#💬│general
1 messages · Page 105 of 1
Purchased again?
Nah I just had to rejoin through my settings on the website
Yesss I think
Oh
I bought it for a year
DM me with an example to test
I have it monthly, thanks to dropshipping🙏
Dropshipping
Yup
Uh what and how and
I use autods (sorry mispelling), also dropshipping is just reselling but with having no actual physical storage
Rejoin using the link in settings
Yeah but how do you get perplex
Ahhh lol
No just to have money to pay for the subscription
It worked lol
I tried rejoining with the link before, and it didn't work, but this time it did
Uh how to imagegen on mobile
They should have a 'perplexed' role for people who report bugs 🙂
Hey guys is it possible to make perplexity model offline??
They did release r1 1776 for offline use
But it requires a lot of memory
Yeah
How can I set it up??
And how much memory??
The full model is like 480 GB of VRAM iirc
But someone might release a variant that's a smaller version
Oof, it's actually about 1.4-1.5 TB of VRAM... but quantized about 480 GB. A 70B distilled version of it is here, which you can run in under 50 GB of VRAM: https://huggingface.co/unsloth/r1-1776-distill-llama-70b-GGUF
PLΞX
UP
ΥOUR PHOΝΞ 
Thanks!!
bro why it change model every message?
Jtjfjfhffj beautiful
Are there others?
Beauty
provided by courtesy of
Oweee
Community mod
How can I become one?
do they pay salary for this?
I don't think so it's community
I would only do it for the vibes
Perplex seems like a cute community
cute community, but only users speak here, developers never respond to queries here or in email, in email support only the chatbot guy SAM replies, he replies confidently wrong information if he dont know answer
lol
@signal hamlet hello
I just wanna make sure I’m not doing anything wrong brother
By talking about andi
It's in the rewrite feature
Yeah I’m pretty sure Sam is an ai
Cuteness is all that matters
try again, only manage your risk this time 👍
I was going for $10k
Got greedy
Ohno don't say that in discord it's a warning if you get reported
What how why
does anyone know why my perplexity is always searching random sources (most commonly "Apple CEO 2023" and "famous birthdays"? I had sent a prompt very unrelated to what it searched (for a fish project). It usually still gives a good answer but why is it listing these as its sources
it happens, try only betting 10% of your account next time and you'll get there
It will take longer tho
How to get invite to this program? - https://www.perplexity.ai/business-fellowship. Is there any chance for the standby list? I work with quite big corporations and I would like to introduce to them enterprise use cases. Please dm me someone from Perplexity team.
Hi - I’m a a pro user on iOS- how do I talk to purely the models rather than just a search engine?
Focus - Writing
Thank you! Never noticed this
Claude is radically different different on Perplexity than on anthropic
Damn y’all really want me to get a VPN
now the fun bit would be when they start sponsoring a reccomendation for <X> VPN provider...
Wow interesting!!! Thanks for your expertise!!!
Are there still limits? Could I attach my documents and talk to claude these days?
Anthropic sets horrendous limits so
deep dive ➤ #💬│general message
check Tutorial I and docs
What is high ?
there is no o3. only o3 mini. And yes it is not smarter than o1 and defnitly not smarter than o1 pro. Even R1 is smarter than o3 mini (low , med and high) . Im working daily with Ai mulitple hours. Only in coding Backend O3 mini is better than R1 and O1. (not o1 pro) But for everythink else o3 mini isnt as good as R1 and O1. R1 goes much more in depth and gives better results in general. O1 does the same but even a tiny bit better. O3 mini is not my wy to go i dont like the anwsers and Think process.
Curious what your read on gpt 4.5 is
4.5 is the best model for writing text. I love it. But for everything else i couldnt really find a usecase.
There is an o3
Api doesn’t have access to it
But there is another way to access it
How would you rank r1 and Claude 3.7 reasoning but in perplexity,not directly at ai providers
openAI announced o3 with o3 mini but it was never released. I am a person which dont really cares about benchmarks. I test the LLM and decide on my own results. If you have some secret information about o3 i would love to hear how to get access.
In perplexity def R1. But also here you should test a bit and make multiple request with same prompt. I still have the feeling R1 is like a "Teacher" or better explaining stuff. It always gives me a very detailed and clean explanation. obv 3.7 Reasoining is Universes ahead in Coding than R1
I also noticed r1 is a bit faster than Claude most of the time
R1 is apparently default digital assistant in Xiaomi phones from what I've heard,or at least they started to implement it in newer Xiaomi models .
Btw,I just realized how much more advanced Chinese are compared to us...their ai in phones can actually be very helpful unlike what we in the "west" have atm
Is there any way I can see a breakdown of API use per day / per key etc? I don't believe my API usage is correct, is there a way I can break it down
What is this supposed to mean LOL?
Yeah but I don't know what I'm looking at lol
And then it will be only a little bit more clear
Can you throw me a bone here? I have no idea what any of this means
That's like complete profile stats that can be accessed only by clicking at this link .
Do u have pro subscription?
Okay but 90% of it just says false and it doesn't tell me anything about api usage from the look of it
Pardon my ignorance
Don't worry,I understand you completely.
It's the inner pain in you caused by user experience in perplexity
😁
Perplexity ai is known for its brilliant transparency regarding every single detail of their product
Now I'm being trolled LOL
Honestly I can't help you much when it comes to API because mine 5$ I get each month for being a pro subscriber are being wasted
Chatgpt couldn't even tell me what it means, makes no sense because it's talking about gpt4.5 and turbo and none of those are api models
Lol last month was 85 and so far I'm at 37 and the month restarted 1 week ago
Deep research, it’s powered by a version of o3
No clue how different it is in pro and plus tbh. I used pro and no issues
No known info on what reasoning level of o3 it’s set to but we can try inferring on it’s performance
Isn't it o3 mini
Seriously Perplexity Team?
I asked in a fresh new chat to recap the old chat, and help me setup my Yubikeys on Linux (arch btw) - as well as my mail certificates, and THAT was the result ?
If you want to set a system prompt, instructions need to be added to a space.
But I don't think that affects what sites it searches anyway
I have everything set up - 2 weeks ago i was able to code with that.
Because the site searching is done before it invokes the model
I feel like the model is not smarter than any 32B model
How long have you been using same space?
No, if it was o3 mini, it would be really bad ngl
O3 mini is so bad
I set up a new fresh space and put my Linux (arch btw) , setup threads in it, because well - it breaks often and i need to re-do things again
This was claude 3.7 sonnet reasonning
Context window is 32k
I had ARI make a guide on using it because ngl I didn’t know what to do with it lol
My project for an automated NLP bot, able to do multiple tasks is running in Space - with more than 100k context, there are threads with a scrollbar, that i can not even see.
So - the model, and the prompting is always the same.
I have even provided more information.
And one thing i recognized: Why is often the first search source the capital of france?
32 is limit for not being autistic
After 32k it forgets, hallucinates,lies,forgets,forgets etc
You have created an amazing thing with Perplexity, but it feels like , - too many request? Even though i switch the space to run entirely with a specfic model its jumping to auto, on every prompt.
I know that its better to ask more things in one message, because it might read everything again and again. So, bit sad. Because i love your App / Project.
Dang
Roberto , you can check if there is any bug if you want.. My acc is linked to actually everything
Hahahaha
it probably greps results first then initiates the LLM
yes, thats what it does always. Especially in Spaces from what i expirienced.
But the most akward search is always "capitol of france", wtf. im from germany - no vpn, no france nothing 😄
stop texting me
btw you guys heard of this bug with sonar 3.7 ? while coding there is a command (dont know if its fixed), that "chown" execution kicks / locks the model out and breaks the conversation 😄
hahahahahahaa
i do this regulary with ai avatars on twitch 😄
Can someone run a deep research comand for me? Would like to know if there is a difference
between high and the old one the difference is in source amount and reasoning quality between search steps
and its quite a large difference
check this - somehow ignoring that my settings are "english" , my prompt as well
the old one the quality is so bad just using claude reasoning beats it...
yeah thats a bug on their side
😭😭
im sometimes getting a different language output
or the reasoning is done in another language but the final result is english
i see this the first time, im just used to it - when Qwen talks chinese to me, out of nothing 😄
yeah deepseek sometimes decided to think in chinese when it was released lol
I have one "high" Deep Research left, about a Domain Value analysis, it literally thought and wrote for minutes, macroeconomic analysis, industry analysis, psychological impact, etc. this was awesome
An upgrade is coming soon
one?
you can see the amount you have left?
subscribers have a pretty high limit
no, i just delete chats that i dont need on a long-run, and i discovered the "high" feature for Deep Research too late, unfortunately
When the upgrade arrives for deep research on perplexity I’ll test it against deep research from chatgpt, gemini, ARI from you.com, grok, etc.
because atm perplexity deep research is at the bottom compared to the others
that would be awesome. But i can imagine, people tend to use the strongest models - and then there is serverside too much load. Btw. someone has a "real" Claude subscription? Would be nice to know if its worth it, on top of perplexity
yes people tend to use the strongest models after paying for it

and ask things like: why cant men give birth
it doesnt matter tbh
if you paid for it then you can do as you please untill you hit the limit lol
thats why its capped to 500 per day
incase someone actually does try to ask something the entire day
you reached the limit? 😄
lol no i run out of things to ask after like 3 requests
other people seem to be around 30
for more serious work
i think with the hand free speech feature you can reach it pretty fast 😄
the waiting time would still limit you quite a bit
and you also have to understand what it spits out
Whats your use case, I use claude 3.7 and claude 3.7 extended for different uses
Both are good ngl
I use clause original platform because jailbreaking is easy there and I like claude extended jailbroken ngl
automation of workflows and creating concepts (in near future) for clients
to have it unrestricted or what do you mean with jailbreak?
the high feature? where’s that
it'll hopefully be out next week
sure with the o3 ? 😄
Yes
Yes
It became easier on claude after a while ngl
so worth to get claude standalone?
Does the paid Claude subscription beat ChatGPT, though?
To be honest, I do find myself pretty limited in the ‘Plus’-subscription.
I find myself switching from o1 to o3 and from 4.5 to 4o, just to prevent reaching my limit too quickly. It’s getting quite annoying.
Claude is alright
Only now do I actually contemplate either going ChatGPT-pro or Claude.
I have the subscription
If you wanna use web search you have to turn it on then say use web search then your question
daily struggles 😄 😄 i exactly know it.. "just dont ask too much with o3 mini high"
@ruby cove good to know, maybe will try it - the extended 3.7 is not in perplexity or was never in, right?
The limit on o3 mini high is quite… high.
I have actually never reached that limit. o1 on the other hand…
And not to speak about their deep research, which takes an eternity before you get new chances…
3.7 extended is the best model in perplexity...
Hello
Ever since ChatGPT introduced ‘Pro’ I find myself unsatisfied more frequently as a ‘Plus’ user.
makes sense considering o3 mini is not much more expensive than 4o
i thought there is a extended version of the 3.7, so exclusivly for the subscribers - thansk
thanls*
thanks
extended is another name for reasoning
Hopefully they add in sokn for the ios app
Website it’s out
although the one on perplexity won't be quite as intelligent compared to their own service
The coding agent by claude is decent ngl
Extended is alright on Claude better than perplexity.
Made me remember a bit of classic video games
Also what I mean by extended is going to reasoning then clicking on Claude 3.7
In perplexity
✅
why did preplexity keep switching to auto after each prompt?
when I select deepseek R1 and enter, and then again when I want to enter something in the same chat, it does not retain the R1 setting. This is really annoying
I have to manually change the setting for each prompt I send
To save money!
this really ruins continuity. the answer quality and the answers themselves change drastically when I dont use the same model again. pls fix this asap
atleast allow continuity for cheap models like R1.
I just posted a question in the quick questions channel that I'm not sure I posted in the right place but getting an answer to it is time sensitive and important. I'll be happy to contact whoever I need to contact if anyone here can read my inquiry and point me in the right direction?
Send email to their support team
OK. I'll do that but I'm pretty sure that's if it's even possible to help me it will not be possible if they haven't had time to address my email fully before about midnight tomorrow
and when I say important, I'm pretty confident based on the conversation that I lost that it may be potentially life threatening
If the technology isnt there to retrieve the data I can accept that I made a big mistake and I have to live with it but I don't want to miss a chance at resolving this because it took someone who is probably working as fast as they can a few hours too long to read an email.
I'm drafting an email now but if you know of anyway to get someone to put eyes on it before to much time has past I would greatly appreciate it. I fully recognize that this was my fault and the perplexity service did nothing wrong, but also that the consequences to not solving the issue could result in my own death and I'm asking for help
Hey @agile quarry!
If you find the original message helpful, please consider reacting to it with the :star: emoji. If the post is appreciated by the community and receives 5 stars, it will go to the https://discord.com/channels/1047197230748151888/1082806833938436228 channel and the post author will get the <@&1082034222778302614> role on Perplexity.
Hello, everyone. pleasure to meet you here.
- Full stack, Ecommerce Speicalist
- Web3 integrating
- 6+ years of experience
- Open to work - full time/part time/contract
- he/him/his from Japan
- member of a Certified Talented Dev Team
Warm regards, Yamada Fujio
so counter productive and coefficient
i hope you quit ai entirely
If you are sure that you are in a life threat shouldn't you go to the hospital already? just to check you symptoms?
not imminently life threatening. the AI had put together a bunch of information I didn't know was relevant and that my doctor's never asked about which all added up to indicate that I may have extremely early onset Parkinson's disease
The problem now is that all of my symptoms that I've been struggling with for the last year and a half are gone and have been treated and I feel fine except for one minor thing that can be treated by getting a little bit more sleep at night and then that goes away too so my doctors are disinclined to investigate further when I have no symptoms
while I've had some medical issues in the past, I had considered them to be resolved and the question that I went to the AI and the original prompt was about potential causes of my right eye twitching occasionally and being annoying. but It kept asking follow-up questions and each follow up question resulted in an answer that led to more follow-up questions
The conversation mostly went like this:
your doctor is right you should listen to your doctor's advice. you should pay attention for the following signs and if any of these things happen you should see your doctor again. and then the AI would list five things and three of them were currently happening to me but seemed irrelevant. so the AI would think about it and then tell me that it was probably also fine but that I may wish to see my doctor at some point in the future but I should look out for these other five things and if any of those happen I should see a doctor much sooner and three of those things were also happening to me
and so on and so forth for about 2 hours
until it finally said that I needed to see a doctor as soon as possible and started to try to give me instructions on how to see a neurologist without a waitlist
because I'm currently on a wait list to see in neurologist and I won't get to see them until 2026
neurologist
and the 2026 appointment is for a general follow-up where nobody is expecting anything to be wrong
anyway, I got the answer that I needed from the support team and while it's not the answer I was hoping for I appreciate that they got back to me as fast as they did so I don't have to wait and worry. I have more information now than I started with and I will do what I can. thank you all
on the bright side , Perplexity managed to call out not one but two extremely rare conditions to warn me about and I happen to have both of them which strongly suggests to me that it's not just a coincidence and that something significant maybe going on. and if it turns out that perplexity saves my life I will do a tour around the country singing your praises and talking people into buying your service lol
All because I asked a question about a twitchy eye
can i have the perplexity chatbar on mac stay like chatgpt does? instead of popping up the main window
i dont understand why deep research is generating answers bit faster*, did they increase CPU GPUs? or did they reduce model quality?
damn that's actually pretty nice for research especially if it is voice assistant chilling in the corner ready to listen to any of your questions
If they haven't done this yet, you can try this on Chrome - https://chromewebstore.google.com/detail/popup-window/nnlippelgfbglbhiccffmnmlnhmbjjpe?hl=en
this extension allows you to make pop up out of any window
\
Hi, I am new to the platform and I started using Reasoning , but it swtiches to Pro while asking follow up questions. Is there a way I can keep 'reasoning' in the entire thread ?
For those who need referral code https://perplexity.ai/pro?referral_code=J9ID1YP6
Also the question (and no, I didn't find an answer in their FAQ)
Are codes like one-time thing? Or you can use the different one every month👀
what is the obssesion of this plataform with Tim Cook?
Ok.
Leet me show you something interesting
last sentence killed me 😄
Team Perplexity - i appreciate you, but what is going on? On Thursday last week i had impressive results, suddenly the model acts like 8B finetuned
hey guys, I need your feedback or help: I'm using Perplexity in Firefox. An "internal error" responds to any request I have been sending for three days. I have switched onto Chrome. A new error occurs today: although the UI behaves as if the submitted file were successfully downloaded, the model (here: GPT 4o) says it can't access the content of file!
If all this happened under a same thread in a same space ,then it is completely normal
how to get that blur bg
I have threads longer than this one, and even if it would be normal (which is ok), then its drastically decreasing .. .. or you had similar situations?
linux, hyprland (window manager), very light weight desktop
damn 😄 its the honest AI model ive ever seen haha
feel like: broo, no worries - happens, its monday
Firefox is generally facing a lot of problems these days
thanks for your feedback @limber monolith
you using VPN? Mullvad DAITA?
none of them
that drives most of the pages crazy because the paket size is always similar..
But if its not VPN, then do a clean browser reinstall. if on windows, i personally would use bleachbit (with caution), and delete all the "Old" temp files, and clean cache
not exactly related to the honest response, but your Space there is triggering my curiosity. Are you being able to quickly apply to jobs with that? Love the idea
@limber monolith do you mean it will also solve the downloading problem I've been experiencing for several hours? The model says that it can't access the content of the json file I apparently successfully downloaded
I sent out about 30-40 job applications per day. Tailored to my CV, plagiarism score very low, no AI detection - and i trained it to adapt more professional german behavior (grammar)
Thats why i say: Perplexity is awesome, i hope their own model gets feeded good,
cant guarantee actually..dont want to guess
This sounds real good. I have been doing some automations for LinkedIn, smart auto connections
Got any success until now?
I managed to get a Software Engineer role at Siemens but no success in proceeding into my carrer until now, seems like most Jobs on LinkedIn are ghost roles
not even a return
after all, maybe it's a matter of patience
Yeah in 3-4 Interviews, but for the effort its still nice.
The idea is to fully automate it, with a dashboard (which is already coded by perplexity), and then put a good backend in.
Use webhooks for n8n workflows.
Its my first coding project, im not in IT. So i learned everything from scratch and mostlikely wrong. but claude sonnet 3.7 can truely help beginners to learn code
takes time 🙂 Linkedin is very busy, always
Keep in mind that 32k window includes everything from that space description and also your own ai profile description .
When you add to that your first question and every other follow up question and every answer to that questions , 32k gets filled up very quickly .
I'd say you have original question plus maximum 5 or 6 follow up questions with reasoning model and you are basically there.
With non reasoning models maybe you can get to 10 or 11 follow up questions depending obviously on what are you even asking.
as far as i understood from the AI even, it includes the inpout and output within a spae
space
- complexity
literally what the AI told me for debugging 😄 nice. .
I think space,your ai profile description and current thread in that particular space
Hello to the entire team,
and congratulations on your efforts if you are currently working to fix the double issue with accessing the contents of downloaded JSON files that I am experiencing:
- for the past three days, any request or file uploaded to Perplexity via the Firefox browser results in a blank screen with the words 'internal error';
- for several hours now, I have noticed the same issue with the Chrome browser. There is no 'internal error,' but the interface behaves as if the file has been successfully uploaded, except that the model claims it does not have access to its content.
I hope this information will be more helpful to you than it is to me, as I am completely at a loss regarding this problem.
Wishing you good luck and a great start to the week.
Marc
yeah but once it was so weird, that it mentioned the project(in the space), in a normal content window..
So its definetely not the same quality as before, thats for sure - but with the bunch of people discovering perplexity, thats kind of normal...(i hope)
So when u get close to 32k limit tell it write simple and concise explanation so that another it can understand exactly how you trained it so far and simply start new thread in that space with his response
Thats kinda how you can maintain somewhat same response quality for much longer.
Eventually after like 10 new threads it will probably decrease in quality but it is what it is
Hey seem to be experiencing an issue the headphone icon is gone and I can no longer Listen to Deep research articles
Have a look in the channel "feedback-web" , it told me literally the configuration which is used for my space, and to adjust it
i should stop posting that much screenshots, but its impressive how this AI works, totally flexible - super exited what the future brings
@elfin gate
Yes and btw space description and ai profile description are not included in 32k context limit.
That's a good news

yeah.. i started to prompt it now to summarize the learnings, gave 2 learning shots - and documented it, will delete the old chats and work in smaller chunks for better results
is the web down?
Yeah, web is down.
Why can't I use it?
Website under maintenance or something.
Getting something went wrong try again later in iOS app too
yeah also noticed that.
I can't connect to perplexity on the Android app
I think It affect both application and on browser.
It's under maintenance
Does anybody have any idea when it will be back up, my project is on the line
I know when it will be back up
?? Huh
As soon as they get it done
👀
What time is the check?
Your question should be directed to perplexity staff ,not to us peasants
It should be working again
yup
I just thought my account was banned 😂
So, never?
I saw the news
It's working for me
😑
working
same problem but a ctrl + f5 flushed the cache and all good
works for me ¯_(ツ)_/¯
Working now?
Good job perplexity.
It still doesn't work on the Android app
ah, just now it stopped saying my thread is gone. took a bit longer than for others. (web)
It's bugs
dead again?
hmm, I jinxed it. only one thread worked... 😩
me too
i think we should wait for announcement
The space and library are still blank
still not working on app and web
is perplexity not working for everyone as well?
yeah
They need to hire a devops engineer who knows what they're doing because this happens too frequently
slowness of system
down for me, nothing working
Looks like the developers could use a box of repellent.
Did they say anything about this here kn discord?
I picked a terrible time to do experiments... half my opened threads (5?) errored out and redirected to home. god damn it
no
nothing new from perplexity lmao
I believe everything Is normal before this happened, what happened that requires them to do system maintenance out of sudden that blackout entire site?
Ik
Is perplexity down right now?
It's frustrating tho
Developers seem to ignore this server. I hope they lurk at least
Bruh I have a big assessment tmrw
time to switch to poe
I wish we could have got atleast a heads-up that they will be performing any maintenance. For future reference, a heads-up would be helpful.
What's the point of having discord with ton of people who are subscribed then....
This has happened several times.
This is very ridiculous gahaha
are we back, chat?
😂
I mean they probably don't care and don't think this is a big deal they would be right if this is first time their communication with users and community was bad
lets all start opening up bug reports
to be fair, it seems higher ups want tons of features tested (on users) in parallel. I saw tons of those flags. a small miracle this doesnt happen more often
Yeah, it's not a big problem.
Here kn discord or?
youre just generally a miserable person, arent you?
My data is not important.
What the purpose of your maintenance exactly, I think system operational just fine before at least could you just told us before you do maintenance because my schedule Isn't funny.
That's nothing new haha
It works on the iPhone app
Agree
They have like 3 people managing us here ,it s already chaos
tons of crap can go wrong specially on a service this much used.
😂
They could hire some pol
Staff is expensive if you're planning to buy TikTok
They have a lot of money for engeneers but problems can still happen
Even though computing fosters individualism, impatience, and a lack of empathy for others' work, I prefer to swim against the tide and wish the team the best of luck in resolving this issue, hoping that its cause is not malicious.
I don't really mind a maintenance, just notify users before because this Is getting unpredictable.
Imagine If I have deadline In one hour and suddenly It poof out.
From third world countries if it's that expensive
This is true
I hope one of 3 mods here on discord is looking at this and does something about it
I think they just ignore our msgs dude
Nooo i was doing a math test 😭
They are community moderators, I think It not related to them.
Or I might misunderstood.
Me rn
What's the problem guys? I'm new and I don't understand a hate against perplexity. Is it common for them to do the maintenance in the middle of the day without giving heads up or people are just angry because they should be?
why is the candle on a teacup
I'm trying to be nice 😂
Lmao ikik 😂
perp perp perp perpLEXITY
Hope the maintenance doesn't dumb down the AI model or reduce its quality
It's common that you will find yourself soon with no proper info about model you're using
Because Imagine If one of people have deadline In two hours and suddenly without head up, we doesn't have materials for researchs.
Some of us have schedule man.
Perplexity back up for me
same
I gonna use other things In compensation and wait until It finished maintenance.
is something new coming by pplx?
It's like accumulative damage we suffered by this minor, irrelevant,small annoying things that can eventually cause a revolution 😂
I hope so
It is even more uncomfortable for pro users
If we all think positive,positive things will happen to us. So let's imagine they are introducing grok3 and gpt 4.5
PPLX started working 🚀
it didnt work for me
All hail the king
App not working
working for me🧐
indian stock market is crazy
Someone should use sonar and ask it what Happend and why we were not informed since they flex with how good it performs with real time information
How do you know I'm trying?
I think something unfortunate happened In system when developers are making changes or something, because this Isn't very often.
Like accident.
😂
perplexity is finally working for me
working for me on the app and web both
is there any way to get supergrok for free or massive discount?
I got it for free
how?
Yep same
.
only for a year tho
what
It will take time ig
sorry for my english
T-T
i have sent u a msg
sorry I dont accept DMs.
ok, but can u tell how to get a year free?
yeah
Operator Activities
what is this?
are u talking about perplexity or grok?
I did it too
no
🤣
It's very sh#t actually when it comes to real time news
Qwen v3 otw
I know these solutions can be used
I don't even know how to choose the best mode
Sonar
I used sonar pro
Not good
ok, i know this but this is for pplx, i was talking about supergrok
They flex how sonar is best for real time searches
After maintenance, deep research is still the kind that thinks super fast but with very few thinking steps. It only performs high-quality, detailed thinking in the first thread response. From the second thread onwards, thinking steps become minimal, thinking speed gets super quick, and the answers end up disappointing! I wish I could use the deep research from a week ago tbh
Oh sorry, I misunderstood
Like for info about things happening right now
Sorry for my poor English
where u from
Don't apologize for that
China😰
do u use vpn to get access to discord?
I even had a hard time subscribing to the pro plan
y
I dont recall exactly from the top of my head but if you have an account with X premium and using that same account to enter an AI coding competition you get supergrok at the expense of Elon
xianyu
the X premium part actually isnt needed
ok
yeah its easy asf btw you dont even need to be an experienced coder
因为之前有教育,校园活动,有一年的白嫖
Brutal American
Yea new name for Americans
(
is winnie the pooh banned in china?
Maybe
yeah
🤣 💀
Xi Jinping doesn't allow us to do this.
Are you Chinese @viscid sundial
y
Cause if you are thats pretty cool
Dont have to respond to my question, I understand
😉
👌
🤝
is he a good person or a bad person?
I will only answer if I want to die, but if you don't know my identity
bad(
But not sure
Fortunately, I didn’t say such things on WeChat.
If you tell the police you might get arrested.
If you use the perplexity deepseek r1 model, you can see the r1 description
dangerous , wish u all the best🙏
oh thank you
Why
curiosity 😅
Does anyone have any tips on how to cancel a Perplexity Pro subscription? My Manage Subscription button does not work.
U for real!
web
real
Why would you want to use perplexity
Okay,I assumed that
Like deepseek r1 qwq
O_o
And we can’t discuss China’s dark history
They are superior to what you get here with pro subscription
How dark you guys are damn, haha
I thought only westerners were war criminals...
Not only is the ability not strong enough, but there are also many restrictions
Just read history
What is context window in r1?
The question answers itself
what
If you go back enough you will find bad stuff about everyone on this planet
Perplexity sucks balls.
Exactly
It answers itself
China is not allowed to use chat GPT claude gemini and other models
Because they can output politically sensitive content
This is extremely dangerous for China.
R1 model from you guys is available here on perplexity also,but it can only be good until 32k tokens. Then autism strikes him and he cantremember anything before that
Our good?
You mean
32k tokens is like main question and 10 follow up questions approximately
Yes
Thats number 1 problem actually .
I think you guys have 200k or whatever . 128k minimum
Probably much more tho
Of course it is. Because everyone already knows everything regardless 🤷♂️
y
Haha, I thought you didn't understand
It turns out that everyone knows
This is sad
Nah im just poking. Honestly I say do what you gotta do to be safe.
This is bullsh#it
As sad as that it to say because words can simply get you killed.
hwha
Or imprisoned.
All this models from western companies is extremely pro western if you get into deeper socio political problems
Oh, that's true
why r u defending china so bad lol
Because I don't like bullies
And r1 is more about Chinese data
u look like paid agent of xi 🤣
To be precise, I prefer Claude's coding ability, Chatgpt's math ability, and Gemini's natural language ability.
That's the most ironic thing I have ever read in my life
DeepSeek R1 knows more about Chinese native information, internet memes etc
Do you like paying too much?
Luckily I live in a free society wherre I'm allowed to express my views
🤣🤣🤣🤣🤣
Wupp. perplexity went down for some time?
y
You can ignore it
If you are not using
👌
Yeah I was using it and doomed because I thought my VPN might got blocked by chairman xi
And then check discord server see others have same issues
It's so sad.
sosososo
have you met the chief security officer ? 😂
China number 1
😂
VPNs market is going crazy these days
few starts up made seed fundings just to change IPs. lol
What should north Korean citizens say if you're complaining.
😂
Would you prefer for us Americans to use Deepseek?
However, North Korean citizens do not understand
Not recommended
Really, why is that
I don't know what data it was fed.
Is it about information
I should do a proper research before defending china . Brb
What do you think it lied to u
That's crazy when even a resident know they were bribed😂
In addition to making up stories, he can only answer questions about foreign politics.
Some North Korean individuals do have access to free world internet, the infamous North Korean hacker that funded by their government
Trust me,you will find nothing good and decent here in west.
where u from?
Freedom of Speech and Education in China
There is also the issue of employment
Especially education
hi guys, how to generate images for my perplexy generations?
Whatever you want
now in the right side I don't have option to generat images again, any help?
ICMP Pong from Beijing at 22:30 PM bro
Choina
No clue what that means but I fw that
Woah we’re twinning!!!
Every country has its rot
The power of China
For real I'm living in Beijing right now
🤡
I want to move to uhhhh what was that weird puzzle city name I forgot
the one with uhh verticality
You are so bold
How dare you tell me your location?
nonono
which state in china u from?
I will not harm anyone I use a VPN with
Zhejiang
Not state, province
y
thanks for info
I thought you wouldn't welcome me talking here.
?
Because I am from China
I thought you didn't like it.
no cot
Gemini 2.0 Flash Thinking deep research output quality feels okay
Expensive and worse
Just don't know why
Do others openly make threats or something when you do?
what
When you said this, made me wonder if people are aggressive for you simply speaking.
I've seen people treated this way in other communities, and I've experienced it before.
Probably mostly a problem with Americans
Ah. Well, I get that. Guess people cant accept the truth or listen to themselves half the time because they don't want to believe it.
I just ignore them and let them scream at the frickin sky. Or laugh at them and bait them cause I have no life myself 🤣
Yep
Maybe it’s because of stereotypes
is that destiny 2 the drifter
That's exactly why
It used to be
it's allover the screen
hell yeah drifter i fw u heavy
What weird questions do you ask with perplexity?
Departure from the original usage
i ask it about concepts
you wouldnt get it
🤣
I invoke the law of silence for that one
A what…
Can you not find that in Chinese -based webs….?
perplexity is my hypeman
Anything image related I usually just use GPT or Gemini, not a fan of perplexity for that
that's fair! I use a mix of both chatgpt and perplexity mostly
Yeah same
I often chat with perplexity in the space
More than just search and research
And then...tutoring?
okay but you really want to play this game? i want the O3 Standalone not a variant which may or maynot perform in Deepresearch.
Hello guys.. Is AI going to kill accounting jobs?
accounting jobs? No,
Accounting jobs as we know it? yes.
being an accountant is much more than just balancing numbers and sorting data.
heyy, im new here, can i get a guide about the server
Heya, welcome to the Perplexity Discord!
There's a pretty convenient guide for navigating around the server here if it helps #😃│welcome message
My first assignment was enter dates in an excel sheet with 10k entries. I was 19 yrs old, I was sent to a bank, had great lunch and snacks. But entered dates all day for a week. I hope such jobs are killed first..
Guys help! I cant see the image prompting window no matter what! I tried both mozilla and edge, turned off all extensions, tried in incognito, cleared cache. NOTHING helps. Last time cache clear and turning off extensions helped. What now? I need to make an image
Open a web search and ask a random question to find the image generator in the gallery.
China number 1
No I am not
😳💥
I'm living here tho
Communism changed quite a bit from the one I learned about in school 😂
Chinese Communism is very different from what you think Communism is
I know in just trolling
You have best tech in the world,best EVs ,soon probably best ai ,and quality of life is better than Netherlands,at least for someone in my profession.
Quality of life is not necessarily
I assumed it's like that already but I actually took some time and checked it
Why
@left stream classic example of abusive parenting not working.
Because they poisoned it's reasoning ability with their propaganda to a level where it simply can't be objective and unbiased anymore . 😂
When I say they I mean American lead western propaganda followed by its minions from EU
Aye
Glad you used the word minions.
I always used to say nato and it's minions, lol.
Minions started to be loud lately tho,don't underestimate us by any means.
Hello everyone, do you know if there are any coupons you can use for subscriptions?
I can send you my referral but I have no idea what discount you will get
what would be forgive me?
I think you will get 10$ of total amount but I'm not sure.
are you Italian?
10$ on the monthly ?
Yes
I think. I don't know because j currently have yearly subscription active for next many months
On the web there are no coupon codes you say?
Wyn
There are
Oh you mean like free trials and stuff?
I mean discount codes for pro versions
I have no idea honestly
I know I can give you 10$ discount if you want
thanks anyway, i don't open link if i'm not sure, thanks anyway for the helpfulness
Yeah how to disabled suggested prompts
Idk why the perplexity devs dont just vibe code a better ui
Check dm
Why are you writing to me in private?
can someone help me with generating imaes. i have been trying from 10 minutes, and the option to generate images isnt appearing
You have to do it on web
View as desktop site if you're on the phone
why is it possible to generate images? with which option?
using from mac m1
pro users get that feature
I surrender then
Have you tried asking ai how to do it on Mac m1
yes I am pro, but the response was that it could not generate images, I always use from the web
i need assistance with my graphic design work. and i uploaded a few pictures as the reference and asked perplexity to give me a prompt for generating images as given in this post. it didnt work
I saw the message, thank you
when will perplexity implement grok 3🤔
What the heck is with the search breakdown and sources removed from searches?
If you find out let me know
grok3 isn't even fully operational yet on their site
better than grok 3 Deeper research?
i'll tell u this, a basic prompt = 300 sites searched and a 30 page report with all good sources
That's crazy
will try it out
@sturdy skiff
Huh. Can you share the prompt? I'd like to test it pls
whats happening
I remember you not being a fan of Gemini and then I see this gentlemens interaction
lol i missed it man, can you summarize interaction for me pls
Start reading from here. Hha
😂
It's already summarized
where is file upload files in research in gemini bro?
great help
That's what I do
is perplexity working for you guys
nope
same
Noep
My Roleplay D:
okay thought i got banned
Again mini maintenance
idk if i should laugh at it
Huhh
lmao
Didnt work, I know. I had one thread that I made specifically for image generation
UGH
now?
Great, how am I gonna finish my website updates for today with perpy cleaning itself 🤪
Comet has clearly gone awol and is attacking Perplexity’s data centers smirk
wait someone check if the app is working
It’s not
fk me
Nope app is wiped as well
POV: using perpy as my brain 🤣😭
I'm so cooked
Sameeeee
I'm having trouble with the "sonar" API model. Perplexity team won't help
like fkn give a hint about it. tell us it's happening
Yes exactly 😩
ion wanna use deepseek 😭
they are doing maintenance again:
$10 monopoly dollars they broke something but are using “system maintenance” as a cover
probably an issue from when they performed maintenance earlier
We’re back crew
imagine us getting o1 pro after the maintainance
i should prolly stop dreaming
someone hear praised gemini, thats why earthquakers are coming and apps like perplxity are not working, please take back your praises about gemini and save the world
Comet was not pleased with the Gemini praise
we are sorry comet
If they give us o1 pro I will run naked for two blocks
Man I have exams 😭 seriously
video proof
Was gonna say the same
i am ok even for o1
need help, which is good for writing report reasoning or deep research?
"0 enhanced queries" this was added and instantly removed just now 
Cannot wait for this moment to arrive 🍿
@perplexity should give us o1 pro and sell his video to cover the expenses
site is fixed?
how is this @elfin gate
app works for me rn
BTW who woke up with deep research high option being added to their perpy👋
lucky bastard lol
Finally it's working 🫡
complexity shows 500 for me
Same here
wait that's peak
they have specific features for reasoning and deep research.
i thought everyone had it
Prolly rolling out in batches to different accounts
had it for a bit but it disappeared
why in original perplexity this is not there
is it good tho?
so complexity rolled out before perpelxity?
Well… I cant tell yet conclusively. Deep research standard was already p good
Seems like
ok going to complexity
for the few prompts I got to test it used like 100+ sources instead of the usual 20 ish
That’s great
took its time tho lol
thats crazy
Yes, it definitely uses more resources, but it doesnt seem to spit out more tokens (still cant get too long texts out of it)
guys do any of you even use gpt 4.5
I prefer it to take some time but to provide better quality answers
pretty useless for me ngl 
Yeah from time to time
But mostly for coding
what do u use it for
Unless I misunderstood, the model selector has been included with uses remaining for months now via Complexity.
why not 3.7 son
Depends, sometimes one gives me better code, sometimes the other
oh
It’s a shame it can’t output more tokens because I find the current one so short in terms of answers
yea the output limit is rly annoying sometimes, force me to split into multiple prompts fequently :/
i want deep research high, i can see that only in complexity, but not in perplexity
ong
Yeah, Ive been pumping our company webaitw with awesome blogs for the target audience, but I have to ask it to expand each section once it gives me the first result
Hope they will improve this aspect
On the bright side, our ranking improved by 15 spots on google 🤣
i suggested a idea once that give us a daily token limit for every model and then let us adjust the max and min tokens for output
like its really annoying
im a dream chaser
YESSS PLSSS
Awesome idea
hi sorry i just joined cause im doing exam prep for tomorrow, is perplexity down rn?😭💔
How logical do men have to be I wonder
It was a few minutes ago
fixed refresh
its back before your heart breaks
deep fry?
Just use Complexity plugin in Firefox or Chrome
bump
yes installed just now
I need to pay for every premium for each LLM
yes and grilled
now its back, hope you are fine
Thank you it’s working again but a bit slow. Alright bye 🫶🏽🫶🏽🫶🏽
ok enjoy bye
Good luck
tysm!!!
Ah gotcha. As I understand the extension will show models/options that are active on the backend but not visible from the UI.
yeah, can be
This is correct
Just use o1 pro in chatgpt lol
he dont have money, thats why he is here
Too expensive
Me too lol
$200 a month is crazy
Yep
LOL
It's worth it tho swear down
its worth it i agree, but there is no worth in purse
If you had enough money it would be worth
At first that was how I felt but now that I've been using it it's actually worth it. I also knew it was coming after openai quarterly report showed a drop in revenue
I bet Saudi prince is not struggling with sonar ,he asks o1 pro stuff like whats the time
he is rich, he will use it that way,
As a EU citizen, I couldn’t afford having to wait 2 months for a new tool to be released in my country but I guess it is worth for other people out of Europe
Wym
perplexxity deepresearch is equivally good as chatgpt deep research, and you get more queries just for 20 USD, when compared with 200usd
I'm pretty sure gpt is miles ahead
it doesnot have features like custom instructions and spaces on top of deep research, so it cant search as cleanly as perplexity, i used both and confirm this
chatgpt deep research uses less sources in my expirience but they are high quality and the analisis is deeper. Plerplexity Deep gets more sources that Chatgpt doesn't since half the world has blocked chatgpt in their robots.txt
They released Sora nearly 2 months (not sure of that) after it was already available in the US. Same with ChatGPT Operator.
chatgpt won by intelligence, failed by sources 😄
Yes lol
Sure should I dm?
Chinese ai will save us all guys
I didn’t upload any files yet
perplexity is really great for 20 USD , perplexity deep research is 50 times intelligent than my brain, 100 times faster than my brain, that is enough for now, and perplexity deep research doesnot hallucinate, in my case it never hallucinated , it speaks only what is in the sources
how much you try you cant upload in gemini deep research, all the best
What's token limit on it?
120 msgs per month, in chatgpt
I meant in perplexity deep research
may be 200K, i dont remember, but i think 200K in chatgpt deepresearch
500 deep researches per day,
sonar deep research it is, got confiramtion on email as 128000 token chat size
You mean even if its not deep research high its good sir?
500 times you can use deep research option per day in perplexity
yes perplexity normal deep research also good
does it still use R1 or claude reasoning sir?
I don't believe it's 128k
it has R1, claude and o3 mini reasoning, your wish anything you can choose
me too, because it focked me hardly in middle of long chats
i mean the deep research sir in perplexity sir, what does it use sir?






