💬│general

[soft dock]

since when was there a limit on deep search? 😭

[granite wedge]

there is a limit on pro search aswell now I think

[glass zenith]

idk. that’s why i said do you

[soft dock]

confused af now 🐸

[zinc pilot]

THERE IS A SONET 5????????????

[twin grail]

let's trust the strawberry man

he's legit fr

u can btw

using third party model from pplx ai

I just knew this

a few hours ago

[zinc pilot]

U...........................

U...........................................................

[twin grail]

?

[zinc pilot]

dms,

now,

or

die

[twin grail]

I don't have the duty to answer ur dms yk

[tranquil orchid]

3. Treat community members with respect.

No harassment, hate speech, or malicious behavior. Keep things positive and constructive.

[crisp kestrel]

Yes lol since last week

[granite wedge]

I cant see that

[crisp kestrel]

Maybe you have outdated website

[coarse dome]

Sonnet 5 isn't real until Perplexity has it

[zinc pilot]

please.........

[coarse dome]

Can't wait to see the posts here and on Reddit where user says: "What do you mean Perplexity banned my account after sending 500 queries to Perplexity with Sonnet 5 in under 30 minutes?

[austere monolith]

[twin grail]

unless u have max

and shared with 100+ people

[austere monolith]

[twin grail]

oh hi karma

long time no see

[austere monolith]

ello Kariyoushu

yeah been busy

[twin grail]

No way

Are u

predicting this

[austere monolith]

LOL

[twin grail]

[crisp kestrel]

Heres a screenshot you can use :)

Wait let me fix the max tag

[austere monolith]

nice

[twin grail]

i want the old yellow color back ngl

the blue variant

[austere monolith]

Yellow felt better on max tho

[twin grail]

is not

[coarse dome]

No it'll be funny

[twin grail]

fitting

[twin grail]

ikr

[crisp kestrel]

done :)

[austere monolith]

on comet atleast u can change theme lol i changed to yellow there

[crisp kestrel]

Yea omg

[twin grail]

wait how

i tried to change

[crisp kestrel]

They removed it

they broke it

[coarse dome]

can you make it sauys Opus 5 too?

[twin grail]

but it still showing me the blue variant

[austere monolith]

yeah F , they broke it really

[crisp kestrel]

4.6 is better

[twin grail]

they broke it

[crisp kestrel]

It was leaked

[twin grail]

i even tried complexity

but no use

[austere monolith]

[dry kestrel]

yes

[dry kestrel]

complexity is probably going to update

since the ui change is quite new

[twin grail]

I really want the

[coarse dome]

brb about to make the sounds on Reddit

[twin grail]

yellow variant back

[dry kestrel]

hm

[crisp kestrel]

https://discord.com/channels/1047197230748151888/1461233564623306872

[austere monolith]

[coarse axle]

wtf changes with complexity, i never understood it, when i run the extension i just get the normal perplexity chat windows

[crisp kestrel]

They even removed the header

[twin grail]

so

[coarse dome]

https://www.reddit.com/r/perplexity_ai/comments/1qvvcqo/claude_models_on_perplexity_are_now_a_max_feature/

LOL

[twin grail]

we can't do anything about it now ?

[twin grail]

WTF

LOL

[wicked berry]

LOL

[twin grail]

That dude

Holy the speef

speed

[crisp kestrel]

Lmao

[coarse dome]

End of an era

😃😃😃

[crisp kestrel]

They are even making it that you cant open stylebot on the perplexity webpage in comet

they really dont want us changing it to our preferred theme

[twin grail]

so basically

there's nothing we can do... ?

Or

We can make a

wrapper for pplx

[crisp kestrel]

https://giphy.com/gifs/meme-farox-igfarox-KdO18xc1YpTippchHl

OH NO

DISCORD CHANGED ITS MAIN GIF PROVIDERS TO GIPHY

I HATE GIPHY

Tenor was better

[twin grail]

save this

[dusk oar]

lmaoo

Got mee

[granite wedge]

why

[celest light]

here's your answer

[granite wedge]

right

mb

I havent been afected yet by that update

[final summit]

is not perplexity free for students now?

[wicked berry]

Only discount around 10$ per month

[zinc pilot]

😭

[final summit]

😭

[green crystal]

@dry kestrel anthropic gives new model and oai sent me mail that they're updating google drive and one drive integration

[humble kettle]

When is comet coming on linux??

[cedar elm]

With the perplexity assistant does anyone know how to get around the academic dishonesty wall

[twin grail]

U can’t

[coarse dome]

holy crap guys my phone keeps vibrarating

Reddit and Gmail are blowing up my phone

[olive gale]

guys why did perplexity lower research and labs limits?

[green crystal]

by having it extract all the content so you can quickly copy pasta it over to another model

[twin grail]

many reason for that

[green crystal]

aravind just went all in like a wallstreet bets member

[dry kestrel]

Lol

[short narwhal]

Real

[crisp kestrel]

@coarse dome https://www.reddit.com/r/perplexity_ai/comments/1qvvcqo/claude_models_on_perplexity_are_now_a_max_feature/ its already a warzone

[coarse dome]

I started a digital riot

[crisp kestrel]

I like that half of them believe it second half fake and the 1% are yapping abt their subscription

[coarse dome]

It’s perfect

[twin grail]

Just after 2 more hours

It’s gonna go viral

And become world war 3

Trust me

@crisp kestrel

I got a

Scary image

Wanna see it ?

[crisp kestrel]

Yes

[crisp kestrel]

Ahhhh

Not the facebook for employed ppl

[short narwhal]

If only it said Sonnet 5…

[distant aurora]

I know how you use the phone.

https://tenor.com/view/dexter-gif-1878760417600646737

[coarse dome]

lol

[gloomy cipher]

Use kimi 2.5 my beloved

It's really good

[distant aurora]

Glorified ass kissing platform.

Echo chamber of lies and sham.

[green crystal]

its the domain of ai agents now

[distant aurora]

Ironically, the influence is so strong that I check LinkedIn to see if an organisation is legit.

[green crystal]

automatic ass kissing

[short narwhal]

Kimi hasn’t worked half the time for me in pplx (was the same with k2). Sometimes it answers to me and says “hmm, let me start searching” or it just outputs its answer in its reasoning

[crisp kestrel]

bruh

[twin grail]

Go for max

Why not

[crisp kestrel]

I dont have 167 $ to throw around

[twin grail]

Let it ride !!!

[crisp kestrel]

Lol

[arctic eagle]

lol

[twin grail]

Best method

[crisp kestrel]

Will claude sonnet 5 release today? Im betting no

[heavy portal]

last I read Sonnet 5 was supposed to release today, but something delayed it (per TestingComputer and/or XDA on X)

[bright glen]

Hi guys

[lean notch]

Has the main layout of perplexity from website changed? Such as positioning for model selection, in-depth research and labs

[vestal ice]

Hi everyone, I have a question for you. For the last 5 minutes when I try to past an image why does it keep telling me to go for perplexity MAX ? i have pro version... why ?

[short narwhal]

yeah new UI rolling out

[sharp pebble]

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

[coarse dome]

I think hourly and daily limits are great but weekly in monthly limits exist as a psychological tactic to make people not use a feature or rely on it. With normal use they could be suddenly locked out and they don’t even get to know the limit

[crisp kestrel]

Yea

[short narwhal]

https://vxtwitter.com/perplexity_ai/status/2019126571521761450

Perplexity (@perplexity_ai)

We've upgraded Deep Research in Perplexity.

Perplexity Deep Research achieves state-of-the-art performance on leading external benchmarks, outperforming other deep research tools on accuracy and reliability.

Available now for Max users. Rolling out to Pro in the coming days.

vxTwitter / fixvx · Feb 4, 2026, 7:11 PM

huuge

https://x.com/perplexity_ai/status/2019126586675720619

Perplexity (@perplexity_ai)

This upgrade pairs the best available models with Perplexity's proprietary search engine and sandbox infrastructure.

Deep Research now runs on Opus 4.5 for Max and Pro users. We'll upgrade to top reasoning models as they become available.

X · Feb 4, 2026, 7:11 PM

those limits are starting to make sense...

[crisp kestrel]

Welp

[prisma apex]

https://www.perplexity.ai/enterprise/pricing

look at the joke of the limits now 🤣 ridiculous. muggy move from ppx

[coarse dome]

It is good that they are providing a reason for adding the stricter limits but also it is very hard for a normal people to hit the limits that have been tested so far. I wonder if Perplexity wlll give up on Sonar and switch to K2.5

[crisp kestrel]

@short narwhal so it runs opus 4.5

https://x.com/i/status/2019126586675720619

Perplexity (@perplexity_ai)

This upgrade pairs the best available models with Perplexity's proprietary search engine and sandbox infrastructure.

Deep Research now runs on Opus 4.5 for Max and Pro users. We'll upgrade to top reasoning models as they become available.

X · Feb 4, 2026, 7:11 PM

[short narwhal]

yepp

huge

[crisp kestrel]

Hmmm

I shouldnt have wasted my research queryies

[abstract seal]

can i still use my referrals links even tho you are not pro anymore?

[short narwhal]

😭 I used 5 today as well due to it asking me what the thing I was talking about was instead of searching for it

Hope they reset them when it rolls out

[glad blaze]

Im kinda confused. Normally i can do deep research but never paired with a model

So how come it runs on claude opus

Or is that the upgrade

[crisp kestrel]

I still dont understand how they work

Bcz it just randomly resets and changes

[short narwhal]

Also study mode is now out for everyone?

[glad blaze]

Ive had that for a few days now

[short narwhal]

yeah perplexitys limits are usually on a rolling basis and regen back

[glad blaze]

(im a pro user idk why i dont have the role)

[short narwhal]

#💬│general message

[glad blaze]

Aha!

Thank u i will soon look cool

[glass silo]

Any take on how many deep research we will have each month for pro users ?

[undone kraken]

Whats Up guys

[crisp kestrel]

For everyone i think

[heavy portal]

good news about Deep Research, but is Labs only for Max users now?

[glad blaze]

Wth is it just me or is deep research queries the same as labs queries for yall

[short narwhal]

It's still for Pro too

[sonic quest]

I'm once again asking if anyone having issues with Perplexity on MacOS, with a request to update the app but it's already up to date? just asking, otherwise I'd have to file a bug report.

[short narwhal]

Shouldn't be the same

[brave valley]

Hi guys, I am new here, how do you create a support ticket on here?

[heavy portal]

where/how to select it?

[stark hazel]

Really?

[short narwhal]

On which UI?

[stark hazel]

So for deep research you can’t select models right to run research with?

[heavy portal]

I'm in Comet for Windows

[crisp kestrel]

Now we need labs update

[glad blaze]

I could swear the deep research queries for pro were a lot more

[twin grail]

Ye

[heavy portal]

ah... "previously known as"

[sonic quest]

Prince?

[silent ocean]

how do i know I got the deep research update

cause now I got 20 a say not sure if that's it

[tired idol]

./

[heavy portal]

see screenshots above, the second one is rolling-out to Pro users

[short narwhal]

I'm guessing it'll say "advanced" once it's out

[sonic quest]

I'm guessing 20 per day or something like that. It used to be hundred.

[dry kestrel]

cool

[heavy portal]

it was nice back when it actually listed how many remaining for the day/week/month/etc

[short narwhal]

also whoever took this screenshot was in a hurry and had the labs text still fading out

[silent ocean]

so education pro is just discounted pro and nothing more?

[sonic quest]

I mean, ensh*tification is a thing.

[short narwhal]

yeah

[silent ocean]

it originally was 500 then went down to 250 and today switched down to 20

[sonic quest]

Hopefully 20 per day, right? RIGHT?!? /NataliePortmanAskingFace

[silent ocean]

20 per day is great

I don't have an issue with that

[pliant solstice]

It's normal that i cannot upload images as pr?

[sonic quest]

if it's 20 per day at least, not 20 per month, then I'd might as well uninstall and remove any account I have from Perplexity.

[craggy fiber]

Where can i cancel recurring payment?

[pliant solstice]

bc i'm trying but it says to upgarde, the account is ok, and it's still on time

[silent ocean]

what?

so if it's 20 per day then good

if it's 20 per month then it's bad

which I agree with

[sonic quest]

Yes, if it's 20 per day I can still manage, if it's not 20 per day, but per month, then I'd just cancel anything I have from Perplexity.

[silent ocean]

yeah 20 per month would be ridiculous

[short narwhal]

It's either like 20, 50, or 250 a month

seeing all kinds of numbers everywhere 😭

[heavy portal]

I'd be fine with 20 per day too, curious what the labs limit is though

[solemn salmon]

My pro version cant even read my holdings from a mobile screenshot. I have to go to web version of my demat to share the screenshot. I tried the same with gemini and it caught everything correctly in one attempt. But perplexity juat didn't make it nearby to whats expected in 5 attempts.

[pliant solstice]

me too

[silent ocean]

I have 20 right now on my phone not sure if it's the new one

and I'm on pro

[sonic quest]

well, any brave souls willing to test it?

[short narwhal]

it stops at 20 right now, but not sure how they regen back

[dry notch]

I had 20 at the start of the day.

[solemn salmon]

Its 50. Lets say you use one question. That will be available for you to ask again after 30 days. Does not reset after 30th of a month. It resets after 30 days of each question has passed.

[silent ocean]

50 a month??

yeah that's dogshit

[sonic quest]

It didn't say: 'tomorrow'? or anything like that?

[heavy portal]

Read somewhere it's on a rolling 24-hour basis, if that's the case it would also be nice to have that date/timestamp shown somewhere (i.e. "next one available at...")

[short narwhal]

@crisp kestrel had it say this

[solemn salmon]

50 tokens available for use. Once you use once, it becomes 49,it will become 50 again after 30 days. Basically that very token gets locked for 30 days.

[glad blaze]

Thats fucked up but whats ur source

[sonic quest]

right, just found out that it's indeed 20 per month... now that is some grade A dog poop.

[crisp kestrel]

[short narwhal]

It's using Opus 4.5, that is more than generous

Pro Search is just as capable as old research imo

[glad blaze]

This guy is sayin 50 a month with no source

[sonic quest]

Well curious to find out when next month they'll make it only 10, and after that 5.

[glad blaze]

Sure he a bug sniper but like dayum

[crisp kestrel]

It kinda makes sense now because the classic model selector is for normal and the research is for the really complex work

[glad blaze]

Wtf why

We literally pay monthly

And how do u know

[silent ocean]

@short narwhal is the new research agentic like Google's gemini deep research thing or the chatgpt research?

[glad blaze]

Im seeing people throw 50 and 20 with no proof

[sonic quest]

Right, well I wish them the best of luck with their path of ensh*tification. If they want to be just as sub par as Claude, ChatGPT etc. then that's on them.

[solemn salmon]

Yup. Its reduced to 25 here in app.

[glad blaze]

And now 25

[short narwhal]

I don't know about that, but the previous research was, and in https://www.perplexity.ai/rest/rate-limit/all you have:
"remaining_agentic_research": 0,
"remaining_labs": 25,
"remaining_pro": 600,
"remaining_research": 20,

do what you will with that

[solemn salmon]

Last month it was 50

[sage gale]

Can someone sum the recent perplexity changes pls

[glad blaze]

Yeah for me it says 12, no proof

[twilit crown]

Who is giving AI access to time?

[crisp kestrel]

They dont have generally on their own models they use googles and chatgpts models

[heavy portal]

Odd the toggle doesn't switch between the flash and thinking versions, and the text "does not have" is incorrect.

[silent ocean]

this is not the same model

pro and flash is very different in their logic

[sage gale]

It is because they are different models

[silent ocean]

flash is linear and pro is more complex in future predictions for example

[short narwhal]

Oh and Research had this added some time back (sorry for the blurring)

[sonic quest]

Yes, that's what API calls mean. I'm not talking about their lack of quality, but the fact that they're asking more for less.

[heavy portal]

so be it 🤷‍♂️

[silent ocean]

this was interesting

I didn't really like it

it reminds me of the shopping feature in chatgpt

[glad blaze]

Yeah so be it imma not be pro no more

[short narwhal]

I liked it besides the fact that it gave me like 50 seconds only to reply and then skipped it

like dang are you in a hurry??

[heavy portal]

I kind of liked that versus how some answers now end with a paragraph like "if you can tell me about this and what you want for that then I can take this action..."

[silent ocean]

lmao

[glad blaze]

200 usd A MONTH is ridiculous bruh

[silent ocean]

did you get the new minimalized menu picker?

[crisp kestrel]

Do you pay for pro or did you get it as a free trial?

[crisp kestrel]

Yea really cool

[sonic quest]

Free trial thingy

[silent ocean]

it hasn't been released to me just yet (last checked 3h ago)

[crisp kestrel]

Yes exactly as hundreads of throusands of other people

[short narwhal]

yeah imo research should 100% ask you questions before going in. Cool part would be if it didn't use your usage at that questioning section (not sure if it does that)

[crisp kestrel]

And perplexity gives you acces to top tier ai models

[silent ocean]

can we get a shopping mode?

[sonic quest]

That sounds more like a problem for Perplexity if they were that generous.

[twilit crown]

Why does Perplexity do this?

[crisp kestrel]

While they pay the usage

[crisp kestrel]

What was the paste text?

[short narwhal]

there is some shopping feature for users in the US

[crisp kestrel]

Yea but its not really good

[twilit crown]

Minecraft crash logs

[silent ocean]

us

🙄

[sonic quest]

I'm curious if there are paying users of Perplexity getting the same nerfed usage.

[crisp kestrel]

Idk maybe it just thought

[silent ocean]

yes

[heavy portal]

both actually, having it ask questions up front and also at the end I wouldn't mind. At least we can still prompt it to do so (i.e. "ask me as many questions as needed to perform this task, at the end you can also ask me follow-up questions to refine this task")

[silent ocean]

I pay and complained many times

especially 2 weeks ago it was so bad it was unusable

[crisp kestrel]

I think they want to put pro users as free users and max as paid users

[sonic quest]

Then they get shafted the hardest, completely dry without any lube. I'd cancel the paying part instantly.

[silent ocean]

here @sonic quest

[heavy portal]

don't forget the ad-supported users (like the eventual ChatGPT Go)

[icy parrot]

Why is sonnet 5 behind max

tf

[crisp kestrel]

I think research should have atleast 100 monthly and labs could be like 15 or 10 bcz nobody uses it much

[crisp kestrel]

I would rather trade my labs credits for research ones

[sonic quest]

Then they just can remove the pro segment and the paying 'pro' users remove their paying status, since apparently 20 dollar or so per month has no meaning to them.

[green crystal]

If gpt-5.2 xhigh only loses by a few small single digit percentage points to ppxl deep research but has thousands more queries
Then gpt wins lmao

[silent ocean]

yeah but that situation from 2 weeks back was about them eating themselves without me doing anything

they just all went down to 0 from not even using them

same for pro credits

[crisp kestrel]

Idk their usage credits are so confusing

They jump up and down

[short narwhal]

whar

[silent ocean]

I demanded some clarification in emails but they haven't responded at all

[green crystal]

Look at the announcement
The graph

[heavy portal]

unpopular opinion, but I'd rather labs be pay-per-use instead of a fixed number for a fixed time period

[silent ocean]

they only gave me the Ai response guy

[sonic quest]

Yeah I wouldn't count on any support.

[icy parrot]

So whats the deal with that

[short narwhal]

xhigh does not have thousands more queries besides maybe in like the pro tier (which OpenAI is losing money on and will likely rate limit in the future as well)? and it reasons for a longgg time

[short narwhal]

fake photo

[green crystal]

Hm yeah just high for plus

[silent ocean]

pro support btw

[green crystal]

But it's 3000 per week

[silent ocean]

@short narwhal this is new

[short narwhal]

no one is using 3k a week

[icy parrot]

When will they implement GPT Image 1.5 though§

[green crystal]

I just meant value wise

[undone kraken]

guys

question

[short narwhal]

yeah was added recently #💬│general message

[sage gale]

No it isn’t

[sonic quest]

Sure, pro support, like @crisp kestrel said that Perplexity is to treat the Pro users like free users. So again I'd not count on it.

[silent ocean]

i still appear to have not gotten the new perplexity look

the minimized one

[short narwhal]

Ehhh, you should be comparing ChatGPT Plus's Deep Research to Perplexity's instead of that

[green crystal]

They didn't do that

[undone kraken]

what do you think guys ?

[short narwhal]

Same... my free tier account got it before the paid one...

[short narwhal]

?

[crisp kestrel]

But i agree with you too they change their limits how they like without telling the community

[green crystal]

It's a bit higher up lol

[glad blaze]

Deep research is genuinely the most used tool i use as a pro user

A monthly credit decrease would be horrible

[green crystal]

5.2 thinking

[icy parrot]

dxcff

[serene sand]

please contact [e-mail]

[tall rain]

New deep research?

What’s the daily limit now

[molten heart]

looks really good

[short narwhal]

20-ish a month perhaps

[undone kraken]

thanks

[tall rain]

🥀

I mean what did I expect

4.5 opus isn’t cheap

[sage gale]

It supposed to be decreased from 600 to 200 (it ok) but I think someone forgot a 0 and now it is 20.

[silent ocean]

nevermind you also get extended access to... whatever this is

[short narwhal]

exactly!!

[tall rain]

It’s more likely 20 lol

[short narwhal]

That's just this afaik

[silent ocean]

no idea why we need other pages for it

[short narwhal]

yeah idk

looks cool ig

[tall rain]

Spacex xai merger

wtf

I just learned about this

[short narwhal]

Anything to inflate that IPO

[heavy portal]

kinda makes sense as they want to launch xAI datacenters into space

[tall rain]

Is this a real thing he said

[heavy portal]

that'd be a weird mission control room of a single human with Grok doing all the other work 🙂

[tall rain]

I deleted Twitter a couple weeks ago

[molten heart]

i dont want to bother anyone, but i do have a question about the new UI of perplexity, on web

would it be appropriate to do it here?

[short narwhal]

cooling in space is hard tho so I wonder how worth it that even is

[dry kestrel]

@short narwhal do you have the new ui?

[heavy portal]

don't know if he said it, but there's articles about it https://www.notateslaapp.com/news/3560/spacex-acquires-xai-building-data-centers-in-space

[short narwhal]

on my free account only lol

[dry kestrel]

interesting

[sonic quest]

That's a very optimistic approach, lol.

[short narwhal]

nvm

[short narwhal]

just got it

[molten heart]

i wish they kept the old one lowkey

[tall rain]

I already have the new look on my iPad

[short narwhal]

I like that it says the model name in full now. The menu in the left tho is whatever

And that font still :(

[molten heart]

that part is nice

[dry kestrel]

yooo

hang on

[molten heart]

but i dont like how the sources have a extra step to change

[dry kestrel]

nope

[silent ocean]

got it

yeah same @short narwhal

[short narwhal]

I had it on my phones browser as well before on desktop

[tall rain]

How do I get the new deep research

[twin grail]

@dry kestrel

[short narwhal]

wait till they roll it out

[dry kestrel]

yes

looks very cool

[twin grail]

My prediction was correct

That’s why the limit got reduced

[tall rain]

By how much is the question

[twin grail]

Down to 50 a month

[sage gale]

Which platform?

[twin grail]

I will try the new dr tmr

[tall rain]

Dayum

[short narwhal]

Web app on desktop chromium browsers

[short narwhal]

I would not trust that site

[pliant solstice]

?

[short narwhal]

more like 20 right now seemingly

[twin grail]

It was from pplx official documents

So idk

[short narwhal]

yeah and that site is saying different numbers for everyone

[pliant solstice]

if it is a rate limit, it is absurb

[tall rain]

Icl I’m not even mad if we get 20 a month

[short narwhal]

same

[tall rain]

I got ts for free for an entire year lol

[twin grail]

The point is dr got upgraded

[short narwhal]

It's just saying they retain your files for 7 days, nothing about limits

[twin grail]

Much more powerful

[tall rain]

Yea it’s actual deep research now

[short narwhal]

"grrr give me 600 a day of opus 4.5"

[sage gale]

Think in the people that paid

[tall rain]

Not the search 20 more links type

[pliant solstice]

srry I'm spanish, but i can't upload files it just do nothing

[pliant solstice]

It's buggy right? It happens when I try to upload a file

[short narwhal]

You hit a limit for file uploads

[sonic quest]

True, 600 per day of Opus would be overkill, but 10 to 15 per day is fair. 20 per month is not fair in any kind of direction.

[coarse dome]

Omg Perplexity Reddit is still taking the bait

[tall rain]

Is perplexity even profitable? Lol

[sage gale]

It is ridiculous having less deep research than labs

[sage gale]

Probably not

[pliant solstice]

wow, even students are 💀 , and price doesn't match the student one

[short narwhal]

20 opus researches is a lottt imo

[pliant solstice]

but it doesn't say when it is recovered, like chatgpt

[tall rain]

Even Anthropic doesn’t give that much

[coarse dome]

I’m sure perplexity will be devastated that they’ve lost another user who uses stolen coupon codes.

[sonic quest]

Yeah not where I'm sitting, if they want to treat paying users like free users, be my guest.

[sage gale]

But opus was specialized in coding right?

[short narwhal]

It's a smart model in general

[twin grail]

I have 1K DR on claude opus

Love it

[tall rain]

What do you even use it on?

[short narwhal]

Pro Search gets the job done well enough most of the time for me

[twin grail]

I do research at lab

So

[sage gale]

What happens if I want more dr even if it’s with a worse model?

[coarse dome]

I am not sure but Perplexity does not train its own LLMs, they do rent capacity from Amazon web services to run otherLLMs so they will have an easier time becoming profitable then say ChatGPT or Claude. Google, META, Microsoft can all run as a loss leader. No one else can.

[dusk pollen]

Is it just me or does everyone else on "best"/auto mode always choose the sonar model?

[tall rain]

That’s the point of it

[sonic quest]

https://tenor.com/view/simon-cowell-two-thumbs-up-bravo-nice-happy-gif-12522517

[tall rain]

Same reason YouTube always default to lower quality

[sage gale]

Yeah, it is a very interesting business model. It is like perplexity is the browser and the models are the search engine.

[short narwhal]

Use your pro searches then 🤷

[dusk pollen]

shouldn't the "best" option auto decide whether question requires thinking (so it will be slower) or does not (so will be faster)?

[sage gale]

Pro searches are normal prompts right?

[short narwhal]

it can do similiar step-by-step searching as current research

[sage gale]

👍

[coarse dome]

That moodel actually has proven success it's called Google. All Perplexity has to do is charge enough money and make sure the searches per month don't cause them to lose money.

[sage gale]

Which model do you recommend for that? Gemini 3 pro?

[short narwhal]

Gemini 3 Pro or GPT-5.2 Thinking

[coarse dome]

Not exactly. Perplexity uses retrival augmented generation. They have their own search algoriithm which runs on its servers and gets snippets of text from the articles that are important. Your prompt and the search data is provdied to ChatGPT , Google, whoever to generate a final response with a detailed prompt from perpelxity before yours

[sage gale]

Yes but this have AI

[coarse dome]

Yes paying other people to run LLM queries is much cheaper than hosting a LLM yourself

Perplexity does run a few open sourc emodels too

But traiuning and having AI engineers build their own model would be very expensive

Their core product isn't an LLM, it's the research tools they built

[shadow lily]

@vague pagoda respectfully I’m sure you know it I know I am a loyal perplexity pro member for almost a year I recommended it to all my friends you know the service has gone down

Now look I keep trying to use it in hopes it will be back since I paid for a one year plan

So if it’s not can you find a way to tell me either anonymously or in here or a dm whatever way you can tell me the quality will improve or not

EVERYONE HERE HAS SEEN HOW HORRIBLE IT IS NOBODY CAN DENY IT

[gentle frigate]

About the (new?) Research mode, I also have a monthly limit of 20 with my Perplexity Pro subscription. I've been trying to figure out what's going on since that limit is MUCH lower than the previous limit. To try and find out what happened (and find alternatives if need be) I used the normal Perplexity Search mode as a Perplexity Pro subscriber. I noticed that if I included in my 'Search mode' prompt to do Deep Research and give a long, detailed response, Search mode behaves similar to what I'm accustomed to with Research mode of maybe a month ago. More recently, 'Research mode' responses became extremely long. Very well structured and very detailed, but an extremely long read. The Conclusion was the most helpful part of the whole response, albeit at the end of the whole response. Asking for slightly less verbose responses didn't do much. So it seems in my experience Research mode got renewed, with a very long detailed response similar to Gemini Deep Research.

I'll spend the next few days prompting for Deep Research style responses in Perplexity Search mode (the one with much higher usage limits) whenever I use Perplexity, to see if I get responses similar to the responses I was used to when using Research mode until approximately one/two months ago. If it appears to be a big downgrade however whilst the subscription price stays the same, I'll try out Anthropic/Claude Pro subscription as an alternative.

[shadow lily]

Give it up ngl unless he replies seriously

😭😭😭

ChatGPT research is a lot better atp

If I’m being honest

I shilled perplexity so hard only for them to betray me

[twin grail]

That’s not the case

[shadow lily]

They lie a lot

😭😭😭

[twin grail]

Also u got claude opus for deep research now

[twin grail]

No?

[heavy portal]

I still wish they had a bring-your-own-api-key (maybe they do and I missed it) ability rather than have query limits.

[shadow lily]

You are right I am a gold fish

Sorry for questioning you

[twin grail]

I used max and they added claude opus for awhile

[shadow lily]

[lament imp]

Whatchu going on about?

[twin grail]

They are rolling it out now

[shadow lily]

Perplexity is horrible and everyone knows it and should I keep trying and hoping and praying that it works

Will it be restored to its formal glory

That’s my question

Former *

[lament imp]

How has it changed?

[night phoenix]

Check your rate limits

[shadow lily]

Bruh do you work for them

[night phoenix]

They gutted rate limits and made some features Max only

[lament imp]

"Bruh" I'm asking a question. You haven't provided any clarity to what you are saying. I am curious as to why you think that way

[shadow lily]

Are you a part of perplexity’s team

[true sphinx]

now WHY have they changed the menu

[shadow lily]

Trying to appease investors basically they blew through all the investors money and are now trying to recoup

Investors are pissed

[night phoenix]

The same reason anything ever gets worse: Money

[shadow lily]

Told works for them and trying to deny it

Gwolf*

Bruh literally hire me if you want to make it look real gwolf sucks

[true sphinx]

yeah its about money

[shadow lily]

Fr

[true sphinx]

😔😔

[shadow lily]

Gwolf all these people wrong ? Or is the ceo screaming down your neck 😭😭

[lament imp]

Are you only going to ask a question in response to a question? Seems more like you are trying to avoid providing logic to what you are saying.

I am simply asking: How has it changed?

[shadow lily]

Simple question do you want me to go on Reddit and review all the posts calling perplexity shit ?

[coarse dome]

They might for enterprise customers but for normal users I don’t think that’s a thing. But there are plenty of open sourced Web apps which you can create an account on and store your account keys in I believe it actually uses your Web browser as local storage so not even a server operators store copied your keys and I’ve almost everything goes directly to the AI provider so it’s super secure but running API queries with your API keys can get extensive to keep that in mind

[shadow lily]

I’ll send you a picture if it appeases your heart

GWOLF AKA PERPLEXITY SHILL

he a spy a informant

[lament imp]

If you go into any subreddit for any AI company it's only people complaining about the company...

[shadow lily]

Perplexity’s are alarmingly high and everyone in here complains

Listen tell your ceo to hire someone better

You suck

Respectfully

You should be countering me by asking difficult questions to which I do not know

And that’s how you make someone look stupid

Instead I’m countering you

[lament imp]

Your argument is "I think Perplexity sucks and has changed because people on Reddit have told me so".

👍

[shadow lily]

Does the ceo approve of your messaging

[heavy portal]

https://tenor.com/view/jon-stewart-eat-eating-popcorn-watching-gif-12567454887078541885

[shadow lily]

Because I think you suck if you are internal team member

SELL UR SHARES BRUH

sell them

Perplexity is GOING TO 0

[worthy sage]

I’ve not gotten a response from perplexities customer support when I wrote a email several months back complaining about the new update to the software/app on my other phone. Unfortunately, that complaint landed on deaf ears and now they’ve updated my current phone with the shitty update.

I emailed perplexity support about a the issue with the current updates. I personally believe the old talk button was way more efficient because when the AI is giving me a long winded answer or misunderstood the question,
I could cut it off. Now I have to wait till it’s done with its dumb ass answer before I can ask another question or clarify the previous question for the AI.
I’m afraid this update is going to result in me abandoning perplexity as an option for an AI when I’ve been enjoying it. It is currently the only AI I consistently pay for each month now I’m choosing not to.

Attached is the email below. If there’s any way for anybody in this chat to forward it to customer support, I would appreciate it.

Urgent Feedback on Recent Update – Request for Reverting to Previous Layout

Dear Customer Support and IT Team,
I am writing to express my deep dissatisfaction with the recent layout update. The previous version of your software was far superior in terms of ease of use and user experience. The new rotating sphere feature does not function properly and disrupts the flow of conversation. Users need a reliable, straightforward button that works consistently.
This update leads to overly long responses, and it is difficult to interrupt or control them. This negatively impacts the overall experience and will likely cause a significant drop in your user base.
I strongly urge you to revert to the previous layout or provide an option to switch back. User-friendly design and smooth interaction are crucial. Please address this issue promptly to avoid further frustration.
Thank you for your attention.

Sincerely,
Lex

[shadow lily]

stay on as a “consultant”

[coarse dome]

You might like this

https://vercel.com/docs/ai-gateway/chat-platforms/librechat

[shadow lily]

Lmfaoo

[merry meadow]

can someone tell me how to cancel my subscription pls

[shadow lily]

Gwolf the hired shill says your lying and everyone complains 😭😭😭

[merry meadow]

like I dont think I have the option even

[shadow lily]

Go to India and find the ceo there yelling at his employees and hiding away from investors 😭😭😭

[coarse dome]

It should be under view details. Sorry to hear you go.

[lament imp]

Go to View Details

[shadow lily]

Builder io rug pull

[shadow lily]

Good employee

[merry meadow]

when I go under view details it just brings me to this

and I have only ever payed 5 bucks

[shadow lily]

Gwolf I’m telling u bruh leave this company

[merry meadow]

it only charged me once

[shadow lily]

Or charge them a higher fee

I’m telling you truthfully this company is shit

[merry meadow]

like there is no definitive cancel subscribtion

[dry kestrel]

Press update subscription

[short narwhal]

It is cancelled already

[merry meadow]

yea it takes me to this

[merry meadow]

wait its canceled?

[shadow lily]

Don’t pay it shit

[coarse dome]

Here is how!

https://www.perplexity.ai/search/bdf8536f-4af5-4b02-9e83-544af88da5d7

[shadow lily]

Aravind Srinivasa IS THE GOAT SCAMMER

[lament imp]

On the left it says it will cancel Nov-7

[shadow lily]

better than adam Newman

Aravind Sriniva=goat scammer

[merry meadow]

yea but i dont want to pay the 200 lol

like at the end

[shadow lily]

Don’t bruh

[merry meadow]

yea so how do I cancel

like even if you delete the account

it can charge you lmao

[shadow lily]

Just charge them back if they fuck you bruh

[short narwhal]

You've already cancelled it

[merry meadow]

what type of BS is that

[shadow lily]

They gonna scam regardless

[merry meadow]

thats doo much work

[dry kestrel]

This is false.

[shadow lily]

You a paid shill also ?

[lament imp]

That is weird, when did you pay the $5?

[merry meadow]

arvind is p chill

[dry kestrel]

I don't work for Perplexity

They don't pay me

[merry meadow]

I met him when he came to talk at my university

[shadow lily]

I know he’s a g

But he’s a goat finesser

The investors were super horny off ai

So he capitalized and scammed them all

[merry meadow]

i wouldve done the same

[shadow lily]

Fr

[tranquil orchid]

1. Keep the content appropriate.

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

[shadow lily]

How can I hate human

[merry meadow]

some VC's are fish lol

[shadow lily]

How can I hate him

Yeah bruh

I don’t hate the guy

Dude is the goat

[dry kestrel]

It was never 600.

[merry meadow]

it used to be like 200

[coarse dome]

Bro I wanted Gemini to find the video not wathc it for me!

[dry kestrel]

Please do not mass ping. You may post in #1111765634267742408

[shadow lily]

Watch he’s going to exit this scenario as the one who was 100% legit but his company failed because of “investors misnegiligence”

Not enough “due diligence”

[keen root]

That may be the case but I have never run out of them. But 20 per month, means I will go through them in a week if not less.

[merry meadow]

i mean he is pretty smart, I think if some people make some investments without 5-10 year cash flow scope its on them

[shadow lily]

Bro HES DONE FOR LIFE

[merry meadow]

nah he is gonna run some new human ai lab method like the rest of them

[shadow lily]

LIKE HE CAN LITERALLY LIKE WATCH ONE POECE FOR 200 years and still be rich

[short narwhal]

You can use Pro Search which functions very similiary to the previous Research

[shadow lily]

Nah fuck that I’m watching one piece

[merry meadow]

bro alr has the one piece

[shadow lily]

I get a few mill I’m going to oda

Holding him hostage

Getting the finale

I can’t wait this long

[keen root]

Is that so? The research was accurate and got you around 50 - 100+ accurate references per search. Will the new pro give the same?

[shadow lily]

Ngl I’m really impressed with Jake Paul’s investments dudes actually the goat

[short narwhal]

Yeah not impossible at all

[pliant solstice]

just a question how much time it is required to upload files again after you've been rate limited

[dry kestrel]

hmm @short narwhal https://x.com/sama/status/2019139174339928189

Sam Altman (@sama)

First, the good part of the Anthropic ads: they are funny, and I laughed.

But I wonder why Anthropic would go for something so clearly dishonest. Our most important principle for ads says that we won’t do exactly this; we would obviously never run ads in the way Anthropic

X · Feb 4, 2026, 8:01 PM

[keen root]

There is no more pro search though. It now is just search and deep_research? The sonar model itself never was all that accurate. And none of the other models got the same accuracy range as the research one did before. Is there even anything that pro users get extra from free users besides the 20 searches per month?

[stark rapids]

that first sentence makes him sound like a robot

[short narwhal]

Search is the same as Pro search

[still star]

whats happening with this update, i can't use the app (says update required) but there is no update available in the app store

[tropic fiber]

Lo que no me ha gustado del navegador de perplexity / Comet es su actualización automática. Por ejemplo, si estoy llenando un formulario y cambio la ventana hacia un documento se pierde toda la información que ya había ingresado en el formulario porque este navegador hace un reinicio automático y no le veo la finalidad

[short narwhal]

he's not wrong, anthropic loves to play dirty

[formal birch]

#👑│pro-lounge message

[short narwhal]

and those ads were funny and good too

[dry kestrel]

Mhm

It was humorous but portrayed in an ill manner

[raven marlin]

What would deep research even be useful for?

[short narwhal]

deep research

[dry kestrel]

Excels at expert-level analysis across complex subject matters including:

Finance

Marketing

Technology

Current affairs

Health

Biography

Travel planning1

https://www.perplexity.ai/help-center/en/articles/10738684-what-is-research-mode

[short narwhal]

the classic travel planning example

[keen root]

Sorry it's not. The search doesn't even look for academic sources anymore and get's around 20 references. You people truly screwed us over. I used it mainly for it's accurate meta-review searches with academic sources. And now I can't do that anymore. Or at the least only a few times a month..

[twin grail]

@dry kestrel

[dry kestrel]

nicee

[twin grail]

I will test it out

[coarse dome]

okay wow perplexity

They removed research and labs from the website

it'sjust gone

[twin grail]

no ??

[stark rapids]

click +

[twin grail]

there u go

[stark rapids]

Deep research

[coarse dome]

they renamed it

[stark rapids]

and?

personally im not a fan of moving the buttons and hiding it behind the + because you force people to look for stuff. maybe they did it to force people to discover the other tools within the +

[dry kestrel]

It was first Deep Research when launched, renamed to Research mode, then reverted

[coarse dome]

but now it's Deep Research

[royal estuary]

Guys does perplexity use vector databases

[coarse dome]

Its possible

[royal estuary]

Doss it api support vector db or only sql

[shadow lily]

bruh they dont even pay these moderators iheard

no wonder the customer support so bad

💀

they been waiting like 3 months

[dry kestrel]

OpenAI doesn't pay their community staff either

[shadow lily]

goat scammers ngl

how can u hate them

i dont hold amnosity well i do but like at the same time

i admire them more than i hold amonisty

[coarse dome]

why would they when they can make a few people think they're special for free 😄

[shadow lily]

L

🙁

welp

[tranquil orchid]

1. Keep the content appropriate.

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

[coarse dome]

No limits on iPhone another reason to pick Apple a leader in Artificial Intelligence 😎😎😎

[gilded vigil]

nonsense.

[coarse dome]

Do you see a li
It cause I don’t

[dry kestrel]

limits are shared across devices and has nothing to do with Apple

[coarse dome]

Pretty sure if limits affected iPhone they would tell you

[shadow lily]

ngl we should find a way to finess perplexity

lets find their investors and say they doing some lazy shit

the blackmail them

idk

better than just complaning in here

[tranquil orchid]

4. Provide constructive feedback.

Make sure your feedback is specific and actionable to help us improve the product and this Discord. Vague negativity isn’t helpful to anyone.

[shadow lily]

pay me so we dont blackmail the investors

or get your ceo to get me connectons to dumb investors

either or

and then help me finess them

[coarse dome]

https://www.reddit.com/r/perplexity_ai/s/pLIeSiEFf5

Apple best AI platform

[dry kestrel]

This is false.

[coarse dome]

Do you see a limit because I don't

[dry kestrel]

Does ChatGPT show you the limits? Does that mean there's no limits?

I don't see a limit counter in Gemini - does that mean there are no limits?

[drifting sinew]

Hello everyone!

[green crystal]

https://x.com/i/status/2019139174339928189

Sam Altman (@sama)

First, the good part of the Anthropic ads: they are funny, and I laughed.

But I wonder why Anthropic would go for something so clearly dishonest. Our most important principle for ads says that we won’t do exactly this; we would obviously never run ads in the way Anthropic

X · Feb 4, 2026, 8:01 PM

Sam hypeman got a point here

Claude free user base is almost non existent

[coarse dome]

it's likely perplexity free its too limited to do aything

[green crystal]

Chatgpt even my plumber knows chatgpt

[coarse dome]

I believe if you have the ChatGPT Go or Google Plus plan they do

[green crystal]

I haven't used free Claude but hearing how paid tiers are so rate limited surely means that free tier is hellish to work with

[coarse dome]

I'm a Google AI Pro user; also Education Pro is now $9 so I'll klikely be cnaceling after my free tiral urns out lol

<@&1098365634498285709> There's a spammer

Spammer info

[dry kestrel]

This is false

[coarse dome]

They disclose the message limi8t on the plans page

[dry kestrel]

That has nothing to do with what you said before

[coarse dome]

https://help.openai.com/en/articles/11989085-what-is-chatgpt-go

[silk palm]

Hello I am pro user but perplexity stopped generating me images... because of region resttictions...

[coarse dome]

I would recommend talking to your local government they set the restrictions.

[agile maple]

Am I the only one who doesn't like the new design changes to Perplexity?

[vale temple]

I thinks it’s ok ish

Some parts of the UI go off screen though

[silk palm]

I am living in Spain... I didn't hear about any restrictions

[coarse dome]

Talk to the European commission who has set many regulations on AI companies that they have to follow to operate in your country

Unless they back down the European economic area will continue to face restrictions from United States trch companies

[silk palm]

But why today? I didn't hear anything also I am able to generate with grok or chatgpt why not with perplexity

[agile maple]

In Spain, if it lets me

[silk palm]

Yes I am living in Spain

[agile maple]

It does let me create one, and I haven't heard of any restrictions either.

[unkempt kindle]

Dope

[meager rover]

So true

[silk palm]

I am starting to think that they are limiting from the shadow pro users

[meager rover]

Prolly

[silk palm]

I don't think we will have perplexity for a more long time It's starting to die slowly

[green crystal]

It's not just EU iirc
People from other countries also had issues

[rocky yarrow]

Yo what the hell lol.

Logged on today and it says I have 20 deep research queries a month.

Is that a glitch?

I'm gimped at work rn what's going on. Will they push a hot fix soon?

[meager rover]

That’s how it is now, they made it “SOTA”

It uses opus.

Better model means more cost which means more rate limits

[rocky yarrow]

What the hell are you serious

Did they make a post about this? I probably run 25 deep research queries a day.

I've never even received perplexity for free, I've been a paying customer for over a year.

That's gut wrenching if they just capped my usage to like less than 1/10 of what I used before.

[heavy portal]

I could (and probably am) wrong, but this could be an over-reaction and they'll settle on a middle ground between the old limits and the new limits. It'll be tough until then, but every company goes through this.

[dry kestrel]

They do not show the limits in real time. What you sent has nothing to do with what I said

The limits are listed clearly by Perplexity, just because they don't show it on the iOS app doesn't mean its unlimited.

Stop spreading false information.

[earnest egret]

25 deep research per month?!?!?!?!?! are they nuts?

[dry kestrel]

Where's your proof?

It should be easy for you to prove

That's not proof

Where does it say "No limit"?

Where does it say "No limit"

Proof?

Where does it say "No limit"

How about you find some proof where it says limits don't sync across devices?

[earnest egret]

but they change this limit like twice a day,
to this day i had like 200deep research per month, now i have 25 hard cap(now 0)
its just unfair, you buy something and they change it like that

[dry kestrel]

Why doesn't it say "unlimited" when it shows for Max users?

Why does it say unlimited for max but not for Pro?

[pseudo ferry]

https://tenor.com/view/harvey-specter-suits-suits-tv-suits-usa-pilot-gif-8279369896926227136

[dry kestrel]

Do you see a limit? Does this mean Grok is unlimited?

You are wrong then https://www.reddit.com/r/grok/comments/1o62blt/what_is_the_daily_limit_on_free_grok_account_and/

Does this mean it's unlimited?

You are wrong

https://support.google.com/gemini/answer/16275805?hl=en

30 prompts isn't a lot

[balmy radish]

is there a support channel?

[dry kestrel]

What do you need help with?

[balmy radish]

basically last year, i redeemed a $200 perprexility giftcard and i had to redo my billing recently

[dry kestrel]

Do you think other people agree with you?

[balmy radish]

at the end of the 1 year is there a way to automatically cancel it to not get charged?

[dry kestrel]

I don't think so, you can always get a refund if this accidentally happens.

[balmy radish]

oh okay? what are the steps so i could be prepared?

[dry kestrel]

I don't see a limit counter, is it unlimited?

[dry kestrel]

[e-mail] https://www.perplexity.ai/help-center/en/articles/10354288-refunds

[balmy radish]

thank you!

[dry kestrel]

Why?

I don't see a limit counter

on my iPhone

I don't see a limit counter. Is it unlimited?

[distant aurora]

This better be a good upgrade.

#📢│announcements message

[dry kestrel]

It's powered by Opus 4.5 - it's a great upgrade

[distant aurora]

Rolling out to pro soon. Hmm.

[distant aurora]

That's for ai plus. For free users it's like 3 pro per day.

[dry kestrel]

I was talking about Ai plus

[distant aurora]

Oh. Okay.

[dry kestrel]

It's Claude

Is it unlimited?

[distant aurora]

Claude and their atrocious limits.

[rough stag]

Thank God (maybe that's a poor choice of words), I thought I was tripping.

[potent silo]

Has anyone else's deep research queries go down alot

For the pro version

[rough stag]

I saw it go down to 13.

[rocky yarrow]

In convinced it's a bug

[potent silo]

Mine went from 20 yesterday to 3 today i barely used it

[rough stag]

It was 19...I ran a few deep research queries and it went down to 13.

[rocky yarrow]

Has the team here addressed it?

Like I might have to do a charge back on my visa because I feel scammed

[brisk hazel]

Hello , I am an MBA-Student and now I am start using pro for doing my independent study and research for my paper. Is anyone have any recommendations on how tonuse edficiently the research ? I means tips and tricks for faster and accurate, research.

[mental dew]

😭😭😭?

[potent silo]

LOL

[dry kestrel]

#💬│general message

[rocky yarrow]

I'm just kind of annoyed about this because I have been a paying customer for so long, and I never got any Pro for free, but it feels so unfair to me that they would downgrade Pro after giving it away to tons of people

[mental dew]

Tf is going on 😭

[random summit]

It happened to me several times too. Just ignored it.

[shut gyro]

Perplexity trying to speedrun bankruptcy

No warning. No indication of how many deep research prompts you get. Just "UPGRADE FOR MORE!" 😅

[dry kestrel]

@coarse dome

[shut gyro]

By all means offer a higher tier, but don't do it by degrading the lower one.

Yes

[quartz wave]

This MONTH now? They dropped from 600/day to 25/day, got backlash and put it to like 200/day. And now this?

I forgot I set my name to this here but... still works ig

Pro is 100% a scam and I need to double check but pretty sure at this point has violated a bunch of laws in my country so... easy chargebacks ig

[twilit crown]

You guys actually used it?

I found the normal mode better

[quartz wave]

Research is so much better for... most things, not all.

Well. At least their AI model narrowly avoids calling it a scam.

In the legal sense, it is not a scam as they can (under most jurisdictions anyway) change the usage pretty freely. But I would say in the colloquial sense, it very much is a scam as they don't announce it ahead of time. They want people to pay and not notice or not bother to ask for a refund, meaning they get away with giving a very small % of what people actually paid for.

Specifically this much less

[foggy cradle]

hi

new guy

here

[quartz wave]

Honestly since the limits only show accurately on mobile, I don't know how much I used before seeing this. We only get 50/month if it is the same as the Enterprise Pro level. Perplexity's own post on it hides the numbers for Personal Pro. So saying they cut 99.7% of the value and not 99.85% as I suspect is giving them the benefit of the doubt.

[quartz wave]

Hello!

[naive sage]

Pro is usually half of Enterprise Pro's limits. But the pricing page for Enterprise Pro lists 250 Research queries per month for Pro...

[quartz wave]

I don't think that's up to date as I was looking at a page updated today. Give me a moment, I'll find it again.

[short narwhal]

Pro search is basically capable of everything that research was capable of previously. And now research got improved and uses Opus 4.5, which is wayy smarter but also wayy more expensive

[quartz wave]

https://www.perplexity.ai/help-center/en/articles/11187416-which-perplexity-subscription-plan-is-right-for-you?utm_source=perplexity

[short narwhal]

Pro has 20 a month for research

[quartz wave]

Wow. Even worse than the 25 I suspected

[green crystal]

they could've used cheaper models

like 5.2 and gemini

[short narwhal]

They’re not opus tho 💅

[green crystal]

those 2 are very close to opus for allot of things

[green crystal]

[quartz wave]

Both have been improved. I agree. If it had stayed at 200 or maybe even 100/day, I could understand. Or if they announce it ahead of time instead of sneaking it into articles? Sure. But pro search is still much more surface level and covers less pages than research did before for it absolutely does not make up for changes

Plus if I understood correctly, we aren't even unlimited on searches anymore. Though it does not show a limit

[dry kestrel]

Best mode was always unlimited

The other models had limits before as well

[quartz wave]

Yes. But I'm saying if I understood their intentionally vague word soup correctly, it might not still be unlimited. I don't really know though.

Perplexity truly is the least honest out of the frontier options. That support article I linked above is so intentionally unclear on what it even means and was updated silently to avoid users noticing.

[dry kestrel]

Best Mode

For quick searches, this default mode intelligently selects the most appropriate model based on your query type. “Best” mode is available without quota limits.

https://www.perplexity.ai/help-center/en/articles/10352901-what-is-perplexity-pro

[heavy portal]

can't find it at the moment, but someone earlier posted a URL one can visit and shows (I think) JSON listng the limits for Pro searches, research, labs, etc

[dry kestrel]

It says clearly in their documentation

that best is unlimited

[short narwhal]

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

[quartz wave]

That wasn't updated today like the other one was.

[heavy portal]

Thx!

[dry kestrel]

updated this week isn't enough?

[heavy portal]

That link for me shows "remaining_pro":600 but doesn't specify a model, so is "Best" really that unlimited?

[quartz wave]

When they changed the limits of multiple AI tools in the past 24 hours? Not particularly.

[naive sage]

It is rate limited, but that 600 recharges rather quickly... generally a minute after the query, it's back at 600.

[dry kestrel]

The updated limits were there this week lol

People were talking about it a couple days ago

[rocky yarrow]

Aza

I'm honestly mostly sad because I'm such a fan of perplexity, and I have recommended it to so many of my friends and family.

[quartz wave]

Yeah and they stagger changes. Which is why not everyone had the 25/day limit when it was being discussed here. "lol"

[rocky yarrow]

I even bought a sticker pack and hat from their store lol

[dry kestrel]

I don't see an issue with the documentation updated this week

It's up to date

[naive sage]

I feel bad for the documentation people... they probably feel whiplash from all the changes lately, and now they need to go update the pricing page[s] again, since the primary Enterprise pricing page is now wrong again.

[quartz wave]

It may be. But it's hard to confirm that. But even if you were entirely correct, it wouldn't refute the point of this topic at all, which is that they cut 99.89% of our Research usage. You can say Research was improved, Pro Search was improved. But they hid the changes hoping for people's payments to go through and that they won't ask for a refund/charge it back.

[opaque stump]

Using it with Gemini 3 and my limit refreshes very quickly to 600

[dry kestrel]

I'm not talking about that though, I'm saying that only best mode was unlimited

I said nothing regarding Research

Plus how is that not up to date?

It's been 3 days

[quartz wave]

Glad you agree then

[novel arrow]

Guys i am on max and it's heaven not to brag but it's really day and night compared to pro

[opaque stump]

How do we know Pro is better? All I’ve seen in their posts on the updated Deep Research and nothing on Pro

[dry kestrel]

I'm not agreeing or disagreeing with the other limits, I'm talking about your comment on models being unlimited which was partially wrong.

[heavy portal]

This probably explains why I've never encountered a limit-reached message as I usually leave it set to "Best" for everything (including shortcuts). A few times I've changed things to Claud Sonnet only to come back later and see it reset to Best. Don't know what's up with that.

[quartz wave]

Well, tbh. Anecdotally I think it has improved but yeah... haven't seen solid proof.

[quartz wave]

You literally just said only best was unlimited. I also said that. So I pointed out that we agree. And now you're like "Waitwaitwait no ewwie >.<* I don't wanna agree with you"

[dry kestrel]

Lmao? You responded with saying the documentation was outdated

[quartz wave]

If you agree that only Best was ever unlimited... then cool. We agree on that. If you don't think only Best was unlimited then we don't agree. Which one do you pick? It was funny the first few times. Now it is getting annoying.

[quartz wave]

Even if that weren't stretching what I said, it wouldn't be relevant to whether or not Best was unlimited.

[heavy portal]

Flash or Pro(Thinking) ?

[dry kestrel]

You added

You can say Research was improved, Pro Search was improved. But they hid the changes hoping for people's payments to go through and that they won't ask for a refund/charge it back.

I clarified saying I wasn't commenting on that last section

[opaque stump]

Flash Thinking. Same thing happens with Sonnet and Gemini Pro.

[quartz wave]

Unless you can directly state where I said other models were unlimited, I am not feeding this trolling further. It is literally "I think Best was the only one that was unlimited" "NO YOU DONNTTTT I THINK THAT OMLY ME"

Anyway. Done with that guy lol

[dry kestrel]

Plus if I understood correctly, we aren't even unlimited on searches anymore. Though it does not show a limit

[autumn raven]

Exactly this point

[dry kestrel]

Documentation supports this. Other users report of seeing this as well.

[opaque stump]

Yea. Doesn’t seem to be any different to me so I’m sticking to that until I see proof

[heavy portal]

this limit talk, is it only about for Research? seems multiple conversations here one of which about the 600 resetting quickly (almost) regardless of model.

[sturdy willow]

I'm bummed because up until yesterday, I was using "Research" to do job analyses (copy-paste job description, and have it actually look at my resume and the markdown file with specific instructions on how to do it), and it was GREAT at that, then next thing I know I used Research a few times today for some more jobs and BAM I'm all out of the "deep research" allotment.

[dry kestrel]

Was talking about "best" mode.

[quartz wave]

It is just so strange that people would defend a 99.89% decrease. Especially when done in such a shady way.

[dry kestrel]

I never commented on the decrease in Research usage

[quartz wave]

Research is the one we were discussing decreases to.

[naive sage]

It's going to be based on a much pricier model, for one... they said it uses Opus 4.5 now.

So... some restructuring makes sense when they release an entirely new model

[dry kestrel]

Hope that they update the FAQs to clarify some things

[sturdy willow]

I think someone said "Search" was improved? A few days ago when I tried to use "Search" instead of the "Research" mode it basically ignored the attached instructions (saved in the "Space") and my resume and just started doing an independent analysis of the job itself without any context and that's obviously useless to me.

[naive sage]

That said, massive Deep Research limits was one of Perplexity's strongest selling points... maybe there should be an option to select the old one too, but running both might not be feasible depending on their infrastructure.

[opaque stump]

I think they underestimated how mad people got with the 200/week and my hunch is they’re going to go 600/week.

[sturdy willow]

I like Opus a lot (I have a separate Claude subscription just to get access to it), but not at the expense of losing access to the former "Research" model, now I'm really bummed.

[quartz wave]

Literally before you even got here... that is what this has been about. You came in and said "Umm acshully, only Best was unlimited" and I agreed and you got mad because I said there was no clarity yet on whether or not they had changed that. I said it sounded like they had but that it was unclear and may not be the case. Then you kept getting mad when I agreed with you about stuff.

[naive sage]

With Sonnet 5, I wonder if Opus will also see massive efficiency improvements when Opus 5 comes out

The rumor is Sonnet 5 will be up to 50% more efficient

[quartz wave]

You are literally only here to try to incite drama and I have 0 respect for anyone like that.

[opaque stump]

Have you tried using Opus on Claude for the job listing task?

[lapis hamlet]

tell your manager or ceo that comet assistant should be unlimited for pro and max user's

[dry kestrel]

I don't work for Perplexity

I wish the limits were higher

[dry kestrel]

Opus 4.5 with Research mode on Claude is decent, just hard to use with the rate limits

[lapis hamlet]

i don't want to copy paste by myself so i can go watch smth else

[sturdy willow]

Now that the old "Research" mode on perplexity's gone or at least SEVERELY limited (I believe I accidentally burned through my quota for the month without even knowing), I'm going to try that, yeah.

[quartz wave]

Staff are effectively inactive here.

[lapis hamlet]

not for pro user is still nott worth it

[dry kestrel]

The Pro plan on claude is quite limited which is unfortunate

[sturdy willow]

yeah it's the rate limits on Claude that bum me out the most.

[lapis hamlet]

i was reported a bug that i am in goverment account for no reason and keksu just dm me

[dry kestrel]

It's annoying that Research mode is token based instead of query based on Claude

[quartz wave]

Interesting. I never even use Claude. Well not for Perplexity

[sturdy willow]

As someone in #👉│feedback-general said, it feels like we're stuck between "low effort" and "MAXIMUM HIGH EFFORT" where "Research" (before the update) was a perfect Medium tier between the two and now it's gone? I think?

[opaque stump]

I have Max. Claude is my most used LLM for work.

Between Gemini and Perplexity for web search now

[dry kestrel]

Yep, the Max plan is nice, I just find it a bit slow at times for Research

[quartz wave]

Ah. Well I never really see them around. And my point was just that unless it says people are Perplexity staff, they're not. Berd isn't, I'm not, etc. It would show in roles like Kesku has.

[lapis hamlet]

you should use your 200 dolllar to buy a chatgptpro

[opaque stump]

I have the in-between $100 max plan

[autumn raven]

Same here. I've just voted with my wallet and cancelled my premium membership. I recommend everyone else does the same that's the only language they will understand.

[dry kestrel]

Staff just solved a bug for users #1468705567819301222 message

[opaque stump]

More AI companies should offer that

[dry kestrel]

3 hours ago ish

[dry kestrel]

True

[quartz wave]

Not to state the obvious, but bug reports is an entirely different channel. So my point still stands.

[lapis hamlet]

btw i tell him about the information he need from me and he give me a student discount(5 dollar per month)

[quartz wave]

That's good

[autumn raven]

The more people we can convince to leave and cancel their subscriptions, the more likely they are to reinstate it. I'm South Indian myself and I understand the culture... The only language important to us is money. This was a very financial decision without giving a shit about customers. If we take it and bend over, it will be seen as weakness and they will continue to do this

[lofty shadow]

I’m in a similar situation but reverse, I went to perplexity for the $5 a month student discount, saw that they raised it to $20 and thought “hell naw, I’m switching to ChatGPT” lol

[autumn raven]

I'm also a coder, and a similar age to the founders... Ultimately creating a mass exodus is the only way to ensure they return the old limits back

[quartz wave]

Unfortunately last time I tried that didn't have much success. I don't think the limits are coming back. But hey. If they wanna get overtaken by Kimi, that's great. Kimi is way better.

[opaque stump]

Kimi does web search?

[quartz wave]

Ye

[dry kestrel]

Didn't specify when you said "here"...

[heavy portal]

I asked this awhile back in a Comet channel here, but those channels seem dead. When it says it got disconnected is a Comet issue or a model issue? I recently changed a shortcut from Best to Sonnet 4.5 and it disconnects fairly quickly, then changed it to Gemini 3 Pro and it has yet to disconnect.

[quartz wave]

My apologies. I thought most people would agree that "not here" and "here" are different things

[sturdy willow]

Claude Opus is probably undoubtedly the best for $20/mo you just run out of tokens real quick.

What do you guys use as a replacement for Perplexity's old "Research" model since apparently we're stuck with only 20/mo now?

[dry kestrel]

I assumed it was the server overall

[dry kestrel]

I've had it happen less this week, weird that it changes with the shortcut model selected

[quartz wave]

Well don't assume. It makes you look silly. Especially since I keep specifying that I don't wanna talk to you. First time I've had to block someone in like 4 years lol

[heavy portal]

thus my question, don't know know if it's coincidence or related.

[dry kestrel]

Afaik the model selected isn't the one that Comet Assistant uses which is weird that it's affecting the disconnection issue

[dry kestrel]

Should have been more specific

[heavy portal]

Yeah, this all happened in the past hour. I'll try Sonnet 4.5 another day and see if it repeats.

[opaque stump]

I’m a fan of Gemini’s Deep Research. ChatGPT’s isn’t so bad either.

Perplexity’s is lightweight compared to the output I get from the other two

[dry kestrel]

Only issue with Gemini is they view and train on your conversations. ChatGPT has a couple deep research requests on the lightweight model for the free version if you run out on perplexity

[sturdy willow]

has ChatGPT gotten better at not hallucinating as much? I used to use it and then I quit it cold turkey a solid 6 months back.

[dry kestrel]

going to do some tests myself, just seems like a weird issue