#💬│general
1 messages · Page 238 of 1
I already gave good example where Google platform + Gemini 2.5 Pro got a question wrong, despite all identical conditions.. Perplexity + gemini 2.5 Pro however got it right (with or without web enabled)
Yup. Same prompt. There seems to be an issue with my account. Claude 4 Sonnet Thinking now never displays thinking in any interface, and both 4 Sonnet and Thinking now have the wrong knowledge cutoff date. Neither of those things were true before, and the base 4 Sonnet on the Anthropic site doesn't have the same issue.
They submitted a bug report already which staff is pretty active about
Can you explain why chain of thought is completely gone now when I use 4 Sonnet Thinking? Web search on, off, in a space, not in a space, etc. New contexts and old. Never displays CoT any more.
Have we considered maybe the AI is just sick of him?
The chain of thought should show when you have web toggled off, mind sharing a screen recording?
AGI is already here
My https://claude.ai usage is maxed out but heres claude on T3chat
With web access off
Imagine in the personalization the model remembers "user hates me for seemingly no reason"
so he's wrong in basically all ways lol
if even Claude platform behaves same way as Perplexity with identical prompt etc
oh i see
yea that's the issue with not using identical settings to reproduce bugs
again tho, it may not be surprising...Perplexity might just have different way to interpret prompt and it bugs out etc
regarding chain of thought/displaying thinking, is that consistent?
that it doesn't produce it. like do 5 separate windowed prompts (different ones), and all 5 times is blank?
Agent mode
Hey Harvey, mind helping me out here?
Claude.ai has their own way of web search and system prompt which could be the reason to why its working
What happend eddy?
is a bug without repro steps really a bug 
Most sites like perplexity has in their system prompt that tells the model it must search the internet for specific things
Someone here claims Sonnet 4 is another model/3.5, told them they were wrong and why, which resulted in me getting blocked 😭
Oh wait they left the server I think
Really for saying correct answer?
Well, I'm pretty dang certain that the user was just experiencing model hallucinations with Claude 4
I did experience this last week with Claude while building a project
Perplexity should really consider telling the model its name in the system prompt
But it happens
when sonnet 4 first released, I was trying it out on the API when this happened
did a bit of testing and it turned out it was just claude 4 hallucinating
Hey Joe, haven't seen you active recently
At least Claude’s hallucinations are ok, we don’t talk about grok’s
People needs sugar coated answers eddy, when we say that they like them more
Oh well, it is what it is.
Noice joe came, how is your gym day joe?
so much free stuff for us indians this week lol
there's free perplexity pro for 12 months with Airtel (india) for all pre-paid, postpaid and wifi users and free google one gemini 2 TB pro for 1 year too with student discount.
Better not to prove our knowledge to a random Unicorn
I need super grok too
Where to check the perplexity offer?
Honestly, if you have a perplexity subscription already, there is no point in getting SuperGrok
You get more uses of Grok 4 (300 a day across all models) compared to 30 uses every 2 hours in SuperGrok
airtel thanks app
But grok can search tweets but this can't
Is the offer live at present?
or is this upcoming offer
My only issue with this is that it isn't as hard core batshit crazy as Supergrok lol, so it needs some prompt modification imo.
nah just saw someone on twitter redeeming it like 15 mins ago
definitely live i think
That is true, supergrok is able to search tweets much better than perplexity due to the native integration. Curious to see what you use it for
Haha, gotta love Grok 4 claiming to be that Austrian painter
i think grok introduced real-time search api with their native api
I generally use it for news research and trending topic for my youtube videos
anyway pplx is also a search engine too so you must be able to get real time data easily
would be simple to refute by simply actually putting his prompt in the actual Claude AI site
Would be cool if they could somehow get their Grok 4 api service to somehow communicate with their tweets, not that X's data is top of the line lol
@cobalt widget Yes 100% correct! It's LIVE
nicee
I think so too
Ah, I'd say check out how Perplexity does and then if it doesn't work out, get SuperGrok. perplexity is also able to search for tweets (albiet not as good as grok)
That would be awesome
oh right, so Grok 4 is limited at 300 a day on Perplexity? Are there any other known limits for other models per day?
Maybe #1111765634267742408 or #1358914210339360798 ?
you get 300+ searches across all models on Perplexity Pro a day
Much higher then SuperGrok
I ran out of usage on my Claude Sub, gotta wait a bit
I think pplx should add mcp support
They already have it on the Mac app (introduced this week)
hopefully it comes to PC as well
this goes down by 1 with each pro search or image gen made:
but it also seems to increase by 1 every 1-2 minutes
In reality, once you hit 300 searches, you start getting rate limited (I tried this already)
maximum seems to be 600
Its more than enough for 99% of users
ah right...I feel I've tried it too, but maybe wasn't hitting 300
pretty sure most of the last few days, I've done >1000
but didn't officially count
wait, how do u mean "rate limited"?
how would you know if that was happening?
You would get a blank message if you do hit the rate limit
How are we doing tonight
Ur wild for reaching that 300 daily limit lol
ah okay..definitely haven't experinced that
and it's 300 separate threads eh
follow ups don't count right?
I just want GPT 5 😢
was plexing a lot
Agent mode
Ive seen that a few times.
lol
You can check your usage here https://www.perplexity.ai/rest/user/settings
Maybe allowing pplx to search/reference all threads is a bad idea.
I've definitely gone over 300 prompts
per day tho
just dividing my total prompt count by number of days I've had this account
already far exceeds 300
Assuming it takes 1 minute to recieve and then send a prompt, thats 5 hours of continuous usage
several tabs open at once 😛
and in reality, the time in between prompts is much higher
"On the Settings page, you will see your current usage for the day. Perplexity Pro offers a high number of Pro Searches (currently over 600 per day"
I think I understand the price of "Max", on learning about some of the features in Comet. If Perplexity, make Comet available to Pro and Free users, wont these versions become shadows of the real thing and almost flavorless compared?
600 per day would make more sense from my experience
Interesting, I think thats available to Pro users rn?
Perplexity need to update their official documentation more imo - for example, 25Mb file attachment is still stated everywhere
I believe the 300+ per day has been 600+ per day for at least a few weeks
i have 4 invite left dm me if you need one!!
Its out?
I wish I had pro to know lol
This is different
Deep research and agent mode
Oh? I might have misinterpretated?
I know for sure that connectors support DR already, agent looks interesting though
Agent Mode announcement 👀👀👀
︀︀
︀︀"Agent mode can do even more complex tasks, like spreadsheets, research, and reservations, all on its own."
Quoting TestingCatalog News 🗞 (@testingcatalog)
︀
BREAKING 🚨: OpenAI is planning to announce "Agent Mode"! Agent Mode will likely be a mix of Operator and Deep Research, which can use the browser and connectors at once.
︀︀
︀︀"Find, analyze, and synthesize your Drive files to create comprehensive reports"
︀︀
︀︀Deep Operator 👀
like an improved operator and more
Connectors are neat though tried them on my team account
oh nah
Def aligns with the tweet that they posted, very interesting
lemme guess
curious if this will be pro only
"Releasing to Pro first, coming to Plus later" or insanely low limits for plus
meaning comet still wins lol
If it comes to teams i will be happy
smh just give us gpt-5 sam
🥵
Fr
how did they even get access to this
Pretty sure editing backend code using inspect
so its still not cemented what tomorrow announcement is
in any case its most likely not a browser lol
they cant make something so insanely fast
I think it will be
Doubt it
At best just operator V2 which still operates on their servers
What else would the cursor be for?
Looks like the cursor in ChatGPT Operator recordings
they're buying cursor? /s
lol no the Cursor in the video
"Plus, perplexity imposes many arbitrary and nontransparent limits. For example, the context window is much smaller when you use the model through perplexity than when you use the ChatGPT subscription directly, and the scary thing is that you don’t even know when and how they change it behind the scenes."
trying to interpret what that means tbh
This article looks like its been written by ChatGPT
true
Hi everyone, I would like an opinion/help regarding which one would fit better for my daily use.
I have access to GPT-4.1, Gemini 2.5 Pro, and Claude 4 Sonnet and would appreciate some guidance from a non-technical perspective.
My needs are straightforward: everyday assistance with things like emails and planning, and help with my academic studies, such as research and writing support.
ive said this so many times, last couple years
hey
It's a possibility that they were working on a browser like comet the entire time and just not releasing it i guess
yeah, I meant more generally speaking, dev has gone light speed!!! always someone or some team releasing something
maybe im just getting slower 🤣
looking at my /rest/user/settings
per month...
my usage is 2128 (query_count)
allowance is 9999 (pplx_alpha_limit) for everything apart from gpt4
gpt4 allowance 598 (gpt4_limit)
is that right?
which browser are you using for that?
edge
tbh, that wasn't how I interpreted it. I actually think gpt4_limit is basically allowance for all models
it goes down regardless of which model is used iirc
ah ha... i see that now
The most capable all around model would be Gemini 2.5 Pro, if you want a quick response use GPT-4.1, and for coding Claude 4 Sonnet
follow up question mate,
is it best to stay on "Best" option or manually select depending on intent?
is this per minute, per 5 minutes?
I notice it goes back up after I stop querying.
I personally prefer to use Claude 4 sonnet, if you just want a quick answer I'd stick with best or GPT-4.1
not too sure, from what I see it just every couple minutes, hence the 300**+** searches
probably 90% of my queries are like a google query!
Would recommend using 4.1 then since its a good mix of intelligence and speed
I'm going to try Sonnet 4 in vs code, hey thanks for the recommendations, helpful info.
Np
Hello
Hello
Perplexity doesnt integrate with VS Code, via cline, continue or similar extensions? Is that right?
in the context of, querying things like.. look at my code and do something.
No, its not exactly meant for coding
I basically have been using Gemini 2.5 Pro as default, as I feel it's overall the best. However, as pointed out, it can be a bit slower.
I actually think it's the best for scientific/literature/medical-related stuff, especially when attaching related documents
gotch
again, I believe it's 600+ "per day", but would be nice to get official stance from dev team etc
intuitively, the fact it never goes above 600 also supports that
if it was 300+, then that number should cap at 300
When coffee is over, cactus takes the role
that's surprising - i was told Edge displayed it in RAW format
I love eating cactus
i have a json extension
thanks that explains it
Firefox has that built in, and displays it like that by default
Staff confirmed it was 300+ a day
when was this please>?
I personally feel 300+ is more than enough for a user
Nope
you've heard update since then?
It was never changed, they always claimed that you get at least 300 searches a day
your link points to 2 weeks ago
i see, but how do you know it hasn't changed since 2 weeks ago
You wanted official confirmation?
Because I started using this service before 2 weeks ago.....
yes, but more regarding whether that had changed. Much like the 25Mb to 50Mb attached only changed in last few weeks iirc
yes but if something changed 7 days ago, that wouldn't matter?
I know for sure that the query limits were not changed at least for this past year
What are you trying to prove here
that makes sense, but the 25Mb limit also hadnt' changed for a while right?
whether something changed since last 2 weeks ago regarding prompt limits per day?
I'm talking about the query limits, not the file upload limits
and I'm telling you it wasn't changed
yes but just like the file limits hadn't change for a year (and then only recently did), isn't it possible the query limit could also (recently) change?
This was in 2023
It was always 300+ searches a day
actually would be reasonable to have increased it too the longer it stays the same
yes I not disagreeing that it's been like that for a long time
perplexity refs is crazyy
just whether recent changes have occurred, just like the file limit
Isn't that exactly what you were questioning?
I don't know when the file limits have been changed
big strength of perplexity
not at all
fr!!!
i was questioning if something has changed in last 2 weeks regarding query daily limits
You can read the changelog here https://www.perplexity.ai/changelog
I don't believe that changelog documents everything
Hi - I have Perplexity Pro, and used the link from the settings, but don't seem to have access to #No Access.. any guidance ? Thanks
@serene sand Please wait until staff gives you the Pro role.
Pro role?
sry but thats all I know so far, you can always read though #📢│announcements to see recent changes
in fact, I'm pretty certain the file upload changed twice in the last 2 weeks. it was 40Mb for at least a while...then became 50Mb last week or so
If you've subscribed to Pro, you can get access to some specific channels
wen for me
I dont have access
@serene sand
me too
Give staff a second and theyll give you your role
Check out #👑│pro-lounge and #1111765634267742408
damn, thank you
appreciated Ed
this is demonstrably true for me, as I uploaded 31Mb file about 10 days ago, but failed to upload a 45Mb file. However, since last few days to 1 week, I manged to upload that 45Mb file now. Point being that things can change quickly and without official documentation or announcement
Np
You can leave some feedback for the team regarding this in #👉│feedback-general
I was actually pleasantly surprised it allowed 31Mb file too, since the official documentation said 25Mb. And now even more pleasantly surprised it went from 40Mb limit to 50Mb quite quickly
hopefully these recent changes means their "limits" for other things will also increase 🙂
But I have the tendency to touch the cacti
, also it looks delicous
persumably due to increased resources on their end etc
I liked when it had the trending news articles above the prompt input bar. Did they remove that entirely or am I just not seeing it anymore?
I think they removed it? not seeing it on my end at least
I guess you'll have to use https://www.perplexity.ai/discover now
wasn't the point though. I am saying, just like limits like File Upload has suddenly changed within literally 1-2 weeks, perhaps other limits could also have changed, like Query limit
anyway let's test out and try exceed 600 queries 😛
I thought you wanted more detail in the changelog?
I guess 🤷♂️. I just liked the convenience and I felt like it spiced up the Home Screen and made it more interactive
Hi I'm wondering about Pro license. Pro can use the model from other brands. if other brands update their model, will Perplexity update those models to the system?
ah right, yea already fed that back a while ago though
same, could try posting in #1358914210339360798 or #👉│feedback-general ig
probably better just to update official documentation imo
If say OpenAI releases a new model, perplexity would have to manually add that model. Most recent case is Grok 3 getting updated to the newer Grok 4 which was done within a day of release
the 25Mb limit has been like that for at least 12-18 months based on reddit questions from people asking how to uploaded larger files
so it's nice to see that increase after so long
Sounds great!
sorry I forgot to point out (had to afk when u first wrote that) - the 9999 only came about 12 hours ago, and I'm not aware if anyone knows what that means. It doesn't seem to change regardless of what you do
do they plan to have other models, except the exist models?
I guess I need to buy one for my desk
They'll usually add new models when they come out. Provided its a a reasonable price
coolio, I dont get hit by any limit as far as i know 👍🏼
Currently the selection includes most of the SOTA models
thats one of the best things about pplx
You get 300+ queries a day and after that, you might start noticing blank responses
Its incredibly hard to hit that though (and time consuming)
is it purely queries as in... 1 request 1 response.
I would have thought the token count would be more relevant
Perplexity’s gotta get better on the image creation train ngl
https://youtu.be/AcRsHrDu72k?si=yBO2OKuEiwUwZ-3t
Neo doesn’t approve
Mr. Anderson sure needs some unplugging, albeit in his own way. #iconicscenes #thematrix
the Grok 4 was impressive how quickly it came to Perplex
Yep, pure queries
doesn't seem that hard to hit 300 limit if wanted to - just open a few hundred tabs with perplexity
then just copy/paste a prompt in each
would take a little while but if you had a few spare hours lol
possibly could be done in <1 hour actually
Please don't do that
In that case, I've been doing something which doesnt matter...
When I have a long running, back and forth query, I summarize the conversation thus far, and create a new thread in order to reduce increasing token count.
(saving the trees 🤣 )
Planning my site ima make with gemini, what do yall think?
just saying that it can be done lol. However, it appears like one recover a query every 1-3 minutes or so, so have to take that into account in terms of estimated time to complete 300 queries
and in fact, if u take 12 hours to do it, you probably have to do it at least 450 total queries (assuming the limit is 300)
Looking great, you should def check out something like Figma, makes life easier
I wonder how good it would be, if you uploaded that image and prompted... make this.
actually if it recovers 1 prompt per 2 minutes....that's about 30 recovered prompts per hour
Yeah well thats not normal usage and you could very well get your account flagged
indeed, would literally be spam
maybe if Max user, would be fine 😛
No, stressing the service degrades it for everyone else.
No point in spamming queries
yea but i mean for being flagged...just a joke anyway
joke being that since paying "enough/more", spam is more tolerable lol
which of course, it is not
in those cases I would think the account is used by many, used for commerce and on wrong plan, or the queries are very wasteful and should be optimized
I wasted 60 credits for this

tokens well spent... that one way to be more sharp
Q - I have approx 200 electronic forms, and 300 documents spanning over 7 categories. Many have a relationship with each other...all these forms/docs relate to an over arching Manual, which itself is govermed by an ISO standard... any tips of how to best structure and use Perplexity to Manage.
Any input/advice ..
Someone on Reddit said that they were generating video in pro account, anyone seen that? Mine says it’s making a video but after all the steps just provides an image.
Afaik video gen isn't available yet
video is max users only
pro users dont have video ;-;
It's extremely expensive
yes
in time prices come down
in my country they gave perplexity pro for free for 12 months today
check out this new prompting tool https://promptweaver.me/, literally it writes a single shot prompts to get the whole UI done, whether your'e usind bolt or lovable, ide like cursor/windsurf it got your back
"Perplexity can now create videos, but with important limitations: this capability is currently exclusive to X (formerly Twitter) and not available directly on the perplexity.ai website or in the main app interface"
when on a project, usually my first prompt to pplx is: whats the best prompt to... 🤣
is that actually written somewhere? I couldn't find that even Max allows video gen
u made this with perplexity?
was this made with perplexity?
or google voe 3
veo
(sora... to add another flavour)
where's that ?
it was an alert in pplx web
wow...u mean like when an updated version happens sorta alert
Not the place to advertise you can post in #😎│cool-projects @serene sand
they should advertise that more widely..I'm sure it would attract more people to Max
Veo3
That guy is spreading misinformation https://www.reddit.com/r/ChatGPT/s/TUzwBQ3Q49 It also will only be on max when it releases
looks like video generation has literally just arrived I guess? might explain why the image gen was down about 24 hours ago
they needed to do something at the backend perhaps to implement video generation
Video gen has not yet been released yet
would you use video? I think I would rarely use other than funny social posts!
I missed the AMA, anything interesting come up?
purely for fun only yep
even image gen is purely for fun
for me anyway
we just need to convince Elon Musk to use the money he putting into the 1 million-GPU cluster into increasing the hardware/servers of Perplexity 😛
is Elon connected with pplx?
This doesn't make sense
Lol
Pplx just get's it off api
If you're good at prompting, you don't need these fancy websites for building your prompts.
Ah, I guess they are doing a slow rollout then because Aravind replied to me citing this week
No
lol yea...he probably doesn't even know Perplexity exists
Probably too high to care
the only thing I dont like about AI, any LLM, is the tight restriction around religious questions, albeit I understand why they are probably necessary.
altho on business level he probably does i guess
@dry kestrel anything else intersting come from the AMA?
Hi all
what about this then?: https://x.com/AskPerplexity/status/1935759056209428531
is that actually official Perplexity?
yea so u can generate videos via that apparently
but i think u can only share them
Yep
Hi Any idea with airtel perplexity pro subscription, do I get any free api credits?
what do they use to generate video?
"Although the company hasn’t confirmed which large language model powers the feature, based on its functionality, it appears to be on par with leading video models like Google’s Veo 3. However, this is speculative, as no official model details were shared."
i guess i created a blunder, i already had a pro subscription, but i tried to claim the pro through airtel app, it showed me a invalid code, and every next instance it shows me i have already claimed the code. :} am i cooked?
Close and open the airtel app and click the recover option. As soon as your phone redirects, try to get the URL link and open in another browser.
If you feel grateful today, you can gift that to me 😄
😂
the first time it showed me promo code applied, i couldnt copy the link the next time it was this code used. argghhh
Theres option to recover account - give that a try
maybe you activated it in one of your google accounts.
umm nope, the thing is like i have an existing subscription is it due to that i am getting a failed response?
yep probably
does any one has 1 invite for comet??
Aravind adressed security/privacy concerns (tbh the data they were collecing were to be expected)
iam a pro user of perplexity i have also joined the waitlist i wasnt got one?
You get $5 worth of APi credits every month
You just have to wait
If everything is local, how does it work w cloud hm
do you have any invite left
No
Even then, transmission to Perplexity's servers is tightly scoped and purpose-limited. All these queries can be deleted on your history or done incognito to ensure it stays local to you and only you.
they still send data to their servers
incognitio is yet another good idea by Perplexity. I'm not aware other AI platforms have this
chatgpt, grok, most of the platforms out there
ah right
I don't think Gemini does
which is funny, cos pretty sure Google Chrome browser popularised the term "incognito mode"
Hello Team
Would it be pssible to get an invite for Comet for my pro account? I just had two day ago a right arm surgery and can't use my arm and it would be amazig if I could interract with the browser with my voice
do the others incognito mode work similar/same as Perplexity?
thing with perplexity is that u can still access the thread even in incognito mode after 24 hours - might actually be nice to have an option to adjust that - for example, only 1 hour, or even instantly removed once close the page
Stop repeatedly posting your referral code here
sry
He seriously went first to comet general and sent there and came here too
Can i get comet invite?
if any one want 12 months free perplexity pro DM
Hi @torpid mortar can i get a comet invite. TIA!
Sry I'll dm u the link it would be easier for u
Good reminder!
Appreciate it 🙏
just saying... I am open to 5 year full access max trials 🤣
That's not a comet Invite buddy, it's a referral link!
My bad
Does anyone have a comet invite?
You can try your luck in #☄️│comet-general
@signal hamlet also here
hey guys
I've been using ChatGPT then switched to Perplexity. I prefer Perplexity but I want more conversational or on-going dialogue like ChatGPT instead of Perplexity treating every query as a standalone Google Search. Is this possible or do I just need to accept it as it is?
Gone
ask it !
Hi everyone. I have a Perplexity Pro, and for the first time I get an error message when trying to use the Research mode, asking me to subscribe despite showing that I'm already subscribed.
Is there a quota of research uses that I exceeded? I can't find any quota info on the help section.
Or is it a glitch?
Thanks I did cancel and renewed my membership. @viscid karma
i think i had a similiar bug, try to clear cache/cookies and login newly
You’re welcome
#1393681130842226698 hello, i am now on the Comet waitlist, i am still waiting for 4 days , when website will provide specific notification, are there any news ?
Been on it for 47 days
doesn't provide the answer
Thank you, but no luck. I tried with a different browser as well and it's still the same
https://www.perplexity.ai/rest/user/settings
check here
dunno if you hit limit
I see, that's insightful. Maybe I did hit the limit, but the error messages don't say anything about it, and there's no mention of a limit in the FAQ either..
ye it say's unlimited
i never had any issues even with high ussage
never hit limit
hi
it always replenishes as well
alright, thanks for all the info! I'll contact support about it
good luck
Has anyone compared Dia and Comet yet?
not familiar with that feature, but its all in the prompt right!? tell it what you wan tfrom the start
heyy
Pplx gonna get a huge customerbase after this
It's only gonna get more people here
Hopefully they fixed the promocode abuse issue tho
maybe you are already denied)
What's the abuse issue?
Promo codes for samsung was leaked
ah shit why is like that or users are so many?
are they accepting manually each users)
No but in batches
what are batches what did u mean bro
WHAT ABUSE?
good job we dont have an energy problem on the planet 🤣
yeah
just made an tweet tagging comet! I have been on waitlist since APRIL! and still didnt get any invite!
Chill, they are rolling out acess slowly, you can always check on twitter for people posting invites
anybody in here have any invites?
is this basically a massive infrastructure problem... not enough hardware for all users
you talkin about usin comet? think they wana test it properly no bugs and stuff?
basically that
yeah, 500k on a waiting list, and imagining the workload of comet.
xd
Hi I had bought perplexity pro from AIRTEL APP offer ...
First I bought from one number ...
Then I had to buy again from another airtel number which I had ...
I logged in and clicked to get offer ...
Thing the subscription added to same account ...
I wanted to buy it for different email ...
Can you extend the offer by one more year or give me the subscription for another account
Anyone help here
I got fkeddd
is it formal data i meant where did u get 500k on a waiting list
Actually very few use AI for actually heavy work. So most ppl who have pro they still use AI for basic things and all. He wants to increase the user base maybe there's a trick to it. Or he's just okay with burning his money.
lmao why do you want 2 accounts?
yea hopefully i get an invite
dont worry about it, when it comes it comes
I just checked, apparently Airtel has 330 million customer base, and all of them got Perplexity Pro 1 year free.
Imagine the number of folks redeeming it 💀
yea most of them are gona never use it tho - even if they use it, it ll be as the same normal usage from the free plan
Yeah gonna get so chaotic
They'll need to get the rewards program claim for pro
But you're right
Most gonna not use much
yep! its just another marketing shit? idk what to call it
It's increasing userbase
but I M GONNA USE IT like for everythin 🙂
Hopefully just not to sellout 🫣
yea hopefully not sellout
probably this
I believe in kesku
co-founder? or something
wait when was perplexity launched?
A member of staff
ah well he cant stop a sellout
I think they're about to strike a deal with Apple and Apple will just inherit this cost 💀
XD oh
Apple's branding doesn't suit pplx tbh
WHAT Apple?
yep
I think it is very apple like, from their website design to their assistant
Apple had bought Shazam!??
Well not polished enough
Lmao yes
Yea, an eternity ago
Old stuff
perplixity is basically layered on other ai models right
I'm an Apple user and pplx more polished right now 🤣
but also adds to other models right
wait since now I have the PRO subscription can I use Claude via API?
yea me too
You can use sonnet and sonnet thinking
Opus needs max sub
WOAHH!! THIS IS ALL I NEEDED
Just open pplx and choose the model from the drop down
Pplx limits them to 32k context tokens per thread
Keep that in mind
the entire citation feature was pplx creation wasnt it?
oh but I cant use it via API?
No
Pplx provides you model through api, you don't get that api ofc
Waiting peacefully for a huge crowd from Airtel users
👀
i m one of em
Pplx model
oh just the pplx model
Anyone confirm did they reduce daily message limit somehow due to now such huge pro subs to everyone?
nah ain happenin
My stupid ass got paid pro last week can't even get refund now I'm gonna get one more sub from my mom sim 🌚
perplexity has added grok 4 into their app - now i can access it for just $20 there? and normally i would pay $300 for it on grok itself?
lol
they made it free for 330 million ppl now, insane
well it increases userbase, and also by default, you are "opt in" to share your queries/data to train model
Getting Claude API is getting access to VIP club of Billionaire which is not easy lol Anthropic have limited lot API access, Perplexity is just wrapper with VERY GOOD wrapping on existing API models
Harvey 🥲
hi
Where can I submit request regarding a language/grammar pplx is using? I want pplx to be more neutral, supportive and comply with WHO and IASP recommendations in the field of mental health.
alr
??
lowkey sup
woah
crazy
when is perplexity app gonna come for linux
ah wait its $30 not $300
😔
woo hoo
use the webapp?
yea
lmao same thing?
xd
any specific usecases, which sets it apart from other models?
wait how do i intergrate it here
uhm......
Grok is 30usd
Something will definitely go wrong I'm sitting on it and watching what and how, either reduced messages, or quality or model degradation, they need HUGEEEEEEEEEE VC BACKING which they have to hoard so many free loaders
idk it depends on what research you are doing
Grok heavy is 300 usd
yes exactly this thing
I don't think Gemini even has this option to turn off:
anyway, the point is that by increasing userbase, they are also able to use more data
yea yea mb
wait How do i enable it
i don't know, i don't use windows
sucks to use windows
lmao frfr
So 2nd code will be activated when the 1st one expires?
well i do have linux installed too (kali one)
Yo
exactly, they'll bleed millions
but i m too lazy to transfer my files
No both will get expired on same date
who uses kali
lmao no!
Lol, can we change it by taking to support team??
idk I just wanted to learn about hackin and shi
aha
woah
still
what is the difference between the Free version and Pro for an normal user?
liteRally nothin
you also do realise that by "participating with these large bases (mobile companies etc), Perplexity is also likely to gain millions?
i'm sure Perplexity will get a cut of their profits
level of coolness
imgs can be generated in pplx? ???
exactly...
just ask it to "generate image of.."
just the inception of being a pro user
I was saying that could we request team to activate the 2nd one after the 1st one expires?
Well study what has happened on Cursor and all coding IDE with free trails and what Chaos is happening even if a company worth trillion dollar can't afford GPUs unlimited with good models, Google is doing because they are already safe for being data hoarding company with own TPUs
yep exactly
weew
yea but how to change the model
like this
i dont even know what model it used
hm
Yeah ask....can u dm me how can I contact them directly
hi!
if you didn't select any it chooses based on query but in my expirience at default it always uses gpt image 1
in the settings ?
yes
Default often uses GPT Image 1 yep
gpt image 1 is goated
got it thanks
fair one, i wonder how different the others are, or better for what
being a student is so COOL istg, I got perplexity + google gemini + cursor + framer (all with 1 year subscription 🫴 🤩 ).
DALL-E 3 is the best?
test it
how do i go back now....
GPT Image 1 by far the best for real things
for artistic etc, maybe others
claude 4.0 is the best accordin to me
for images?
yeah when is it gettin updated?
nah
Dm me bruh
for normal deep research
Yeah overall claude and grok 4
alr!
idk Ive had mine on "Best" for years
unlimited img generation right?
yes and no
GPT Image 1 (or GPT4o) is used up to say 50-100 times a day. After that, it will use some other method, but it's not bad still
yeah but quality sucks even on gemini pro and pplx pro until they update the API
to new gen chatgpt one
oh okkk
and img gen in pplx takes a bit more time
btw try this prompt
Write like a smart human, not a machine.Be clear, confident, and conversational—like you’re talking to another thoughtful person. Avoid robotic phrases like “in today’s fast-paced world,” “leveraging synergies,” or “furthermore.”Drop the fluff. No filler. No AI tone. Skip dashes (—), unnecessary quotation marks, and buzzwords like “cutting-edge,” “robust,” or “seamless experience.”Use natural transitions like “here’s the thing,” “let’s break it down,” or “what this really means is…”Vary sentence length and rhythm like a real person would. Prioritize clarity, personality, and usefulness. Above all, think critically. Don’t just say what sounds good—back it up. Give value-driven, reasoned answers, not vague claims or surface-level ideas.Every response should feel human, intentional, and built on solid thinking.
spammer
tf?
i just joined this server mate!
sad
lmao
sucks to be that
lol
Welcome all new users
lmao
quality was actually really good....just they have had to dumb it down a bit for some reason...probably traffic flow.
I have a few images which are 2.5-3Mb size - they are another level above the 1.5-2Mb size
I aint getting any of that!
Hope you have a pleasant stay
damn the theme
crazy
I think when traffic is too much (also maybe based to how many times you have used), they reduce the quality to only 1.5-2Mb PNG size. If you get lucky, you get 2.5-3Mb size which is another level
but rainmeter consumes a lot of battery right?
woah i like this one
mines a live wallpaper of cool imgs
windows-EW
animated btw
xd nice
so good!
i m changin my lock screen to this thanks lol
what's wrong wit windows?
i don't gate keep
it's bad
shit what which ai are you usin for this? api ?
yea it is
wdym ai 😭😭
though it ll give you short answers
XD ALR ALR
it'll
You can try to contact them by going to your profile and click on "contact"! Wait time will be 5-10 minutes max.
wait
Nah mate it's not its just more freedom than mac! im a mac user btw
if i type a question
it'll open my browser
and do the thing
google search
alr+space = pplx how do i do that in windows
hmm
some toy u had
by ms
idk
some shit
i can prolly make that myself sometime
neovim btw
yup
# Define the path to the config file
config_file=$HOME/.config/hypr/UserConfigs/01-UserDefaults.conf
# Check if the config file exists
if [[ ! -f "$config_file" ]]; then
echo "Error: Configuration file not found!"
exit 1
fi
# Process the config file in memory, removing the $ and fixing spaces
config_content=$(sed 's/\$//g' "$config_file" | sed 's/ = /=/')
# Source the modified content directly from the variable
eval "$config_content"
# Check if $term is set correctly
if [[ -z "$Search_Engine" ]]; then
echo "Error: \$Search_Engine is not set in the configuration file!"
exit 1
fi
# Rofi theme and message
rofi_theme="$HOME/.config/rofi/config-search.rasi"
msg='‼️ **note** ‼️ search via default web browser'
# Kill Rofi if already running before execution
if pgrep -x "rofi" >/dev/null; then
pkill rofi
fi
# Open Rofi and pass the selected query to xdg-open for Google search
echo "" | rofi -dmenu -config "$rofi_theme" -mesg "$msg" | xargs -I{} xdg-open $Search_Engine```
i got no idea tf this is
try something like this for alt + space
ah this is the code for the search
yes
i no coder tbh
itoskayy
do you know stuff about runnin models locally
ablated version of them
ohh
atificial engineer?
do you have CUDA?
deepseek cool
it's nice
try the pplx version of deepseek
uncensored
but its not ablated and shit when i ask stuff it still says i m not allowed to say that
xd
but i downlaoded the ablated one
0 what
i know this dolphin mistral
xd
llama has an unrestricted version
4060
ou cool
huggin face link?
how much vram?
i got LM studio
thing is, the things that are restricted are things we never search for anyway, fun for 10 mins then downloading a proper one
0 learning curve
8gb right?
uhm cool
goddamn
32B
bruh
lmao wt
Open WebUI is an extensible, self-hosted interface for AI that adapts to your workflow, all while operating entirely offline; Supported LLM runners include Ollama and OpenAI-compatible APIs.
that's too much
oh lol idk
even ts is too much
alr i ll check it out thanks
its finding a balance on your host, biggest for acceptable response time
ddr5?
yea
exactly
okay
@winged dome
try 8B models
8B-12B is ur range
4B 6B will work best
8B will be fine
12B will be slow asf
but 8B - 12B ur machine can handle easily
deepseek wouldnt be my choice for general stuff
Is there unlimited pro searches for pro users?
ig
300 per day
"Essentially unlimited "
i just wanted to ask stuff which normally other models wont tell
this ?
ah dang it
What do u mean exactly about it?
It should be plenty for people per day
8x7b
i mean 300 is fine i thin
exactly
It also refreshes fast
when you get setup mate, youll be downloading models every day!!!
use 4bit quants
What are the features we do get in pro subscription?
ah wts that
you'll essentially need a rlly good system prompt
Quantization
i ll ask gpt for that?
alr alr
alr
okay
wsl maybe a better option
hello, how do I join the pro channel? when I click on the link on my account settings it redirects me to this server instead
<@&1098365634498285709> Wait for staff to give you the role
oh, so do I have to request for a role or something?
ah nvm got the role
theres a bug where it sometimes doesnt give you the role automaticlay
thanks a lot!
got it!
check out #👑│pro-lounge
Hi all, is there any specific role for pro access people? If yes, how to get it?
do you have a pro sub?
Yes
Yes
@wraith crow
it's shit
can someone give me pro role?
Thank you
@wraith crow Give staff a sec
how do you know, your on a mac!
thanks
Hey, MLX works pretty good on mac >:(
😃
hi
i don't use mac
i hear good things
?
just messing with you, you think everything is shit lol !
why?

yo
Hi
Heyy
@shut coral sorry for the ping but joining from the link in my menu didnt work. I didn't get the Pro role even after having subscribed to it
Whenever I search for a prompt and select a particular model, the ai reverts back to the "best" model , is anyone else facing this? Please help
U using grok?
known issue for mobile app
already reported
No for any model possible
What kitten said then
Is it new or always been this way?
Daam
What does the Perplexity Plan cost?
What do you guys use perplexity pro for?
i don't know when i saw it the first time but it is for 2 weeks if i remember and saw last time
I claimed it just now
Can Perplexity Generate Videos?
How is it free if you are paying 17'000 🤨
It's free bro
It says worth 17000
