#š¬āgeneral
1 messages Ā· Page 214 of 1
Ik
Ahh then it's alr I don't think I spend all
Hell no imo
So with lab you create apps ,and is there option to publish to Google play store or we need technical knowledge to do that? I am thinking still labs is only for basic purpose. Whenever I used it , i observed it is doing thrash
what even is lab and when do I need it.
U need technical knowledge to do that
also
for me
I used lab
aww... did sam get hurt by zuck's mma arm-bar manuever (i.e. luring away talent) 
it cover 70% of my expectation
Dear PPLX you're a search engine... I want you to search stuff.
Everything else I dont care
If u give it good enough amount of info
I want perplexity to replace chatgpt, but they end up being just for search and research
So who here got max?
I switched from ChatGPT to Perplexity. It took me two weeks to get used to it.
I'm max annoyed
I'm curious to see how o3 pro does anything different
And opus
Especially since they gave us the excuse for the context size and the nerfed models that they are a search engine and not a general llm service
Perplexity always forgets what's going on in chat conversations , so I still use chatgpt
Turn on
This will not work, this is only for memories, I am speaking about ongoing chat context. It's due to low context windows
Apparently the system prompt is so large that it takes a large chunk out of it
You could use free AI Studio with 1 million context window:
https://aistudio.google.com/prompts/new_chat
No bro I need extra security, so I bought till now only one of these in different phases of life
- Chatgpt teams
- Perplexity enterprise pro
- Google workspace
It looks the same and searched 15 sources, talking about o3 pro
But it takes a few mins
I think this option is not present in enterprise pro, where can I find that generally?
Account -> Personalization
Ok thanks and it's not available for enterprise pro, so sad
Depends on how satisfied you are with labs.
Not even 1 time broz I felt it is overhyped
wth is this shit
i just realised pplx released a "max" plan. had been inactive on discord for quite a while.
Damn, now the pro plan is gonna have reduced limits to make the max plan look more appealing !!! it has already happened with chatGPT before
who was the magnificent creator of such shitty idea? š
lucky me i just use github copilot
how can I use the labs feature of perplexity properly, I haven't used it since it got there
any suggestions?
money money my friend sir
We will be launching an Enterprise version of Max with unlimited Labs queries in the near future
https://www.perplexity.ai/hub/blog/introducing-perplexity-max
Perplexity Pro ($20/month) remains available with generous usage limits! This is just a step above Pro for those who demand limitless AI productivity and immediate access to the newest products and features from Perplexity
When will Max get comet?
yeah pro is good enough for me. its just that i think in the coming weeks / months they might reduce pro limits to push users towards max. as of now i have no complaints. but these generous pro limits might not last very long is what i am saying
within the next week or so - still prepping and making sure itās as good as it can be
Can you help with a problem in my account?
Y tried support at web but it's too slow
Mind sharing more details on the issue? Feel free to DM
There are some things we cannot assist with via Discor (i.e Billing), but happy to take a look for you
wait actually
i may honestly try it for amonth š
@lime kite
How much for the max?
Perplexity Max costs $200/monthly or $2000/annually
May be worth reading this
š¢
how do I request a comet role?
This is exactly why I came here. I'm looking to join an existing group or create a new one, focused on creating a great investment underwriting tool.
I think it's kind of lame that they've been hyping comet for months and now all of a sudden you won't be able to use it unless you have Max. Who knows when us pleebs with Pro or free will be able to use it.
Will perplexity max not have any promotion codes like the Xfinity one with pro? I mean $200 per month is outrageous and it's like just copying chatgpt
https://www.perplexity.ai/page/new was also an pro only feature first, then eventually rolled out to all users
@gloomy finch we are doomed https://www.instagramez.com/reel/DF287Jnv_c-
š¬ 1.6Kāš 0āš 76.8Kāš 616.2K
New viral captions šāļø
.
.
#वायरल #ą¤ą„ą¤°ą„ą¤ą¤”ą¤æą¤ą¤ #ą¤°ą„ą¤²ą„स š¬š§šļø IP ą¤Ŗą„ą¤°ą¤¾ą¤Ŗą„त ą¤ą¤°ą„ą¤! 2024 ą¤øą¤®ą¤¾ą¤Ŗą„ą¤¤ ą¤¹ą„ ą¤ą„ą¤ą¤¾ हą„, ą¤ą¤° यह ą¤µą¤°ą„ą¤· ą¤
साधारण ą¤¹ą„ą¤Øą„ ą¤ą¤¾ रहा हą„! šą¤
ą¤§ą¤æą¤ ą¤°ą„ą¤ø, ą¤ą¤ नया ą¤ą„ą¤°ą„ą¤, ą¤ą¤° #MICHELIN पावर ą¤ą¤¾ą¤Æą¤° ą¤ą¤øą„ ą¤ą¤ ą¤Øą¤ ą¤øą„ą¤¤ą¤° पर ą¤²ą„ ą¤ą¤¾ą¤Øą„ ą¤ą„ ą¤²ą¤æą¤ ą¤¤ą„ą¤Æą¤¾ą¤° ą¤¹ą„ą¤ą„¤ šą¤ą¤° ą¤ą„ ą¤
ą¤§ą¤æą¤ ą¤Ŗą¤ą¤”़ ą¤ą¤° ą¤øą„ą¤„िरता ą¤ą„ साऄ, हम ą¤ą„ą¤°ą„ą¤ पर ą¤ą„ą¤ ą¤µą¤¾ą¤øą„ą¤¤ą¤µ ą¤®ą„ą¤ ą¤°ą„ą¤®ą¤¾ą¤ą¤ą¤ ą¤ą¤° ą¤°ą„ą¤®ą¤¾ą¤ą¤ą¤ą¤¾ą¤°ą„ ą¤ą¤ą„शन ą¤ą„ ą¤²ą¤æą¤ ą¤¤ą„ą¤Æą¤¾ą¤° ą¤¹ą„ą¤ą„¤ ą¤ą¤ą„शन ą¤ą„ ą¤°ą„ą¤®ą¤¾ą¤ą¤ą¤ą¤¾ą¤°ą„ ą¤¹ą„ą¤Øą„ ą¤¦ą„ą¤ą„¤šš«š·
š¬š§ šļø MotoGP ą¤ą„ ą¤Ŗą„ą¤°ą¤¶ą¤ą¤øą¤, तą„ā¦
That's the point of perplexity. You can use gemini for conversation purposes, I believe that google gives a lot 2.5 flash requests per day (for free)
There is no copying that's what the AI market plan price, still Perplexity Max price don't match the offerings since its limited to search and research unless you are a professional who deal with high impactful projects, the Max plan doesn't seem a good option.
Not legit, this would lead to an account suspension @last spoke
Does it work for websites whose content are behind a paywall? Suppose I have the login details.
hello guys i want to ask one question.
what is the currently best IDE / Vibe Codding provider?
Either Claude code or cursor
i have gemini pro cursor pro and github copilot pro
i am every platform at once
https://webscraper.io/
You will have a better experience with this company. They have a free version - a browser extension and good documentation. I scraped 17,000 documents in one week with that extension. Of course, it took me about 1-2 days to learn how to adjust the initial settings. And after that, everything was fine.
Thanks
š¤«
Idk
Of course, you still have to come up with how to feed those documents to LLM (through RAG?).
So far, I use the datasette.io locally with the keyword search.
Only 5 extra sources? 
No search during reasoning
Still the same setup
lmao
- PPLX still refreshing while im reading its finished responses
- after refresh it positions the page at the top
- why devs keep doing this?
- multibillion dollar funding... still refreshing without user's consent
Idk if anyone has asked here but is there a reason why you can't upvote a thread on Android and only downvote for unhelpful?
If youāre a Perplexity Max subscriber, just DM me or anyone with Perplexity affiliate badge anytime here for help! We will work on triaging and helping you get what you want. Some more things like video generation, access to proprietary data, etc coming soon. Itās not just about
video gen woah
sorry but where is my pro when I get it for free for one year?
š
Who bought max
Isn't it paid though?
What are the limits for O3 pro on chatgpt?
i wish there was a /pplx [query] so we dont have to copy-pasta
/pplx what are the limits for o3 pro chatgpt
That's none of your business. Look it up yourself lazy human.

ok too sassy
I would appreciate it if you could at least confirm whether the maximum context length is extended when subscribing to the Max plan.
Considering the substantial subscription fee of $200 per month, I believe more detailed information about the services provided is necessary. I would be grateful for more comprehensive guidance.
If I try to paste the text from a 60 page pdf ebook I get this. But if I try to attach the pdf file it works
it doesn't say how big the context window actually is tho
I JUST DROPPED $200 ON MAX IF I DON"T GET ACCESS TO COMET TONIGHT I AM GOING TO SHIT MYSELF
be prepared to shit yourself š
FUCK
i think I saw someone saying invites may be starting next week
I hope u shit with prepare
Bc that's what getting u
Perplexity max only priority you to get more faster to access to comet
not instant access
...
I get why they made a Max tier (everyone is doing it), but $200/month feels unreasonable.
Also, showing the Max models in the model picker for Pro users is a bit of an irritating nag. Itās already on the left side of the site for those who want it.
Itās for the group elite
Even AI companies have a class divider. š«
All us poors use free or Pro.
True Elite pay for Max.
Hi ! I wanted to know how long is the cowldown on X (twitter) for the video generation?
hi
yo
google has some crazy deep research
it is not even finish and it check 251 websites
i think im going to switch from pplx pro to google one subscription, they also have notebooklm which is actually amazing for studying
@dry kestrel @green crystal did they remove image analysis for free users? Or is it simply not obeying? https://chatgpt.com/share/6865b540-0758-8013-a224-9384e2b83bbd
Google Deep research is better.
Surprised they got a discord
hi
End your life
bet
Lol.

Tell it to carefully analyze the image
Sometimes if the details are detailed enough it won't use tools
It's asking me for detail but won't use the tools š
Hmm
Wasn't tool usage purely a paid tier thing?
Can't remember having anything past thinking
No, it worked for free earlier atleast.
I used it for the electricity bill thing.
If you remember.
Nope.
Please share if you do find one!
what do you mean by image analysis?
Lol because it's human nature to down vote something that are not satisfied with.
I think it shuold still be available to free users
What bothers me more is that feedback doesn't support attachments. I lost interest cuz of that.
I'm not gonna type 50 hallucinations in the response verbally for the sake of feedback.
can i ask somthing here?
Test your luck.
.d
Not working now?
about mcp
Didn't in that thread that I shared.
try another image?
or different prompting?
Either they fucked free users or some weird pattern on my side has broken it for now.
Give me the image and prompt that you were trying to use.
Make a list of characters in this image with their levels.
Now it will name a few wrong. After that I ask it to zoom in and look carefully again.
But it won't.
Way too advanced for o4 mini I can tell you that for sure
Ill run it give me a sec
Electricity bill was harder than this imo.
I not mad about it being right or wrong. I just want it to zoom in manually.
Which it isn't doing for some reason.
Try 12 then.
I was able to get the image analysis working fine
@distant aurora https://chatgpt.com/share/6865bcfb-e5ec-8013-8947-52f4dccb592f
Tell it to zoom in on the character between Barbara and kuki
It was able to zoom into other characters, Ive run out of my limit and I'm not feeling like creating another account
if you click into the chat history you can see it was able to crop into the image many times
what is it
Or turn it off altogether.
Yeah thats going to mess up its tool call usage a bit
Is the gpt 4o limit prompt count based or token based for free users?
query based
Wait so do the people who signed up for the waitlist for comet not get it and have to go through the 200 a month plan?
Seems like it.
Perplexity Max is only early access to Comet - it is not a requirement for Comet access
You don't need to go through Max
Well, pro users submitted their email for early access to Cometāat least that's what I thought it was.
brothers and sisters on the max plan, not sure if you guys have experience with using pro then later upgrade to max. what do you normally use max for? and the use case for upgrading to max? just wanna understand.
If you mean via https://www.perplexity.ai/comet - that is a waitlist
Oh ok so that means the people on the waitlist will eventually get it and its just that the max users get it first? Cause if so im willing to wait
yes
Ok got it
So I'm confusedāa waitlist isn't for early access? If the waitlist is just for when it's released, then why do they ask for an email?
So we can email you when you have access
A waitlist is, by definition, a list of people waiting for availability or an invitation
We will of course roll out to Pro users before a public release (to free users)
What are the limits for the MAX plan?
- Unlimited Pro Search
- Unlimited File Uploads
- Image Generation
- Unlimited Labs Queries
- Access to o3-pro, Claude Opus 4, and new frontier models added as they evolve
I understand that a waitlist is typically for early access, not just waiting for the general release. With the MAX plan, it seems like only MAX users get true early access, while Pro users might have to wait a bit longerāis that correct? Also, if all Pro users get access at the same time, whatās the purpose of signing up with an email for the waitlist?
it seems like only MAX users get true early access, while Pro users might have to wait a bit longerāis that correct?
Yeah, that's right ā Max users are first, but Pro users will still get access before the general public, including Free users
Also, if all Pro users get access at the same time, whatās the purpose of signing up with an email for the waitlist?
The waitlist helps us keep track of who's interested and makes it possible to notify you when you're granted access
A waitlist doesn't mean immediate early access. It's simply a list of people who've asked to be notified or considered when access becomes available.
I thought unlimited pro search was already a thing in pro - or maybe I just never used it enough to hit a limit or is this something being given a limit and the benefit being moved to max
Thanks for clarifying how the waitlist and rollout work.
I do want to share some feedback as a Pro user: when Comet was first announced, Pro was the highest tier and we were encouraged to sign up for the waitlistāso it felt like weād be getting early access as a key benefit.
Now that the Max plan exists and gets priority, it feels like the value of the Pro subscription has been reduced after the fact, almost like the goalposts were moved.
I understand the business reasons, but itās a frustrating experience for those of us who supported Perplexity early and followed the process.
I hope youāll consider this perspective when planning future rollouts or communicating about new features.
I thought unlimited pro search was already a thing in pro
Yes, but they are limited per day (but otherwise unlimited)
You can do roughly 300 Pro Searches per day on Pro
Right yeah okay so I just definitely wasn't hitting the limit.
You get heavily rate limited after 300 Pro searches
Really appreciate you sharing this ā I totally get where you're coming from
You're right that a lot of early Pro users signed up expecting early access as part of their support ā and your support has mattered a lot. Even if Max now gets the earliest access, Pro users are still part of the early wave and will always be prioritized ahead of general release!
I'll usually get about 10 prompts in and make a new chat anyway as the AI will just start hallucinating like no tomorrow
I only really make a new chat about 3-4 times a day at max so I'm sure the limit won't be an issue and hasn't been thus far
Please keep in mind that Comet is still in beta
WHAT
?
I really should learn more about the ai community developments. I literally work with it as my job
Had no idea the model context protocol existed until yesterday
I thought there was no search limit on the Pro plan.
Same here - probably because it's so high you've never hit it
300 pro searches a day is pretty hefty
If you had no idea the limit existed, chances are you've never hit it to have known of the existence
Its my first month using the pro plan
They need to update their documentation
maybe i will
yes
When you hit it, you'll notice that some queries dont get processed and nothing is returned
Are you just asking general purpose stuff. I literally work with AI and it's still hard for me to hit 300 prompts a day (outside of automated testing)
I've been using perplexity for months now and I've never hit that limit and never known it existed
Lol, $20 dollars for a limited search
Then dont buy the sub
for general
It doesn't even give you a "sorry you're out of pro searches xoxo" just straight up ignores you? š
I'm sure you'll be fine
Its pretty generous considering you get access to claude sonnet 4
yeah, 300 a day its too much
You could easily get your money's worth
Yeah
Like youll just get a blank response
You can see your limits here https://www.perplexity.ai/rest/user/settings
Similar to the thing I get with the windows app. Sometimes after a particularly long topic chain I'll ask a follow up query and it'll just go blank and I'll have to start a new thread for it to work again
Assuming that's just average browser based app side effects though big webpage lots of updates something just eventually breaks
You should try to avoid long chats since hallucinations start increasing the longer the chat is
And theres a context window of 32k tokens for most models and once you hit this, responses start getting degraded
I've barely ever had it hallucinate but that's mainly because most of my queries are:
Here's the documentation excerpt:
Can you write an example from the documentation
Don't typically ever ask it to fix anything because then that's just asking for the tech debt of two programmers
Mainly just verify existing logic, generate mock data and write code examples from documentation
money's power~
Dear perplexity Support Team,
I am writing to request assistance with an accidental subscription change that occurred on my account.
Account Details:
Account email: vardhanayyappa@gmail.com
Original subscription: Pro (1-year plan)
Accidentally selected: Max plan
Date of accidental change: july 3, 2025 ist(7am)
Issue Description:
While reviewing my current subscription plan, I accidentally clicked on the Max plan option, which immediately processed the upgrade and terminated my existing yearly Pro subscription. This was not intentional, and I would like to revert to my original Pro yearly plan.
Requested Resolution:
I kindly request one of the following solutions:
Restore my original Pro yearly subscription
Process a refund for the Max plan upgrade and reinstate my Pro plan
Apply a prorated credit for my remaining Pro subscription period
Supporting Information:
Based on standard subscription policies, when upgrading plans, customers typically receive prorated refunds for unused portions of existing plans. Additionally, many services allow subscription restoration before the end of the current billing cycle.
I would appreciate your assistance in resolving this matter promptly.
Thank you for your time and consideration.
Best regards,
Vardhan.
Check with support@perplexity.ai
They don't check issues here
Holy yap
Aināt nobody reading allat
@timber niche I send you a message
?
an LLM reads it
and then they get a fully written reply
by the staff using an LLM lmao
Sam can only wish
Life is good i love work
Bro the amount of people that use ai is crazy š
Bro even in the hospitals when my buddy was showing me his work, I saw so much shit written with chatgpt š
They thought he wrote a script that took weeks long but it actually took a whole day for him š
And he had chatgpt plus
Joe can you check dm @left stream
Tbh gpt plus is the best all around for 99% of people
lmao this isnt going to turn out very well
It is, he landed his job as pacs manager without certificates or a degree š
I was on the meeting quietly when he was going over the script š
Ngl It gave me a wake up call when I saw how working remotely goes. Meetings everyday on teams while you do tickets at the same time š
Bet
A day? That's slow lol
Should be like an hour tops for script writing
Depends tbh because you gotta proof read it and clean it up a bit. This is hospital work so he canāt rush š
Headed to the fair tomorrow
new stealth model on LM Arena
steve - claims to be from DeepSeek
Steve does not perform too great
general consensus in servers is that itās a small DS model or a 3rd party distilled model based on DS
the latter would explain why it might claim to be DeepSeek even if it might not be
lmaooooo
Friends, tell me. If I get a PRO subscription now, will they give me an invite to Comet browser on Windows?
only max users will get access to comet for now, not pro users
that's a good point. Probably don't need that regular reminder of stuff you can't access š
I suppose it's useful to know what you're missing out on though, in case it becomes worth it enough for your use cases etc
No, Max users get priority access
Okey, thanks
Who has max? I want to compare it to chatgpt pro
ngl depending on the amount of veo3 videos you can generate I might actually get the max plan
What the heck you do with all those vids?
Isn't veo 3 on gemini?
For the men of culture things
where my doubt emoji š
aravind had a tweet
Yes saw it
Have fun with it
Veo 3 for max users soon
also potential commercial use
I keep associating pro with $200 plan lol
I'm just saying you can get way more fun for 200 dollos a month instead of some funny vids
Chatgpt pro is stuck
lmaoo
No I meant also commercial use
aside from the occasional stupid video
Ikr that's like 2 escorts
Commercial fine fine, is veo 3 much better than sora?
also shit like this
Its way better
not even on the same level
Best ai video gen Ive used so far
I get chatgpt pro from work, so I'm hesitant to splurge on more AI stuff in my personal life
Why can't I upload a image to Perplexity for research or a quick question
I guess something like perplexity max, claude max, gemini ultra could replace chatgpt pro
so far though chatgpt pro provides the most overall value
I got PPLX Pro from my internet provider for free for a year, not gonna turn that down.
Dang 3 chatgpt pro users
3 musketeers
haha
It's from xfinity right?
chatgpt pro just offers a far more comprehensive setup than the other $200 plans rn
Nah Sunrise in Switzerland, my jaw dropped when I had the free offer in my inbox
"I know that site" š
damn
Nice that's lucky hopefully they drop one for max š
I had pro personally for a year, but stopped because it felt like pplx just worked best with the default model.
Stuff got slow when I'd go for claude models etc.
Do reasoning models suck at meta prompting?
Pro was worth it at the start when you got unlimited API calls
ut not gonna pay for the model and then also for the api
Wait really?
wrf
Dang I missed out
At the start the API was included
you still get $5 monthly credist though
I just cheap out and use google gemini free tier api for my crewai projects
I wonder who ruined it for everyone š
Not me! I only played around with some silly discord bots.
I gotta start using it more tbh
Those $5 credits are op
I feel Gemini got the best free offer with their api
Yes
I don't think ChatGPT has a free tier
they train on your data though
No, the api on openai is crazy
I think after $5 spent they give you free credits
in exchange for training on your data though
That's fine, I mostly feed it irrelevant Werewolves(Mafia game) talk to summarize
gemini's privacy policy is the worst by far
Feeel free guys to train your model on "Sussy bakka over there is surely mafia"
haha
Google is the worst by far š
testing
no option to opt out of model training even when paying
and they view your conversations
So google of them
I gotta have this much of a budget for o3 pro bro š
It's a deal breaker for business I feel
what the fuck
4k?? I hope this is work budget š
Yep, I would have subscribed to their ultra plan
if they let me opt out and shit
š
No lol
It was personal hobbies lol
Could someone help me?
Yeah?
Why can't I upload a image to Perplexity for research or a quick question?
whats the error message that you are getting
Business: "Okay lets go with Gemini, we send our business plan directly to them"
Fr
No error, it just sends the text but won't upload the image
I wish PPLX could search images, I'd love to give it a list of celebrities or something and it goes off and finds a good image for each.
Thats weird got a screenshot or something?
Maybe a job for Comet
that is true, that would be something for an AI browser. I feel the big hole in my workflow process for training models etc. is sourcing material
training a lora etc. have fun getting images
worst part of the process
haven't found any AI product solving that
thats been a pain in the ass for me
never managed to fully learn how to do that shit
Also with building comfyui workflows, i wish i could just get chatgpt to do it
Correction. O3 kinda sucks at meta prompting.
you might want to contact support, or email support@perplexity.ai
Wait so the image stays in the prompt?
if its not detecting the image
Yes
that looks very much like a bug š®
Just a visual bug
the model still gets the image
How do I contact support on here?
ps. if anyone finds any AI tool to scrape/search good images, tell me
Yandex image search.
Nothing beats it.
the model should still be able to see the image
its just a visual bug
hahaha, I use it all the time for reverse image search, and I fully agree best image search out there.
there's the paid one? eyesomething?
Tinyeye?
but I wont also pay 15$ for image search
You are correct that when I typed "What do you see?" and Perplexity described the image to me. Doesn't seem to be right that I just paid for Pro and I'm already encountering a bug
yeah thats a bit annoying indeed
Veo 3 coming shortly to Max users
As much as I think max is overpriced, pro is quite a good deal using it for 2 years now. Besides the stupid bug in pages it works very well for me
https://battery-dashboard-fixed-2c291df4.scout.site/ @distant aurora
Comprehensive analysis of Android smartphone battery efficiency across all major skins and brands
altho I'm still very very frustrated with PPLX for not fixing the pages bug
@merry kettle It just seems like a poor start. I purchased the Pro I think less than hour ago
From my humble expirience it's one of the few bugs you'll encounter š It is my default search provider for a long time now.
@merry kettle I guess I'll attempt to keep an open mind but if I encounter a ridiculous amount of bugs then it defeats the purpose of using Perplexity and paying for the Pro.
ps. I can confirm the bug, just tried it myself.
It reads the image but it stays in prompt
mhm
Thank you both! @merry kettle @dry kestrel
Np
I always feel frustrated if I join a products discord, ask a question and nobody answers š
happens quite a lot š
I'll continue with working with Perplexity tomorrow, hopefully they get back to me! Everyone have a great night!
PPLX would be so good if it could also do reverse image search
We always got stuck on the loop " i'll transfer u to the team etc... bla bla " then the team never actually showed up
I don't think that's possible
how so?
I have always said this.
google image used to do that
Why not?
but now because of new policy
Tell me about this.
It is an Ai
I imagine theyāll have to do something similar to operator but on perplexityās browser
You should just email them
So I gave PPLX an image of a politician, it even recognized the party š
but not who it is
Mostly we use reverse image just to find someone
fast
or smt
google just good enough with that
Due to sys prompt and dumb guidelines
there's no point for that being added
You can give a picture of abraham lincoln to chatgpt and itll refuse to tell you who he is
gotta say pplx deciding to lock models behind the $200 USD max plan and not giving pro users even heavily rate limited access to them might be the final straw for me. i depended so heavily on this product for about a year and now i'm close to totally cutting and moving to you.com because they offer claude 4 opus extended thinking (on their regular plan). i don't know if this has totally killed the service for anyone else, but honestly for me it's the first overstep towards "enshittification".
Yup
Use yandex.
But the technology is there, yandex does it, tiny eye does it, google used to be good at it
Give operator from chatgpt access to yandex or jailbreak the model and tell it to find it and itāll find it
Google is becoming shit as a search engine day by day. š
O3 pro and opus is really fucking expensive
This seems all a policy problem not an technical one
Too much censoring.
and yet other providers manage to not charge $200 for them, or rate limit them appropriately...
true
Jailbreak it
Its the easiest of them all to jailbreak
ehhhh Claude rate limits A LOT
You.com teams plan has heavy output limits
And claude limits are shit
Claude is terrible with rates, even as pro user
i'd rather have shit limits than no access at all
Opus is heavy asf with rates š
like how is that even a point against it
You.com web search is shit. If you won't need it for web search then go for it.
Its almost unusable
You can't even vibe code with claude because mid code it will take a break for 6h š
Just get another subscription at this point if you dont think perplexity is for you
the search is the only thing that made older models on pplx usable, but Kagi has greatly improved their offering too. i'd probably split across You and Kagi for it
You.com UI and UX is horrible btw
What's kagi?
Wanted to ask the same š
Prem search engine
Kagi wut?
Ai?
$10 a month non-enshittified traditional search engine. unlimited searches and a web search assistant. it's like the old google
Ngl theyāre interesting. Claude opus 3 was heavily limited on claude for pro but on you.com it was unlimited š
S e a r c h E n g i n e
I'll pass, not paying another subscription š
Theres also ai stuff baked in but eh
PPLX has basically replaced google for me
You cant usr it to code though
Limits the output length
Kagi also does a free trial, would highly recommend. i haven't had to scroll down the search page yet as the answers i need are always top 4 results. saves me so many hours a month
You mean opus 4?
It has both
Opus 3 before was also really limited on claude
I might give the trial a go, ngl PPLX has solved my google enshitification
yeah pplx showed me the light too, but i still wanted a good traditional search engine that was just good that that with none of the oddities that using LLMs to interrogate the web brings along
the duck, how many?
never heard of them too š
I'm falling behind, I used to know all the AI shit
Duck duck go
Duckduckgo
in all honesty the perplexity max subscription could be a response to cloudflare saying they're going to start charging for ai scrapers
oh I know THAT
Lmao
duckduckgo has OK search results but i wouldn't ever consider them stellar. still a better experience than google and great if you have a specific or ambiguous problem that an llm web search with pplx won't accurately capture
Fuck fuck o3
Very mid. But uncensored.
Brave would be good when they'd recognize crypto is over and just drop all the hustle shit
Im taking a break for tonight my brain is too fried for this rn š
They're just being the broker.
brave did a bunch of affiliate link injection too didn't they
that kinda put them on my shitlist
Integration
I just can't brave because of the crypto crap. Else it would be a decent browser, but you can't trust anything related to crypto.
bing is lowkey kinda goated when you work somewhere that uses sharepoint heavily
clippy gets me the best responses
Fr
I feel I need a pure reddit search engine
I already put reddit into every non PPLX search prompt when looking for a solution
Ah didnt read that other word
so you can do like a custom group of sites to search if you have a set of preferenced sites or you can add a specific search engine with criteria and then blast a query over to that
Wut usenet , "Looking for Spiderman the movie" and I get the bin? š
yes
that's not usenet tho I think, just archive being archive š
there's also a usenetarvchives.com result
ngl it looks interesting, but I'm cutting down on subscription not adding them ://
true, the free tier is pretty good too if you're a light user
i think the fundamental thing for them is that their LLM search is good because their search engine is good
Think like a pirate and youāll have access to the sea
oh not for the usenet stuff š I have a ship already
oh they also have llm based translation which is unlimited too (and really good at mandarin)
I just found it an interesting search domain
which imo english to mandarin & vice versa is about as difficult as it gets among living languages
if you don't use it you've lost nothing but your time
which is admittedly limited on this earth, we are but a speck of dust in the never-ending halls
š
Nice
Oh wtf
Dang
Let me try to email them again
Claude code creator left
Also best not to show your email here
Gone
Totally
Billy and karrie stealing a helicopter.
That's not interesting?
We gonna beat radars with clouds - Modi 2025.
At first I thought Rahul Gandhi might be lying in his speech.
But after seeing this.
i mean a real prompt
Abstract representation of limerence.
Guess the subject
Math.
I had differential something in first year of computer science.
Okay. Wait.
Quantum mechanics?
I read that manually btw.
A.P.āÆFrench &āÆE.F.āÆTaylor, An Introduction to Quantum Physics (MIT Introductory Physics Series, W.W.āÆNorton, 1978).
Yes
Chadgpt?
Yes
Make it more biological/anatomical, include heart, brain, glimpses of memories/imaginations.
Am I the only person who alomst never used deep thinking modes?
I aint got time for that shit
Lol
His ocd powers are way more stronger than mine.
Maybe o3 pro is still as good
deep thinking is a gimmick, reading the thought process is fun once or twice but it's so much bla bla
I have never found any ai model as good as Claude when it comes to following instructions properly.
hell nah
Are you apple per chance
wut?
Bro
Lol apple was trying to cancel thinking models
Where's the glimpses of memory/imagination part? Also, no brain here.
Bro I just realized the title is differing
Ooo
I hate apple (writing this form a macbook air)
Im gonna confirm
Lol
I just think deep thinking is too much thinking, too much bla bla š
O3 pro shit confirmed live
I hate life. (Writing this in this life.)
Wiat what
Like on their on device models?
The title from o3p apparently wrong
They were just recently trying to say all thinking models are pointless
Apple intelligence is shit
Oh that i saw
Talking about that one
Its kinda self explanatory tho
If the model was never trained lr exposed to something like that ofc its not going to have any idea
Yeah
Grok 4 is supposed to compensate for that as per claims.
Marketing
Nevermind right
Crack 4 is never coming lol
The funding and compute to achieve it though.
Whats next? New Apple study says consumer rights are shit? š
Elons political party might arrive faster
I'm not believing anything elon posts until i see it
He said it's coming after July 4.
Self driving cars soooooon
Yeah he's never delivered any promises on time
Lol
He's been saying we'll been in mars from 2020
I'd be willing to try Grok if Elon wasn't involved
Fair.
just not believing anything this guy says
Mood but unfortunately it's good in some applications
Grok itself thrashes Elon.
Especially reading twitt data
Lol.
I don't touch X with a 100000ft pole
Lmao fair enough
or any other social media bs
Bro touches kids on reddit.
When jailbroken you can use it like a monster on twitter
Not true! I mostly post about Dota 2
It's unhinged already can't imagine how much crazier jb can be
Fr. I tried doing NSFW roleplay with it.
Smash.
lol
You have to put it in custom instructions though.
mhm mhm mhm someone did their research
Yes.
I didn't like it though.
It's Aave English vibe.
Can't really mimic the talking style of the desired character.
It would be a monster for osint
time to create my own AI ranking site
based on NSFW roleplay
which model is best at it
Good morning fellas š§āāļø
I'm scared
Essentially your own person for testing given how much irl shit it can emulate
And data it has
Maybe that's why they didn't make the reasoning variant available on api.
For me personally training an AI model on Twitter data is like basing a gourmet restaurant on the English Cuisine.
I think even the regular one should do just fine for this
It has its uses
Lol
Many people like the shitshole
Especially like me who have been using since 2012
Twitter social media's like no other could
@timber niche grok might be Elon's controlled opposition.
Facebook Instagram could never
Nah he can always just change the backend when he's unhappy
That's the worst thing
If you're actively using FB and you're under 40 you've failed at life.
But all companies do it for bias anyway
For marketplace
It's good
guys any way to get free api key of llama 3.3 70b ? i have claude api ut it used 10$ in just 9 md files
usin gmodel 3.7 sonnet
any other api key availaible?
You mean above 40?
Altho I think Facebook Messenger is a huge thing in some countries
Use o4 mini next time.
I use facebook to get rid of social media addiction
Try ai studio for now I guess.
Imagine it pulling random pics from twt for each person for a real osint ground
I fell in love with osint because of rainbolt's video
I legit added mine to gemini cli one
i got claude from student one 50$ credits
3 posts in "FML I quit"
It's always ai posts and ai comments under them
Go to reels to find more ai cat videos
you forgot the angry grandmas
Fuck this shit
Rn it's grandmas and aunties simping for comedians
Young ones lol
What do I do in it? I don't know coding.
Idk for fun lol
It's essentially just gemini but can do more on pc
Like write to files and stuff
Does it have web search?
@distant aurora remember when i told you about filling more instruction to perplexity with file upload. Well it seems like o3 is really good at handling that meanwhile gemini fails to load and use all of them
But o3 is nutz it even loads my personal instructions from the preference lab or whatever it's called
I told it about its 2 mistakes. It corrected them in the response and repeated them in the response after that.
Even O3 has it's own problems though.
It will deliberately ignore what I ask for in my case.
my issue with o3 is censorship
bro is too strict like imagine not allowing blood and gore
O3 is terrible for no shot prompting.
Sonnet does it fine.
I was trying to make custom instructions to make ai obey queries properly.
O3 gave me a hard time with it's gymnastics.
yea but
it worked well when i gave him a lot of instruction to handle
it guess its because it has the ability to reason multiple times so it kept reasoning in the instruction files multiple times
What interface is this?
Oh I remember that site
I think I used to use it as a search engine or something and it was entirely free, and now you can't even access that without going through a process first
thats not the issue
the reason why they fall to follow it
is because the instruction is too long
I'm talking about my case. I gave it a simple task to write me a prompt. O4 mini and sonnet 4 thinking did it well.
O3 needed to be pushed a lot.
Even gpt 4.1 did better wrt the effort.
You mean sign up?
I mean before I remember it as being a enter and search thing
Whatās the max amount of tasks you can do?
Wdym?
welcome
I want to report a payment vulnerability to Perplexity Max who can I contact? thanks everyone
please contact support@perplexity.ai
which vulnerability tho
The 0$ bug ?
They fixed it
They call it, the crisis of internal consistence
yes, but want they will do with all account use it ?
For those who don't know, yes there was a bug allowing user to actually upgrade to max for 0$
if u have pro
But they patched it after 5-10 minutes
ye, and i got this today
before u report
so what i can do right now
i wanna test max
i'm crying
Email them or contact support
did u just convert 1 year pro to 1 month max ?
eyy lil bro you are showing your cards info
So it actually
Deducted the money
on ur card
or whoever card it was
Who is @signal hamlet , he is suppoter ?
=))
okay
Me waiting for my pro to be converted to Max
I wouldn't mind trading my 1Y of pro
for Max
as for now
It suits my need
wow arent you showing confidence in the room
wdym ?
" i would be happy to trade my iphone 3G with a new iphone 16" ahh play š
btw i did my ielts i got 6.5
i got cooked in writing and speaking
On perplexity
Cmon it's a good trade ngl
Told ya
Planning on taking Ielts test again
pulls out an iphone 3G
š
Wait you can convert it?
Iāll convert my one year enterprise for that š
Sadly u can't
ye i can, i now i'm finding to supporter to get back my account
It's for personal only
Xfinity is my hope then

dont think they will convert a trial account to max
it's more like convert it to Max
Did they ever reply back
Why didn't they add a credit system
For email
" I'm fast asf bois " " I'm fast asf bois " " I'm fast asf bois "
For live chat
" I'm on my way, 5 more minutes "
relatable
the one you are trying to convert is a pro account with 200$ paid or a promo or trial for 1 year?
Trial ofc
Do u think
i'm that rich ??
Dude
My nitro
is
" TRIAL "

After this nitro end
imma back to using vencord
nah i mean would they give u a free trial of max?
lets see how it goes
yeah i mean promo
i meant the student pro trial of 1 year
they are 2 different things
mb
nah i was asking for opinion, do u think they might give an option to convert edu pro trial to max
lets give it a try, though hard to think they willš
i don't think so
edu is a different level
or platform should i say
but i don't think trading 2Y of perplexity pro
for 2 months
is a wise decision
bcoz my 1 year of edu pro is gonna go to wasteanyway
my cllg would delete my mail account in few months, as per its regulation
Especially when we donāt know what models might release in the next 2 years
O4 or o5?
Never tried that feature.
The point is
For example
Me
Iām batman
Iāll need the max version right now
As itās a crucial tool to help me boost my studying atm
With access to other advanced models
And unlimited access
For pro on the other side
300+ ask a day
Is enough i guess
For a normal use
No one's supporting you on this one, lol.
before gta6
When o3 fails, o3 pro rises
Or opus 4 too exists š

Hopefully this o3 pro jailbreak worksā¦ā¦ā¦..
Oct 1 2027
Till itās delayed
i think at this point we need to be suprised what comes after gta 6
I trust matt and rob
$200/month for o3 pro & opus 4 .. LOL it's robbery š¤£
when perplexity max for enterprise?
company credit card hates this one simple trick
Has it always been like this or has it only happened recently that the Deep Research results seem to be very weak compared to Gemini 2.5 Pro (especially when you prompt the latter to perform a maximum analysis)?
results became week like since 2 - 3 months ago
not long ago the name deep research got changed to research and nothing got fixed...
not that deep anymore huh
The number of times I write a long prompt and then send it, just to get a blank page (refresh) and having lost the prompt... š¤¬
you don loose the prompt?
Yes you do in some situations. I know they've tried to store it locally (probably localStorage) but it still happens at least once a day where this local stored copy is cleared and no thread was created - meaning the prompt was lost.
šØ China Just Cracked RSA Encryption With a Quantum Computer ā The Global Security Wake-Up Call Is Here
Researchers in China have achieved a major quantum milestone: breaking a 22-bit RSA key using a D-Wave quantum annealer. While small, itās the first real-world demonstration of quantum-powered RSA cracking.
š Why this matters:
⢠RSA encryption protects your emails, banking, health records, and national infrastructure
⢠This method didnāt use a gate-based quantum computer ā it used annealing, proving multiple quantum paths to decryption
⢠Today itās 22 bits. Tomorrow? We may not be ready for whatās comingā¦
š”ļø Experts now warn of āharvest now, decrypt laterā attacks ā where encrypted data is stored now, only to be cracked once quantum is strong enough.
š Source & Credit:
ā āChina Breaks RSA Encryption With a Quantum Computerā
Shanghai University & D-Wave Systems
š Follow
to stay ahead of every leap in quantum computing, cybā¦
GG
i better start praising the red ones




