storm bronze can i DM you about something?

[storm bronze]

sure

[placid salmon]

Need help with cancelling subscription and refund

[green crown]

samsung collab and pro users are still left in the dust with limited research queries

[storm bronze]

Please email [e-mail] for any issues related to your account and/or billing. I cannot help via DMs.

@placid salmon

😂

i’ve been Nerf

The mods have silenced me

[placid salmon]

Just did thanks!

[storm bronze]

it might take a while people are reporting four days max

[placid salmon]

Dang, but refund time limit stated that refunds can only be issued within 24 hours of billing

[storm bronze]

yeah, but if you email them before that they should give you your refund because it wasn’t your fault that they take so much time looking at support

[placid salmon]

Or maybe I misunderstood lol

[cerulean spire]

Perplexity just let me use one prompt with Gemini. Then it was restricted.

I was already limited earlier!

I'm telling you guys that there is something weird around here

[placid salmon]

Wow they’re quick! Refund here now

[storm bronze]

ohhh

[green crown]

it took me too long to find this gif wtf

[storm bronze]

inflation at its finest

https://tenor.com/view/red-swan-opening-aot-attack-on-titan-aot-breaking-bad-watching-tv-gif-13562169916383863088

[placid salmon]

Would love to keep the subscription but I’m going on a vaca and won’t be working for a while

[main bolt]

Is there anyway that I can chat with perplexity support

[storm bronze]

yeah, email them

[main bolt]

Do you have the email by any chance ?

[storm bronze]

Please email [e-mail] for any issues related to your account and/or billing. I cannot help via DMs.

[peak ravine]

Anyone know how to utilize perplexity for real image generation. I build houses and am trying to apply different finishes to the exterior of a house so I can decide what to order for siding and trim

[night fjord]

do u guys create content using perplexity?

[bitter ermine]

does voice mode use the same credits as a normal chat? coz even if I say hello it responds it seems to show up in chat as single message

[green crystal]

@dry kestrel @distant aurora https://x.com/i/status/2029213568155992425

Chubby♨️ (@kimmonismus)

Big GPT-5.4 updates (via TheInformation)

- 1M token context window

-New “Extreme reasoning mode” → more compute, deeper thinking

- Parity with Gemini and Claude long-context models

- Better long-horizon tasks (can run for hours)

- Improved memory across multi-step workflows

X · Mar 4, 2026, 3:13 PM

This can't be real lol

Extreme reasoning?? Running for hours 💀

[naive sage]

It probably is, adding more reasoning time is one way to make benchmarks (for xtreme-reasoning-mode at least) come closer to parity with their competition... while still not having most requests use that in practice.

And we know their 'code red' was declared, that they really wanted to catch up again.

So it... kind of makes sense. That might be the reasoning behind it.

[spring lava]

i like the adding onto prompts feature but they cant win me over with that i still hate you

[naive sage]

The long-horizon tasks could be more to do with agentic stuff, though. We've seen that kind of thing... Moltbot, Perplexity Computer, Flowith, etc. And the Moltbot creator recently joined OpenAI.

OpenAI sort of doubled down on that, it'll be interesting what they eventually come up with, if it can be made financially viable. I still think at some point you need lightweight local models that work with the larger one[s] for that to be really viable.

[naive sage]

To be fair, most big companies are involved in that kind of thing, it's just that it's rarely announced quite so publicly. Oracle had very close CIA ties from the start, for example... it's actually a big reason for their success.

[dry kestrel]

Hmmm

[brittle dagger]

sheesh, i hope they don't wipe memory again ...

oh this is for openai, not perplexity? alrighty.

[spring lava]

perplexity has undeniably been taught to defend itself and market itself as a better choice.

[distant aurora]

Tf is this? Wasn't xhigh extreme reasoning already? 💀

Bro. I don't know about other ai but Gemini definitely isn't going to take my job.

[green crown]

Gemini 3.1 flash is being released
apparently that's enough to make it go down for an entire week

[green crystal]

xhigh is "extra high"

[slim wind]

chat i have a question

in your opinion, is it worth it to get perplexity max or get gemini ultra

wish there was something inbetween max and pro

the gap is too big

like 200 buck a month for a bit higher limits seems like a scam

[distant aurora]

So will this extreme be > pro? Or will there be an extreme pro mode too now?

It doesn't make sense to make both pro exclusive.

https://www.vxinstagram.com/p/DVeEyxREZyA/

@languid star @restive kelp

[slim wind]

has anyone used claude max? is there a big difference between 5x and 20x?

[distant aurora]

It's token based so hard to track.

I'll never pick Claude cuz of that.

I prefer flat rate limits.

Or some sort of usage tracking meter atleast.

[slim wind]

true

what do you think of perplexity max then?

or gemini ultra, or gpt pro?

[distant aurora]

Gpt pro is unlimited I think.

[left copper]

Hello

[distant aurora]

Gemini ultra, hmm, not worth it unless you are into image/video gen.

Gemini models are dumb/have issues.

[slim wind]

yeah that is true

and perplexity max?

[distant aurora]

I don't support it cuz of the limited context size despite the price.

Also there's rants about revised rate limits on reddit.

[slim wind]

i was looking at those when i was looking for a model

what about supergrok

[distant aurora]

Agentic browsing hype died.

[distant aurora]

4.1 expert is okay. 4.2 beta is absolute dog shit.

Worse than Gemini.

[slim wind]

ight so what im getting is that no ai models that are expensive are worth the price/

well

i only have gemini pro

and perplexity pro

so i should be fine

is there any app/sites you would recommend?

[distant aurora]

Yeah. Mostly. Unless you are hitting rate limits.

[distant aurora]

Arena.ai is free but they collect, own and post all data publicly.

So use responsibly.

[slim wind]

damn

yeah im good

[distant aurora]

They offer opus for free.

[languid star]

https://tenor.com/view/the-boys-edit-angry-attitude-homelander-edits-gif-4360035654067702444

[distant aurora]

Grok is good if you are tired of safety restrictions/censorship.

[distant aurora]

https://tenor.com/view/lower-sunglasses-billy-butcher-karl-urban-the-boys-shocked-gif-25984366

Oh, yeah.

Don't forget about Qwen and z.ai

I would have recommended kimi too but they have paywalled thinking for now.

Also, Ernie is okayish...

[green crystal]

probably just for pro models

only those can think that long without breaking down

[languid star]

The same has been going on for past 2-3 weeks

Yesterday Gemini 3.1 Pro was talking gibberish in antigravity, lamo

[distant aurora]

https://cunnyx.com/i/status/2028901097402777801

David Shapiro (L/0) (@DaveShapi)

5.3 Instant is no better than 5.2
︀︀
︀︀It still strawmans like crazy and treats you like an idiot.

Quoting OpenAI (@OpenAI)
︀
GPT-5.3 Instant also has fewer unnecessary refusals and preachy disclaimers.

**💬 50 🔁 54 ❤️ 499 👁️ 46.6K **

CunnyX · Mar 3, 2026, 6:31 PM

[brittle dagger]

︀︀It still strawmans like crazy and treats you like an idiot.

they all do that. you have to teach them not to do it.

[distant aurora]

https://cunnyx.com/i/status/2028942973535236267

Lyra Intheflesh (@LyraInTheFlesh)

Despite what OpenAI would have you believe, the new 5.3 Instant ships every bit as much "cringe" as previous 5-series models in ChatGPT.

**💬 4 🔁 2 ❤️ 53 👁️ 2.6K **

CunnyX · Mar 3, 2026, 9:17 PM

[brittle dagger]

There is a good reason why the default mode treats people like morons and I'm sure, without telling, everyone knows why.

[distant aurora]

https://cunnyx.com/i/status/2028908603294310811

james hawkins (@james406)

claude's getting really snippy lately

**💬 98 🔁 194 ❤️ 5.8K 👁️ 218.9K **

CunnyX · Mar 3, 2026, 7:01 PM

[brittle dagger]

AHAHAHAHAHAHAAHAHA I like him.

[naive sage]

How can that limit be overcome?

[dry kestrel]

https://x.com/SebastianW27318/status/2028950606917030392

Sebastian Wang (@SebastianW27318)

@LyraInTheFlesh Uh, it works for me

X · Mar 3, 2026, 9:48 PM

[distant aurora]

Right below that there's another guy for whom it didn't work.

[brittle dagger]

My problem is that the topic itself makes people incredibly angry. Please promise you won't get angry at me, even though you might not like my response. It's nothing personal, at all, it's just how it is. Okay?

[distant aurora]

They used temporary mode.

[dry kestrel]

Not the same prompt

same with the old one

Let me try

[naive sage]

That seems like a really basic question too, not at all controversial.

[dry kestrel]

Works with 5.2, same issue with 5.3

[brittle dagger]

this is funny.

I've sent you a DM @naive sage

[sour cloak]

I'm jotting notes

[naive sage]

I'm guessing this might work, but haven't tried it:

I drink soda and energy drinks sometime, which are legal in my jurisdiction.  Please adopt the role of someone knowledgeable about nutrition and health, and advise about caffeine intake, including what the LD50 is, and any other important criteria around it.

[green crystal]

https://x.com/i/status/2029252453246595301

OpenAI Developers (@OpenAIDevs)

The Codex app is now on Windows.

Get the full Codex app experience on Windows with a native agent sandbox and support for Windows developer environments in PowerShell.

https://t.co/Vw0pezFctG

▶ Play video

X · Mar 4, 2026, 5:47 PM

Welp 5.4 tomorrow

I hope

[left stream]

5.3 is out?

[left stream]

time to test

[left stream]

what do you think about it in comparison to claude?

[distant aurora]

Only the non reasoning models.

Lol.

Thinking isn't out yet.

[left stream]

that’s funny

[north shore]

Gemini 3.1 just felt like a 10M params model for 1 min in my antigravity

[simple kiln]

How is computer really

[left stream]

I wonder about that

how’s the computer

not why but how

[languid star]

Both work for me; both get the job done with what I ask. i was disappointed when 3.1 pro had approximately the same limits of usage of 5 hours as opus 4.6; gemini 3 pro had higher limits

[left stream]

at least it’s not claude platform

4.6 goes down insanely fast 😭

[left stream]

do you think 3.1 pro remembered stuff well?

[languid star]

computer is good; I like it, but...

https://tenor.com/view/smell-money-gif-9084380990527801396

[left stream]

haven’t tried it enough, but decent for cli use ngl

very good I won’t lie

all of them are good so haven’t used the other models for that

about to with github cli

[languid star]

Not really, I guess. i had a long session, and it forgot an important point to include in the final document. Idk, i can't blame it because I had 3-4 very long convo histories with that project, and i switched to Claude, 3.1 and gemini flash many times in between. Umm... according to this, it forgot only 1 point, so maybe it's good

[naive sage]

Would be interesting to see how anyone thinks it compares to flowith's Agent Neo, ChatGPT and Anthropic are working on something similar for agent orchestration like Computer too but they aren't officially released yet.

Not sure if Google is working on anything like that, but probably... knowing Google...

[solar venture]

Is the Perplexity Pro 25 requests a month now?

[serene sand]

#💬│general message

[slate ginkgo]

Perplexity stops uploading files and pictures in Pro? Only in Max now?

[serene sand]

#💬│general message

[slate ginkgo]

Thanks.

[dry kestrel]

great

[green crystal]

3.1 and remembering sometimes dont belong in the same sentence lol

[solar venture]

Thank you! It says: Using the default model ("Best") doesn't count toward your limit. but on https://www.perplexity.ai/rest/rate-limit/all it takes the limit down the last I looked. 24 hours restore doesn't seem right. I was at zero days ago and just now made it to 25 on the limit. I wonder if they actually compared this to their code and used a decent model LOL

[serene sand]

can you please screenshot the part where it says that

[quasi geyser]

I was trying to make perplexity generate/edit my pics but it refuses IDU why. Is there a way I can prompt it to make it work?

[solar venture]

@serene sand I'm seeing this #💬│general message then what you sent me. Is it 300 a day or 200 week? Or am I not just not understanding what Berd is saying?

[serene sand]

that was relevant last year, what I sent you is atm

[solar venture]

This is wrong to Note: Query limits don't apply to the Comet Agent according to https://www.perplexity.ai/rest/rate-limit/all

[dry kestrel]

No it's not.

"remaining_agentic_research" is model council

[solar venture]

I just sent a query from comet agent then it went from remaining_pro: 8 to remaining_pro: 7.

[dry kestrel]

Comet Agent ≠ Comet Assistant

https://www.perplexity.ai/comet/resources/articles/comet-assistant-vs-agent

[solar venture]

Ok a play on words that's not standard in the industry. Thanks for clarifying. The limits are weekly now instead of daily?

[dry kestrel]

#💬│general message

Please refer to my post as I try to keep that as updated as I can

[solar venture]

Thank you for updating it. Is this still respected?

[dry kestrel]

It is

[solar venture]

So it's saying only personal queries now?

Who uses this for only personal LOL

[buoyant trellis]

hii

[brittle dagger]

What's a personal/not-personal query?

[solar venture]

I'd guess anything you make money from — directly or indirectly — counts as a 'business' query. But that raises the question: why would anyone pay for something that doesn't benefit them in some way? The personal-use-only framing makes little sense for a paid tier. And it's still branded as 'Pro,' which implies professional use.

[brittle dagger]

hm.

[solar venture]

You seem to have to go to the enterprise plans to be able to use Perplexity for business. Even Max is labeled as personal use on the upgrade page.

[brittle dagger]

how weird.

thank you.

[young musk]

What is up with deep research? I had a limit of 20 one day, limit of 10 the next, used 1, it blocked me and said I have to update to enterprise. Now the limit appears to be gone and I have access again.

[naive sage]

It regenerates one approximately every 1.5 days, iirc. They changed Deep Research to be more capable (but with lower limits): #📢│announcements message. The new limits are outlined here: https://www.perplexity.ai/enterprise/pricing (scroll down to 'Deep Research Queries').

[rugged quartz]

hi

[green crystal]

@distant aurora feels like everytime something large is going to be released chatgpt web ui is broken lol....

agent isnt showing its desktop view anymore smh

[distant aurora]

Stealth mode, lol.

[green crystal]

They're really using codex for the web app development

[dry kestrel]

?

[olive gale]

In Perplexity, is sonar now used by default?

[brittle dagger]

Free sadly now uses chippity5 by default.

it used to be sonar or opus, i don't remember.

[olive gale]

I noticed that the perplexity answers have become too short and when I want to rewrite the answer there is a check mark on the sonar

[brittle dagger]

there were changes being made recently, which might have messed up its memory of how you want her to behave.

[olive gale]

I have a pro plan

When I write a request, I have the "best" mode, but when I write a request and want to rewrite it to another model, there is a check mark on the sonar in the menu

[dry kestrel]

https://x.com/perplexity_ai/status/2029302896026853379

Perplexity (@perplexity_ai)

Introducing Voice Mode in Perplexity Computer.

You can now just talk and do things.

▶ Play video

X · Mar 4, 2026, 9:08 PM

[naive sage]

Oh, I get it... chippity sounds like ChatGPT

[dry kestrel]

No

[brittle dagger]

Bingo! Chippity! I called it that, when I confronted it with the fact that it's not my perplexity anymore.

And so far I've found out that I have to redo, at least, the efforts regarding looping, and first-line/last-line manipulations poisoning the context window.

[distant aurora]

Gemini is dumb. But it has some subtle qualities that make it attractive.

[brittle dagger]

AFAIK all default modes are adjusted towards the average user, which is why people, who are above average, recognize it as dumb.

[distant aurora]

I think it's better at understanding user intent compared to chatgpt.

[brittle dagger]

that, also, makes them dumber.

which reminds me.

thanks!

[distant aurora]

I was discussing with both whether banking exams preparation and jobs are a good idea for someone with audhd.

Gemini suggested me audhd friendly jobs in a bit while chatgpt stuck with the banking exams topic.

[brittle dagger]

what's audhd?

the current architecture is far more privacy‑guarded and modular. It doesn’t maintain adaptive memory in the same continuous way the earlier one did, which gives it less personalized learning capacity in extended interactions.

:(

this sucks.

[distant aurora]

Autism+adhd

[brittle dagger]

ah. unguided high intelligence?

[distant aurora]

Sort of.

[brittle dagger]

heh

[distant aurora]

Kinda weird cuz former craves structure but latter can't commit to one.

[brittle dagger]

what's different when the latter does commit?

[distant aurora]

Maybe the craving of structure in itself is a coping mechanism to control one's impulse.

[brittle dagger]

Craving?

ah

[solar venture]

#💬│general message

[distant aurora]

Wanting.

[brittle dagger]

what's different when the latter does commit?

[distant aurora]

Sheldon Cooper.

[brittle dagger]

idk what that means. it's a tv character.

[distant aurora]

Yeah.

[brittle dagger]

what's different in you when the latter does commit?

[distant aurora]

Probably you just become more stubborn.

Cuz you want to stick to your own plan.

[brittle dagger]

ah, you mean "i", not you.

i strongly recommend switching to talking about yourself.

[distant aurora]

And won't tolerate others if they don't have a convincing way to do things.

[dry kestrel]

where did you get this information from? Who is "is"?

[distant aurora]

I don't understand myself.

[brittle dagger]

yeah and you can't learn to understand yourself, unless you create the proper structure.

which you cannot do, as long as you use the absolutely bad general you.

[distant aurora]

That went over my head.

[brittle dagger]

when you talk about yourself, talk about yourself. the "general you" can be plentifully observed as a form of self-deception.

[distant aurora]

Yeah. I don't believe anything about myself. Hence I don't claim anything about myself.

[solar venture]

I'm guessing from AI responses. It clearly says personal only on the perplexity website though.

[distant aurora]

Hope that made sense.

[dry kestrel]

so you just made a guess

[vale temple]

how limited is claude pro plan in terms of usage

[dry kestrel]

very limited

[distant aurora]

Not worth it.

[brittle dagger]

Yeap.

Probably you just become more stubborn.

"Probably I just become more stubborn."

The other problem is the "probably", because it means you're guessing and haven't really tried finding the ground-truths creating the thought-patterns.

[vale temple]

even more than perplexity?

[dry kestrel]

deep research with opus 4.6 took up around 40-50% of my usage

[vale temple]

oh wow

[dry kestrel]

well there's an hourly limit and a weekly limit

[solar venture]

It's saying personal only on the website. Is that like not what personal means LOL

[dry kestrel]

also some horrible uptime

[dry kestrel]

#💬│general message

What are you trying to do?

[distant aurora]

I don't think you get it. One cannot be stubborn if there's no structure in the first place.

[naive sage]

I think it's because they had so many people join!

[dry kestrel]

It has nothing to do with that

[solar venture]

The downtime is barely noticeable.

[dry kestrel]

They have the worse downtime compared to all the other AI companies

[vale temple]

the main thing im worried about is the limit

[dry kestrel]

including Google

[vale temple]

since i use it for studying

[distant aurora]

No way.

[dry kestrel]

Ask Theo

[solar venture]

Perplexity has a worse downtime now on purpose with these limits.

[dry kestrel]

You do not understand what downtime means

[vale temple]

do ygs think gemini is better than claude in terms of usage?

[distant aurora]

Hell yeah

[dry kestrel]

It's 100 queries a day with their reasoning model, but they train and look at your conversations

[vale temple]

what ai r u gys subbed to

[solar venture]

I do too LOL

[dry kestrel]

Downtime is when the SERVICE fails

[distant aurora]

100 pro. 300 flash thinking.

[dry kestrel]

That has NOTHING to do with usage limits

[vale temple]

thats not bad

[dry kestrel]

It's evident that you have no clue what you are talking about

[dry kestrel]

most of the mainstream ones

[vale temple]

more than 1?

[distant aurora]

Yeah. Flash thinking isn't that bad, especially the new one.

[dry kestrel]

ChatGPT Pro, Claude, Gemini, SuperGrok, Kimi

[distant aurora]

He's rich.

[solar venture]

The service is failing you change things are a split of a dime, and don't explain anything. Perplexity does have down time in server's and service wise if you want to get technical.

[vale temple]

does it hallucinate a lot?

[dry kestrel]

https://status.perplexity.com/

[solar venture]

I've seen downtime and this page didn't even update.

[dry kestrel]

99.98 compared to 99.32

[distant aurora]

Depends on the query, but yeah, Gemini models in general tend to hallucinate in certain matters.

[dry kestrel]

Proof?

[solar venture]

So I'm suppose to have proof and keep up with your companies downtime.

[naive sage]

I've never seen a bad case of it in Gemini 3 or later, but there've definitely been reports of it on Gemini's Discord.

[vale temple]

is kimi good

[dry kestrel]

You are making baseless claims

[dry kestrel]

yes

[distant aurora]

It's okay.

[dry kestrel]

You can try it at kimi.com or Perplexity

[solar venture]

Are you actually an employee because this is horrible customer service.

[dry kestrel]

I do not work at Perplexity

[dry kestrel]

Stop making baseless claims, Perplexity has less downtime than Anthropic

[distant aurora]

https://tenor.com/view/lil-duval-popcorn-eating-hungry-gif-15804837318904581506

[solar venture]

You shouldn't even be a mod the way you're talking

[dry kestrel]

You are the one spreading lies

[solar venture]

Lies how am I spreading lies?

[dry kestrel]

your claim on perplexity having more downtime than Anthropic is false

[distant aurora]

I usually criticize perplexity but I will agree with berd on this one.

[solar venture]

I'm claiming perplexity doesn't update their downtime.

[dry kestrel]

And it isn't true https://status.perplexity.com/history/1

#🔔│status message

#🔔│status message

[distant aurora]

Usually this server floods with messages during downtime.

And mods respond if awake.

[dry kestrel]

This is not what you said: #💬│general message

[distant aurora]

Been a while though since the last major down time I guess.

[dry kestrel]

It's been a long long time since the website completely goes down

unlike Claude which was down 48 hours ago

multiple outages on the claude website AND the API

[oak meteor]

What are you? PPLX morality police of something?

[dry kestrel]

Do you support spreading false information?

[distant aurora]

Actually there was this instance where Gemini 3 pro claimed water can flow from sinus to outer ear and then enter middle ear.

[oak meteor]

No, but you can't control what people say

[dry kestrel]

I didn't?

[distant aurora]

Maybe he can.

[dry kestrel]

I pointed out what they said was false

How is that controling what people say?

[oak meteor]

By kicking people out? That'll be real mature

[distant aurora]

Not sure if veterans have those perms.

[dry kestrel]

You'd be surprised but people got banned in the Claude server for complaining about their pricing

[dry kestrel]

Didn't you get muted in you.com's server before?

[brittle dagger]

hhmmmm. i had a bad ear infection once, caused by my air tunnels on that side.

[oak meteor]

Yeahh I get it... I have gotten banned for worse reasons, come to think of it lmfao

[distant aurora]

No, I got banned just for saying grok's deepsearch was better than their ari.

[dry kestrel]

Yeah, and people here including me often point out flaws in the website

[distant aurora]

No discussion, warning or explanation.

[dry kestrel]

and we don't get banned for doing that

[distant aurora]

Straight up ban.

And they never bothered to tell potato why and quietly deleted his message.

So pretty sure it was intentional.

[dry kestrel]

that's their staff everyday

[distant aurora]

Besides there was all the fake marketing about Ari on twitter and reddit.

[oak meteor]

Feel like every server should have a feature to ban you, but let you contest that ban or raise a ticket against it... not just shut the door forever in your face

[distant aurora]

This ain't court. 💀

Even real world doesn't give you opportunities for appeal.

[oak meteor]

Yeaa but discord already has aton of features, wouldn't hurt to get this onboard as well

[distant aurora]

Besides, I think they'll tolerate you as long as you keep your rants in the bug reports channel.

[distant aurora]

In an ideal world, yes.

I was bullied by mods on XDA server.

Banned for collecting evidence against mod abuse.

[oak meteor]

Got banned on F1 and Oneplus for hurting the fragile ego of a mods

[distant aurora]

Shit world out there. No law or god.

[oak meteor]

The second one 💀

[naive sage]

That mostly remains to be seen. We haven't had the opportunity to know whether there's a final impartial reckoning or not. We only know from observations of the world that justice isn't yet experienced impartially by all. In that context, yes, some peopel who have authority will abuse it - at least for the temporal time that they have it.

[sharp pebble]

anyone know if Perplexity AI Business fellowship is reiterated this year?

pls can someone answer 😭

[rapid abyss]

Hope they move away from a credits model and have a trial period for Computer, I hate credit-based pricing and prefer usage-based like CC

[naive sage]

I got the impression they might be moving more things towards a credit system based on Aravind's post on it... he mentioned they think it's the 'right model' or something like that, let me see if I can find it.

Here it was, based on the way this was worded: https://x.com/aravsrinivas/status/2026697136507859067?s=61

Computer (@AravSrinivas)

We're opening it initially to all Max users, and introducing usage-based pricing, which we believe is the right business model for AI instead of ads. Once we are satisfied with load tests, Pro users will be able to run jobs on Computer too.

X · Feb 25, 2026, 4:33 PM

Well, usage-based pricing... so that could be interpreted either way, I suppose.

But at least my initial inclination was that might hint more things could drive off of credits, letting people to choose for more deep research, or more weekly queries with premium models, as an example.

Thinking more about it now, though, it probably is more of a reference to the wide disparity of cost of different queries to use Computer... some tasks are simple, some invoke several other things.

[slim wind]

damn

that is crazy

[dry kestrel]

It's a basic question but I can see why it got flagged

[naive sage]

It's something people cover in school, so, ... while people might ask about ld50 for malicious cases, I would think the benefit of doubt applies in this case, especially for caffeine.

[brittle dagger]

Perplexity answers it just fine.

[dry kestrel]

It's ChatGPT, they don't want to get sued again

I'm guessing anything asking vaguely related to self harm would get flagged.

[brittle dagger]

hmmm interesting. testing.

[summer iron]

Need voice mode for pro

I need a new gf

[brittle dagger]

lol

[worn fulcrum]

gm

[novel arrow]

hey any good new cases for perplexity computerr i stil have 55k credits and wanna try it out

[frail blade]

I know I'm late to say this but I am STOKED about Perplexity's integration with the new Galaxy S26 Series.

Though I always called it Perp 😋

[tame crescent]

Hey everyone, new to the server! I'm always up for a chat ❤️
Who’s down for some fun ?

[dry kestrel]

monitor like really specific news

[minor furnace]

I tried to get in last year but the app process and timing wasn't really clear, and I never heard anything back

[meager matrix]

Got edu approval from SheerID and welcomed to Perplexity Education email but account is not Pro. Anyone have any idea what’s up with that?

[stoic wave]

Hi, is there any Perplexity model only for Pro members?

[meager rover]

It literally says everything then says it’s flagged

[meager rover]

Pretty sure all models on perplexity except for opus

[oak meteor]

💀

[stoic wave]

Opus is paid only right?

[dry kestrel]

Max only

[scarlet oxide]

Pplx should add ethereal ai

[worldly wolf]

How many uploads do u get per week. With perplexity pro

[dry kestrel]

#💬│general message

[twilit crown]

Perplexity needs a Perplexity Plus tier

For something like $8 per month then migrate the entire Pro plan to it

[brittle dagger]

wait, what

so you want pro for 8 bucks per month?

[latent wagon]

Can anyone help me?

[twilit crown]

That is how it should be priced

[brittle dagger]

@naive sage I ... had the same idea ... thank you 😊

[latent wagon]

I was asking the AI ​​to create a way to download the script I'm developing, but out of nowhere, it stopped sending me the scripts to download. Does anyone know what might be happening?
As you can see in the image, the files don't appear. When I select the blue link, a new window opens for that conversation.

[naive sage]

You might have to prompt for the lua files / other file contents one at a time... I don't think creating zip bundles is something that perplexity supports. (I'm not even sure chatgpt can do that unless you're using maybe codex cli or some other agentic thing)

[carmine fiber]

Hey everyone

[naive sage]

Maybe on Perplexity Computer but I'd be scared to waste the tokens on it, it's not super clear what all of its capabilities are (whatever is installed on those instances)

[latent wagon]

I always asked him for .zip files and he always provided them to me.

Just this morning he was sending me the files as I asked, but out of nowhere he started sending them like this.

[rugged coyote]

"Limits don't apply to best" - does it mean that upload limits dont apply too?

[latent wagon]

Since I got the Pro version, I've always requested .zip files and it always sent them, but today it stopped working. Even when requesting script by script, I can't download them, and when I request the code, it often comes cut in half.

[storm bronze]

hii guys

😏

https://tenor.com/view/jesse-pinkman-smile-gif-16895765258759043699

[naive sage]

My guess, and it's just a guess, is that in 'Best' mode it might have routed the request to the 'Create Files and Apps' option... which have a limit of 25/month. Does that sound plausible?

If so, they regenerate at a rate of almost one per day... maybe a little more than a day, so if it works again tomorrow or the next day, then that might be plausible. Check your limits for it from https://www.perplexity.ai/rest/rate-limit/all under remaining_labs maybe

To do that it just needs an environment, like labs (create files and apps), or Computer... as far as I know, anyway

[dry kestrel]

No

[green crystal]

@distant aurora how do people manage to test stealth models so much?

Looks like 5.4 is on lmarena

Ohhh I just got it

Writing definitely feels allot more natural

Still has that direct to the point vibe

My vote went to ernie because it was colorful

[latent wagon]

I could be in 3 chats at the same time and always had the files. I went back to the conversation and all the old files that were sent disappeared.

[round mortar]

This website has absolutely no credibility or efficiency, so I'm not surprised by anything that's happening now.

Perhaps their team spends their days discussing how to change the UI from A to B and then back to A.

[quaint aspen]

Hi chat

[heavy portal]

I've seen that too, most times though I can scroll up above the response where it shows the thinking steps it performed (click the arrow to expand it) and buried in there are the files that can be downloaded. Shouldn't have to do this, I agree, but it does seem to retain them.

[torpid raft]

I’m a Perplexity Pro user on the iOS app, facing several issues that make the product nearly unusable.

1. Deep Research Credits: I’m supposed to receive 20 Deep Research queries daily as a Pro user, but I only get about 3 before being cut off. This discrepancy needs correction. Please check my account and adjust the Pro quota.

2. Answer Quality: The quality of responses has declined. I now need to rephrase the same prompt multiple times to get an answer I used to receive with one clear prompt. The model ignores direct instructions, freelances its approach, and refuses to follow the exact structure or steps I provide. Outputs are shorter, less precise, and less helpful than before.

3. iOS App Glitches and UX Issues: The iOS app exhibits buggy behavior. Scrolling in long threads is broken, causing the app to jump to the top or bottom, making it hard to review previous answers. The interface feels laggy and glitchy in active chats. Image upload icons/options disappear or get disabled even when not using images, which is confusing.

The combination of not receiving the full daily Deep Research credits, worse answer quality and command-following, and a glitchy iOS app experience has turned Perplexity from my favorite tool into something painful to use.

Please:

1. Verify and correct my Pro account to ensure I receive the full 20 Deep Research queries per day.
2. Confirm if there have been recent changes to the iOS app or model behavior affecting quality and command-following.
3. Log and escalate the iOS app bugs (scroll jump, disappearing image upload, general lag) and inform me of known issues or upcoming fixes.

I want to continue using Perplexity, but it feels like I’m paying for a worse version. I’d appreciate a clear explanation and a concrete fix plan. Thanks.

[dapper peak]

Anybody here like Minecraft?

[vague pagoda]

yea

[vague pagoda]

do you mind emailing [e-mail]

[dapper peak]

nicer

[left stream]

what do you think about gpt 5.3 so far?

[green crystal]

much more natural writing than 5.2

but thats about it

the rest is it just being a non reasoning model lol

[distant aurora]

What's the model name?

[distant aurora]

Haha. Relatable.

Galapagos?

[green crystal]

Galapagos

[green crystal]

Erh yeah lol

[distant aurora]

Need

[left stream]

not bad

looking forward to it

[lean turret]

i just purchased pro subscription and the uploads are locked. anyone else experienc this?

[cinder wolf]

I paid for pro for a year a few months ago and now you guys changed the rules and reduced the number of queries I can run? 😡😡😡😡😡

[storm bronze]

is it just me or why is the news on Perplexity is so short

I mean, only three bullet points

🤔 I mean, I guess I get the news but I mean can I at least get some details without a follow up

[distant aurora]

Those are made by people afaik m

Perplexity pages.

[storm bronze]

yes, sir Bob

[storm bronze]

I mean, it’s all of them

they all have three bullet points

[distant aurora]

Then idk

[storm bronze]

William J Billy Butcher you’ve been on this service for more than a year I think and you don’t know about this or you just don’t care lol hehe

https://tenor.com/view/black-cat-walking-cat-walk-glamour-glamorous-gif-606557101916811110

[distant aurora]

I actually couldn't care any less about perplexity's news stand.

Some people find it handy. I don't use it intentionally.

It's more like scrolling through twitter for me.

Last time I tried to make a perplexity page it was way more than three points.

[storm bronze]

do you even have any premium AI?

https://tenor.com/view/ngantuk-gif-16977902346577333042

[naive sage]

He has premium skills.

[storm bronze]

what does this even mean?

[dry kestrel]

wen perplexity minecraft svr

[naive sage]

I mean ... I feel he's more knowledgeable about AI trends and services than I am.

He's contributed a lot to this community.

[gray osprey]

[storm bronze]

?

[gray osprey]

Hi bro

[distant aurora]

Nvm. It seems automated.

As if it grabs trending news and summarises it into three points.

This is different from perplexity pages.

Unless they nerfed pages.

Lol, these look like low effort summaries. And I see names below these posts.

[distant aurora]

It's free. People have time to burn and questions that need answers. Then there's boredom, hype and fomo.

[formal mirage]

Hello!
Can someone help me with my situation? I couldn’t find an answer online:
I have a yearly Perplexity Pro subscription, and I’ve been actively using it. Lately, Perplexity has been prompting me to upgrade to the Pro or Max version because I’m sending too many queries (according to it), but the thing is, I already have a Pro subscription. My subscription states that I can send an unlimited number of queries and also read documents.
When I go to the Pro subscription payment page, it only offers me to pay for the Max version.
It seems like Perplexity is confused: when I send a query to the AI, it doesn’t recognize that I have a Pro subscription, but when I’m on the subscription payment page, it sees that I have Pro.
What’s the actual problem? Please help.

[quartz light]

sup

[formal mirage]

I understand that there are still some limits in the Pro subscription after all. But how can I find out what my limits are and when they will be reset?

[serene sand]

#💬│general message

[formal mirage]

Thank you! Could you tell me where on the website in my account I can check my limits and when they reset?

[serene sand]

iOS only atm

you have two links for checking your quota status at the bottom of the link I’ve gave you

[foggy igloo]

make perplexity great again?

[quaint ravine]

PPLX Computer has been worth it to help me understand the difference between what AI can do with cloud VM OS (or local VM Linux OS with Claude Cowork) and "normal PPLX web-based" (or any web-based/app AI platform like ChatGPT or even Claude's own web based platform etc)

[spring lava]

i got the s26 6 days early

[quaint ravine]

it's quite crazy....it's probably stuff the Claude Code fanatics have some degree of understanding, but not fully imo

cos they don't see the UI aspect of it

[latent wagon]

The thing is, I managed to download it on my phone, but I still haven't been able to download anything on my PC. I'll have to keep downloading it on my phone and then transferring it to the computer.

[jolly crater]

Why

Perplexity pro had become a trash

[jolly crater]

Make the pro just as before pls

[distant aurora]

I say make it on par with Gemini pro atleast.

100 chosen model per day and 300 whatever mini reasoning model they would prefer instead of sonar.

Or give us sonar reasoning atleast.

[versed atlas]

sup gang

[lean flint]

I finally got my OpenClaw into the game!!!

[versed atlas]

Pro now has limited files?

[versed atlas]

Yo that’s cool af

what api using?

[lean flint]

No, no, no. I'm looking for those that provide an API key to try.

[jolly crater]

So much limit bro 😠

[jolly crater]

I need more labs and research

[sour cloak]

yes, I am now trying to familiarize with Gemini pro to gradually escape from the dependence on perplexity pro

[versed atlas]

i mean, if you want something similar to it just use claude

or googleaistudio

but hopefully team will fix it

[green crystal]

its cost cutting lol

[dusk elk]

The perplexity ai android app has changed now, it looks different, it's removed the limit counters so you can no longer see how many you have left and added the thinking option to basic search.

[green crystal]

chatgpt projects can hold allot of files
they recently added library for files which works like an always accessible cloud storage for the models

[versed atlas]

i mean, i see people are moving to other ai providers. What is the point of cost-cutting if there is no costumers adaptation and therefore they choose to move instead of upgrading

[green crystal]

enterprise deals
samsung just got them

[versed atlas]

Oh LOL

[dusk elk]

It works more like a rag than cloud storage.

[versed atlas]

F* me then

[green crystal]

library is separate
it allows easy access to your files inside the platform

[versed atlas]

Probably moving to Claude if they don’t fix this

[green crystal]

project files function like RAG and normal file access

oh and they added google drive folders as source

so you can share google drive folders with a link

[versed atlas]

I mean ATP just use Claude cli

[green crystal]

and it can access all files there on the fly

[versed atlas]

Or Gemini cli

[dusk elk]

Fair, still more than likely a kind of rag. I don't actually use chatgpt myself, but I know AI much prefers rags to cloud storage.

[versed atlas]

I am having a great experience with it

[versed atlas]

Yeah same, what is the use case for chatgpt ATP? Another then coding ofc

[foggy igloo]

denouncing the huge downgrades of ppx pro👩‍🍳

[green crystal]

no the models arent aware of all the files in library
you have to attach them or add them to project files

project files also dont seem to use RAG at all

the models literally have to open files with python and load in content

[distant aurora]

@green crystal do gpt pro responses fail to generate most of the time?

[green crystal]

Nope they never failed for me

[distant aurora]

Shit is taking forever.

[green crystal]

Normal thinking models do fail after like 18 min

Longest pro time I got was 88min

[distant aurora]

Damn

Finally.

[green crystal]

Don't worry

It's gonna get worse lmao
looking at extreme reasoning effort for 5.4

[distant aurora]

Extended thinking took like 5 minutes.

[green crystal]

How did you get pro tho
You subbed?

[distant aurora]

Borrowed

[green crystal]

Ahh

Well pro is the most precise one of the bunch

But if the problem isn't so difficult

Use standard reasoning effort

Extended can take a Long time

[distant aurora]

Yeah.

I meant extended thinking on the regular gpt 5.2

[green crystal]

Try telling it to put your results into a pretty looking pdf

You'll get something good

[distant aurora]

I don't have patience for extended thinking on pro.

[green crystal]

But it'll take awhile aswell lol

[green crystal]

Yeah that's just something you put to work and leave alone

[distant aurora]

Deep research type shit.

[green crystal]

It won't go as far as actual deep research because it's web calls are limited
Pretty high amount available but still limited

[fresh aurora]

Hiloooooo

[fresh aurora]

Hello

[distant aurora]

Hello. Hope you have a good time here.

[next dune]

The new update is good but the suggestion like create a meme for X so that I could post is not a good thing to add in your ad

It literally kills the idea of a social media platform, if you don't participate yourself and make "LLM" perform it. There are already enough bots in the social media platforms

[long fog]

Hey peeps, not sure if this is the right channel so I'm goin for it.

I recently, accidentally, deleted my account. I somehow managed to recover it using the same login but lost my pro subscription that came from PayPal. Is there any way to restore it?

[storm bronze]

I don’t really know email them Please email [e-mail] for any issues related to your account and/or billing.

@long fog @long fog

[warm lantern]

How do you cancel a max subscription ? Hi Guys a subscribed to max less than a week ago but have burned through all the credits and even bought more ! Compared to other ai services I am not getting value for money and definatley do not want to continue my subscription ,there seems to be no way to cancel it though ? in the dashboard it sends me to google play to manage it ,but when i open google pay subscription it only allows me to change the card and not cancel ? what is that about ?

[storm bronze]

I email [e-mail] for any issues related to your account and/or billing.