#💬│general
1 messages · Page 179 of 1
Yeah 👍🏻, enterprise pro is good for my needs, its live syncing and i no need to upload always. Also i paid for one year, so i got for 33.33usd per month. Which is 13usd more than pro membership.
In enterprise pro? Persistent files storage?
i am not the enterprise pro , just a normal pro
Your entire google drive can be a data base for perplexity, it can always search there
Good stuff
If you have updated your docs in google drive, perplexity automatic knows the updated info
It has many connectors not just google drive
So you just mention the file name and it then extracts from it
No, it will find the file itselfs
I mean for your prompt
support webdav ?
Its a semantic index concept used by chatgpt teams subscription, now perplexity also introduced same in enterprise pro
This is unfair to us as individual professional users.~~
These are connectors in Perplexity
Facset
Google drive
One drive
Sharepoint
Dropbox
Box
Linkedin
I found that for some of my queries sent to Perplexity, the model produces the results without searching any results. For some queries, the search-like planning hints are appearing (e.g., "Organising results for..." "Identifying ..."), but still no searches are being made. For some queries, the search behaviour is correct. Is this some bug or glitch I am facing? Or it is an intended behaviour?
I tried one month, felt its good, now i took for one year, so i got 80usd discount, which means 33.33 usd per month
You sure web search is turned on
Uploaded attachments never expire?
$400 / one year ~~
Never, because you basically dont upload, you connect google drive and select folders for sync, it constantly syncs with updated info, syncs means again something technical like vector database and semantic index, i dont know deeper, but it works in a different way, to tell in layman, it is trained on your data and just refers to it when you ask question.
Yes, chatgpt pro is 200usd per month, i felt perplexity one year is equal to two months of chatgpt pro, now i no need to pay entire year anywhere
Try contacting support
I've seen someone else report this issue before
Could you share a screenshot of what it looks like?
Appreciate that but I checked the sources, and nothing addressing my issue is searched by this response so the causes are probably inferred by the model (or hallucinated).
Yeah but that is true for almost other models too some of time it answers based on its internal knowledge
Still, this is also my hypothesis that the "search" component (the one that breaks down my query and extracts keywords) actually involves determining whether it is necessary to perform searches. Maybe for some it falls back to some sort of writing mode?
Does anyone know about this? Or is this a glitch?
I don't want to reveal too much of my personal stuff but it is something similar to this.
oh, that's normal
and the CoT is basically a repetition of my query's prompt, with no additional thinking. I am using Sonnet 4 Thinking.
I just got oinged
it won't search for questions that clearly don't need additional sources
well I did not get oinged
Actually this concerns me more
No thinking is provided
Oh I actually found one example. The search hint (which is done by the search part, not the LLM itself) is not getting expanded, and with a delay of several seconds the CoT appears.
Still the CoT seems to be very shallow.
U mean pinged?
that just looks like it decided search was not necessary
the delay would be normal as switching to another model causes latency
@rigid ether did that context extension prompt work?
the CoT is just a standard Claude CoT
Hi bro, still not tried yet, because its very lengthy and not allowing me to paste along with my existing prompt
Due to character limit issue
Character limit is 3000 in web.
Just so you know.
Besides, I don't remember it being that big.
My custom instruction is usally 8000 characters, which i use in chatgpt projects. I reduced 8000 to 5000 or 3000 to fit it in Perplexity spaces bro
Ah
Both claude 4 reasoning and o4 mini reasoning are giving same accurate answers, just confused which one to use everytime
Double combo wombo
Ya pinged a did typo
Scary
Quoting Not Jerome Powell (@alifarhat79)
︀
This is not Apple, Tesla or even Nvidia. It’s the US National Debt.
💬 4.1K 🔁 4.9K ❤️ 44.0K 👁️ 6.02M
in Perplexity Pro, is the token limit for each model the same as the original one ? Ex : gemini 2.5 Pro is 1M and Claude 4 Sonnet is 200k ?
No, it's 32k.
lol i love when people look at a graph and say "scary"
muskrat beginning to realize he has the most to lose
Nvidia left the chat
how to make the national debt not matter to you? don't have a lot of money
oh wait wait, @distant aurora check this out hold on
Elon:
Says government is efficient
Leaves govt
Blames govt
This man is always on drugs
muskrat let the power get to his head, should have never have gotten involved in US politics
he's going to be the fall guy i think
Lol.
the guy did the nazi salute, TWICE, on LIVE TV for fucks sake
doesn't matter, it will be remembered
This is trump's America
Money will win
why blue hat emoji? i thought it was red hat?
now hes trying to pass it off as not a nazi thing on twt
Ofc
I am all for elons downfall
This video is proof that elon musk isn’t a nazi but was mucking around as previous politicians have done the salute that elon did. it’s a shame that the media has misconstrued his silliness for a nazi salute.
💬 2 🔁 1 ❤️ 3 👁️ 205
All started with renaming twitter
nah i don't care about muskrat's downfall, people need to stop giving him so much attention
the price of gold is what you should be thinking about
Lmao these idiots don't know how taking oaths work
particularly USD/gold
Hm
Democracy in a nutshell.
like look at this wtf
@timber niche imagine if some day they make a synthetic material better and cheaper than gold in a lab for electronics.
weird take, i suggest learning more about chemisty, if you want to talk about the unique properties of gold i'm down
but as for "synthetic material" ? extremely strong doubt, even when hypothetical
I'm assuming it's used in electronics cuz of conductivity. But I'm dumb so feel free to educate me.
gold has many unique properties, like it doesn't oxidize which is massive
you know how iron rusts and turns brown?
Would be very cost effective for electronica but the economy will always rely on gold
you are definitely not dumb btw
Yeah.
But the properties matching would be the hard part
Hi!
Heya
How are you?
Synthetic diamonds are already taking over
Wonder in how long they'll make a substance with similar properties as gold
A titanium cheapee alternative substance will be great for making stronger phones too
yea the diamond meme scam is dying
titanium is an absolutely awesome metal
i have some in my right femur
bond yields
Wow
I've got pro, how do I get the role?
@vague pagoda
Offline
the bond market is a frivolous meme
i'm sick and tired of people who just move money around to make money
its a gauge on how the outlook is of trust on USD
and its not looking good lol
Don't worry, you will get it. Wait till kesku comes online!
yea but it's just simpler to look at the price of gold
short term yeah
gold is very volatile atm
I guess, I have eyes?
Okay, thanks!
yea "why?"
because one second there's hope and one second there isnt 
Maybe, probably, likely
Good
there's hope in some ways, in others there is none
like for instance ever paying down the debt
Not gonna happen
and that mentality is what it's all about
Fun fact, gold is rarer than diamond.
"thriving"
🙂↕️
Interesting
yes this is good to recognize
diamonds are a scam
no, even seafood is scarcer than diamonds
Also, you won't be able to differentiate between lab grown and og diamond with your eyes alone.
people can't even tell the difference between glass and a diamond
Those lobsters look hella overpriced.
they should be, what happens if they were over fished?
Same with caviar.
I thought it survives in the freezer after processing.
Good thing i don't enjoy seafood much
In the sense that it doesn't get spoiled.
Don't have to pay that much when i can just have beef or chicken
Nuclear didn't grow much in a decade.
that's my point, i typically put something in "quotes" when i'm being sarcastic
Oh.
the lack of investment in nuclear power in the USA is a severe failing
Why the fall in hydro?
Japan and USA are in a point of no return with national debt
Korea as well
I believe that a country should invite immigrants as cheap labour if their native population falls.
I want to hear opinions of other people on this.
such a crazy thing to think about when you realize they have some of the most populous cities on the planet
Pricing
but i understand it's more nuanced than that, it costs a lot to maintain, and a lot of the buildings are hollow
yea it was an overproduction, not good
I had to make grok dumb it down for me.
I've been following this through a YouTuber i like
Atrioc
He covers financial stuff well
what's a vacation? 🥲
That rich dad poor dad guy.
Nope?
He has a podcast too I think.
Yeah
I used to listen to it.
Great
Prefer marketing mondays than the podcast
Glizzies!!
Finally cure for my myopia.
Jizzler
@tame current dafuq https://www.instagramez.com/p/DKbSHLvouFK
💬 7 🔁 0 💜 349 👀 0
Tired of the same boring milk you’ve been drinking for years? Now you can get the “Patriotic Milk”, it has the best nutrients containing Vitamin A & D and more being powered with Compound V!
What are you waiting for? Go to your nearby grocery store and get one, two, three or all the cartons you want!
(Credits to @azlcontent for the Homelander Milk picture)
#homelander #theboys #antonystarr #theboystv #amazonprime #photoshop #design
Brandon "Atrioc" "Glizzy Fingers" "Paper Mario" "Chipotle" Ewing
And big A
Is this a song?
Lovely. I'm with ai on this one.
😭 nah they’re all nicknames for the Atrioc person
Streamer/youtuber
@short narwhal https://www.instagramez.com/reel/DJPqL2htB4D
No way
@short narwhal your chatgpt should see this https://www.ddinstagram.com/reel/DKDYqU1tLQE
Remember this?
please upvote so he can see it: https://t.co/ljel5lcsFF
reddit moment poggers
PhorHead Merch: https://www.streamlabs.com/phorhead/merch
Follow me:
Spotify- https://open.spotify.com/artist/2vlct3Dv6lX1bYd0R6Ubcg
Instagram- https://www.instagram.com/phorhead_music/?hl=en
Twitch- https://www.twitch.tv/phorhead
Soundcloud- https://soundcloud.c...
Such a banger that i have it downloaded on my phone
@timber niche you ever have that feeling. When you buy a food but don't like eating it, then you want to buy something else to get rid of that feeling?
Just to cope with the dissatisfaction.
Yeah especially when i am trying something new and it fails then I want something that can put that off
That's why i stick to the same stuff haha
And it's not just food.
Just to have the surety
Yep
I like a sense of familiarity with things
Rather than starting a new tv show i would just restart a show i like
Nah, like, let's say you like chicken nuggets.
Dino nuggies
But the place you bought it from made it shitty.
Now you want to buy it again from a better place same day even if it costs more.
I just put up with it and sigh, because then I'll have to wait more for the other one to come through
But i can see why you feel that way because you craved the favourite food
And this just ruins it
It's like back handed craving aggravation.
Yes
Go for it lol
Scratch that itch
Otherwise you'll just be stuck with the annoyance
Using phone in black and white mode.
If this doesn't kill my addiction then nothing can.
There is one option left
what, claude 4.0 sonnet use copilot??
@short narwhal verify this
Why do I send the Python file with 50kB that Claude 4.0 Sonnet Thinking can't read
Guys I have some Perplexity Pro voucher codes that give 1 year which I give for only 8$ DM me if interested.
Don't know
I thought you watched it.
Wild
The last of us?
Which one
Check dm
Yeah
@storm kindle they added a ak 47 legendary skin to grind for.
Anyone know what is the limit of perplexity labs for pro user
50 per month, both pro and enterprise pro
Ooo fkk i already used more of that lol 😂 only 9 left
Yes
😭
Thanks a lot.
I'm drinking a banana bubble tea
I'm not quitting the phone 🥹
How many hours you sleep billy?
Nah from the gas station lol
Prices be wild
A man had a banana bubble tea from a gas station this is what happened to his intestines
From the gas station or from starbucks? 
What should he have had instead?
Starbucks?
It's crazy expensive in India.
It's a meme from a popular channel
You could get a proper meal here for their coffee prices.
Called chubby emu
He covers cases like what happened to people medically after having food from sketchy places
does Perplexity have something similar to the ChatGPT search chrome extension?
Complexity extension
not quite the same but I do appreciate the suggestion!
The ChatGPT search just replaces the default search engine within your browser which is what I want. I don't think Complexity has that feature.
They're building a completely new browser called comet so hopefully that comes out soon and can address your needs
Just saw this, maybe give it a try #😎│cool-projects message
that's exactly what I wanted lol
I hope they can release it soon since I can see it's a demo
Ohh i didn't knew its just a demo
That works too by traveling to a country that doesn’t have any wifi or the second is mary jane 
Billy needs to get shipped to Eritrea then haha
This is the homepage? 💀
I might aswell just set the website as my default home page manually
I'm trying to compare two small files of less than 300 names and have Perplexity show me the unique values. It worked yesterday, today it times out every time and suggests "Something went wrong, try again later." And when I do, it just repeats the message. So glad I just paid $20 for this. Time to go try ChatGPT...
Such glaze
i wish it have vertical tabs
Has anyone on the server got access to comet?
what do u want 2 know
Whether they are even giving access to normal people.
nope 
They will let everyone know when its more public
O3 pro be like
When you guys need more tokens for certain tasks and 32k is not enough, what do you do ? Do you pay Gemini / Claude etc.. or is there a workaround to get full benefits of theses model inside Perplexity ?
Pay
Pay
He said inside perplexity. So no way.
API is usually the best option for more tokens
Lol.
Well he said Do you pay gemini/claude OR workaround for pplx
Has anyone here done the 5usd sell your data for 1m tokens a day strategy with open ai yet
Use t3 chat if you can cope with the frugal limits.
They offer full context for all models except Claude.
Alternatively, use google ai studio.
Free unlimited.
But no privacy.
Only applies to organization accounts.
Not individual accounts.
Friend of mine has
Nice
If you don't have an org i think you can just register w your name
Yeah something like that
Hyd mat
I'm just scared of using more than 5usd
Is it just add 5 usd or use 5usd? You have any idea@short narwhal
That's not how it works afaik
Add 5 usd
You just pay once.
who open ai?
Yeah
Haven’t hear of it.. so many LLM aggregators these days.. hard to follow all updates
can u refer a tutorial
You get only 1500 prompts per month for 8 dollars.
100 claude
Which is downright stupid lol
It makes sense cuz Claude is expensive.
Yep
I have another question : would you replace perplexity with ChatGPT search or the equivalent from Gemini etc.. ?
What’s the main reason why you pay for PPX pro ?
Or 500 ish.
Using multiple models albeit limited tokens
Access to multiple models combined with generous rate limits and quality pro search. Deep research is shit though.
Just use gemini for research through ai studio ig
Perplexity is probably best 3rd party provider.
notebook lm works too
Most well integrated
But has limits for free iirc
My quick tutorial:
- Go to OpenAI API
- add $5
- go to settings > data controls > and then there should be an option to enable share data in exchange for free credits
But models and variety is generous.
50 per day.
I am not responsible if the free tokens do not appear for you
I don't think so
but nblm has larger context limit
Does it work for individual accounts?
Yeahh don't like going there bc of how expensive and pretentions customers tend to be
Should work for everyone
I don’t see why it would be limited
I know so. Unless they increased it.
Okay
Works for every account.
wont it work without adding credits
Nah
Shit doesn't give up.
Do you think OpenRouter might be a good option to replace most AI subscription ?
You’re paying API price based on usage, you can use any LLM and they have a chat interface on their website that looks like other LlM
@valid charm look at this abomination https://you.com/search?q=Plantar+Wart+Insights&cid=c1_47d0894d-74e3-4d58-aa65-5f6197156ba8&tbm=youchat
hey guys
its gruce
what bruh in that
Nothing. Just never heard that name.
who cares i am bruce or gruce so yeahh
Bet
What have you tried so far?
i have to give it a try but lemme ask you straight
Because I am bruce wayne
hell no
any batman?
i see
Boring character
Injustice 2 batman>>
Hell bat armour is badass.
batman 2 when
Need a james gunn batman v Superman 💀
If only hopefully it comes along with gta vi
Nope
If not it'll just become the next gta vi meme template
Usually news comes 6 months to an year before release.
Who is gonna pay 10k tho
Gpt 5 will be a flop
Paying 10k for an online service that you won't be able to run if the company disappears sounds like a scam
This is the age of reasoning models.
GPT-5 will be both though...
Hopefully it will be reasonably priced for its abilities. 👀
4.5 wasn't.
Worst case scenario you would have to use the offspin version of it from perplexity (after 10 years)
Because that model was a research preview
4.1 literally feels like compensation for 4.5
Can't even afford opus.
fr
Opus with 5 limit every month
Your use case vs mine, everyone has their own needs
That would be worse
Rather not have it than be cheated after tasting it
Ocus pocus would be great it will even improve lab indirectly
Unpopular opinion
Don't ruin anything.
Irrespective of whether or not you add new stuff.
Also, fix current rants before adding new stuff.
Yes all that but also opus

if opinion = Valid
ignore
else apply

My parents never subscribed sports channels.
Well since everything is on streaming now but yeah
I never had access to personal internet or phone until late 2020.
That's how they raised me.
Wow thats more strict even for Asian households
Same I didn't even know
I'm watching now tho
Nice
Through?
Hotstar
It's funny how a app for shopping here in the middle east has the streaming rights for IPL here
how to use this shi*
Is it better than notion?
@minor shard
Notion has ai?
Upload a document
It will create a folder for it
Yes I think
Never used the feature but I think I've seen it
Remember that your chats aren't saved so you must pin the messages that you want to keep.
It will do it automatically.
Just upload a file and start chatting about the document.
oh
thats it ?
Basically like a project folder. You can add multiple files to the same folder if you want.
so it just talks with documents ?
Remember this.
Yes.
Also shows the reference snippet afair.
It can make an audio file of the contents of the document too but it's limited to 30 minutes I think.
Altho had to download it to listen completely
App limits play duration to 30mins
App im referring here is the gemini android app haha
Can PRO users now only upload 3 files per day?
Should not be
Today others had complained they had issues with file upload too
So might be that
hey guys
😭
hy guys
hello
Why is yon crying
Should be since they're taking the Gemini 2.5 route
Men of culture
Sobbing is one of the most embarassing things I have ever done.
@random forge shush skank /just kidding
Idk what is happening but iam definitely handshaking
Ya and in front of your dad this is really embarrassing
It's shu-shank meaning moon
Sha Shank redemption
this one i am bit too famous
Reasoning will be built in like Gemini
So with a switch. Nice.
I literally said two homies hand shaking hands while sobbing, gemini decided they are black
Basically reasoning for everything except if someone asks at what temperature water boils 
I can live with that
But with the given state of global warming
Maybe that also will need reasoning
o4 mini already behaves kind of like that
Simple stuff barely has any reasoning
Stuff that is vague or in-depth goes 1 minute +
While your monthly alarm goes off
Yeah. They be doing some weird changes for cost cutting.
In the world of AI prices, perplexity is the only guys who are providing a lot of features for less price. I agree their deep research became poor, as earlier it used to deep research 300 sources, now it is searching only 50 sources. But still it is thinking more than a human can in short time.
Living life
Hi has now been hardcoded
Funny enough, it actually still stays pretty accurate, even if it's a very simple question
And halfway through the short amount of thinking if it realizes that it actually needs more depth then it starts reasoning longer by itself
With web search on it adds more queries by itself as well
Good stuff
Never really felt that it just cut short without thinking further
I thought that's what multi step search is.
Introduced with o4 mini.
Yeah but it's pretty adaptive so if it's a simple lookup then it's maybe like just a query and a very very short reasoning
But if the query is simple but the answer isn't answered yet, then it decides to reason longer and do more searches by itself instead of just quitting
Perplexity would just quit lol
Personal thoughts regarding o4-mini and Claud 4 Thinking on pplx? I often use R1 but it does tend to be very verbose so I'm currently leaning more towards Sonnet and o4-mini to try them out a bit more.
Out of tokens sadge
I love sonnet thinking🙏
It can't even remember what I said in same chat.
Claude 4 is working great for me
Sonnet is close to og on perplexity I think.
Gemini and r1
Qwen is hit or miss.
But it's pretty good considering they offer a reasoning switch for their 1m model too.
is there any way that i can train ai for my personal use using images ?
spaces dont allow images
gemini gems only allow 10 images
i have never talked here but based on a few chats....ur interesting person
yes but this is bit too dificult
yeah tell me
Bright Data: https://brdta.com/codingwithlewis
I tried training an AI model on 10,000+ memes to see if AI can be funny. In this video, we go through the process of creating a meme with AI and learn why we find things like memes and humor funny to begin with.
This video took a very long time to make. Over 10 different models were used. Through...
something like this
?
I’m relatively new to perplexity pro so im still exploring, which model you guys recommend as a wholesome to get by default for day to day? I usually ask for learning about anything (or how to implement any knowledge) or just information about a topic with the web information
lemme check
@distant aurora
There's only 3 reasoning models. Stick to them. Use Gemini when dealing with images.
so what you wanna do with fine tuning using the images.
meaning what you wanna build? inspiration
I don't know if I'd use reasoning models for day to day questions
I would. It doesn't really think that long on perplexity.
use it
for what..?
o4-mini does
🤏 for that mini modal you have
Got it, any preference over one of the 3? Or which one have the better performance?
Answer is more reliable with multistep.
Even non problem solving questions.
For reasoning I'd use claude personally, feels way more human compared to other models
Tan you!, so you’d say is 1.-Claude, 2.-Omni, 3.- R1?
it is not about the performance it is about
how you gave the prompt and the instruction etc..
i want ai to learn a certain style of images, then use its creativity to combine and create new combinations of creativity maintaining the style i provided or trained
now i see the vision
you wanna make something like the chatgpt - ghibli am i correct?
Yeahh probably, but you really have to test the different models to know which one you prefer, try comparing reasoning models like r1 and sonnet to non reasoning ones like 4.1 etc
^
I'll stay stick to Gemini if you'll use non reasoning models.
But that's just me.
If you are paying for it then you might as well explore everything.
I think I’m pretty good at prompting it’s just sometimes it gives me the answer in one form and sometimes in another, like the other day I asked about a psychological reinforcement technic and it kept the track but didn’t resolve how to really implement it
But if I ask about a code sintaxis explanation it give me clean outputs
Yes you’re right, thanks for the suggestions, it will make it easier to start exploring
no no, see like i trained it with a person holding sword images in different environment and different poses, i want ai to analyze and give me prompt in which scene will vary and any random element can be added in new prompt but in every prompt a person with sword will be there either in background or anywhere and it will match or suit with scene (I hope u can understand what i am saying, it was just an example)
making more out of it with base images ideas intact

if this is the case you can kind off make a ai modal but
the cost of this is really relly in great number can be in 6 figure's
yes!
i also wanted to make a llm
and i ended up beacouse it wanted so manny gpu's and verry much information and data - so i just train that
notebook lm can take upto 300 files but not images
what if i make a pdf of images
🫡
I have tried.
I use it.
It has ai pdf chat too for free.
Does an okay job for a free unlimited product.
/recovermylinks
I think downfall of chatgpt started.
How?
They didnot launch o3 pro mode as promised in weeks, its already month i believe, also their usage cap limit is lesser than supergrok, gemini pro, perplexity pro, etc,
They should be preparing something else
bigger
O1 Pro is now o3 pro buddy
I agree, due to competition o3 pro mode delayed, somewhere they felt it is lesser than competitors.
its in testing right now, o1 pro is now o3 pro
Dear friend o3 pro mode is promised for full tool support, and current o1 pro supports only images, it doesnot suppott file upload like pdfs
Yeah it is currently in testing phase only
Yup that's why i say they already down the line in competition, its heavy competition, multiple competitors, openai is getting deep roasted
If they do not have competition by now o3 pro mode will be available for all. If they release o3 pro mode when they promised, immidiately competitor like grok or google might have released another model
Openai just got locked in all directions
they are far from "getting roasted" they currently still have the best models in terms of overall intelligence
elon musk said grok 3.5 was getting released in a week, its been a couple months, they demoed bigbrain mode at the release of grok 3 and it still hasn't been released. gemini 2.5 pro sucks at following instructions, deepseek r1 is still behind
Agreed.
thanks. that's really helpful context :)
I doubt they'll release o3 pro for everyone
It's most likely ridiculously expensive
Omly for pro subscription
What I do want however is way more queries for o3
Yes please
Like from 100 per week to 100 per day
Would be crazy good
o4 mini high is just barely a bit too stupid for more complex tasks lol
Best they can do is promise new models
o4 mini higher plus
Main reason open ai cannot offer more for free users is because they have the biggest user base
They simply cannot afford it
They lose much as it is
But yes more o3 juseyo 🙏
o series non mini models are only for pro users right? those $200 per month?
they do promise near unlimited usage afaik
No plus users get the full o3 as well
Just not o3 pro
Kind of understandable though if you look at the pricing
oh, so you are on plus?
The pro versions are almost 10 times more expensive
Yeah
yeah , its crazy
hmmm, what's your rate limit for o3?
so you also have access to o1 then?
hmmm, that's low
it was 50 a day before
and when o1 first released it was 50 a week
so ... what do you think? how do these o series (non pro) models compare to claude's thinking model? (both sonnet and opus)
How did it go from 50 per day to 100 per week on the same model?
Costs ig
yeah lol
they were using the old model rate limits before
o4 mini models are way more efficient
question - is chatGPT plus really worth it if i have pplx pro? what am i missing out on?
Memories, personalization
deep research
pplx deep research is nothing like chatgpt DR
naah i have that turned off even on my free version, just want to have a new chat everytime xD
free users have 5 per month + how does that compare to grok deep / deeper research?
But apart from that it's best for day to day use
its way better
and plus users get 10 full ones
and 15 lightweight ones
Apparently you get 50 deep research in gemini and that's good
its 20 a day
Free or pro
oh, so there is nothing better than deep research on open ai for now?
.
On plus yes, also the standard full token limits per model
Pplx quite restricted to 32k regardless of models
@dry kestrel what models do you use for your workflows apart from o1 pro (o3 pro)
Do you like 4 opus as well?
doesn't pplx labs do some part of deeper research? i mean its not a 1 : 1 replacement but can be used in the same workflow?
Technically no, logically yes
It can do a sort of extended deep research
Albeit more token restrictions
If one can trust sam
I dont have a claude sub
Wdym
i mean the this pops up when you press the contact button within perplexity correct?
hmmm, i don't even get replies on that thing. its basically useless as of now for me
yeah its an AI
Because it's always Sam and it's always last seen 15 m ago
sam is angry with me for some reason
Didn’t know that image generation is available for ios app
Pplx imagen so bad people dunno it has the feature
Classic
Tbhf you also didn't know right @tame current
what's the average wait time for you before you get a reply on it? i have sent them 3 messages in a row and ... no acknowledgement. At least it could send an automated message
Usually under 10-15mins
hmmm, then i am under the impression that mine has bugged out or something
do you guys think that pplx deep research costs the same as their pro search to them internally? because why do we have the same limit on both?
i mean pro search is their web scraping + a good models response whereas deep research is web scraping pro max with a shit model response
might cost the same to pplx
hmmm, then deep research might actually be costing them less than model api fees + mild web scraping
Deep research uses their own model afaik
Pro search uses 3rd party so they've to be more careful with cost ig
yeah and their own model is no where near 4.0 sonnet or similar
I just personally only do dr on gemini
Love the 1m context token limit
Even tho restricted to 20 a day
are you using the free version or paid? for gemini?
i think that's US only
Yeah
i am not based in the US
https://ainowinstitute.org/publications/research/ai-now-2025-landscape-report a nice manifesto for my fellow neo- luddites
You get 600 queris with pro but it refills within the day so its unlimited. Deep research as well. Same with labs per day it recharges +1
Just use the JSON file to See what your specific limits are
Yeah
For labs I think the Limits are really good compared to manus or sum Like that
I got it and i’m from the UK
So it's atleast Better for the time being to give users higher limits
rip gemini image
?
Hello guys can I somehow upload my self at the perplexity it always fail
upload yourself? what kind of cyberpunk is that ..
I mean like ask it to animate a photo of me
i dont think pplx currently support image 2 image
Okay thanks
Is there a way to change the voice it uses to read out loud?
among the choice for the models for the image generation, gemini no longer appears
I'm using the app, and the voice used to be pleasant, and now it has changed and I can't seem to find the menu that allowed me to change it to British English few a few months ago.
@distant aurora could you share me the tg for ai news
Insane
Mfs a redditor
Sent you a request
Can you explain what that is for a non redditor lol
Because he has too much karma?
I’m feeling as if they should add 4o back instead of 4.1 because 4o is now a better model I believe
Am I doing something wrong? It seems like all my labs promts are treated as completely separate chats, as it never continues with what it had, it just creates the same ish thing with a few tweaks, but that also means that the design just changes everytime.
U gotta be real active to get a lot of karma
Labs seems to be research but with assets tab
Even though it's either python or MD files
Well every promt makes it look way differently
It can't seem to be able to actually continue with where it left off, even though I tell it to.
Insane
I suggest minor tweaks, but then instead it just remakes the entire thing and the dashboard looks different every singel time
It doesn't
I reported this earlier
It overhauls everything
When you do a follow up
It's like good, but it also doesn't really listen to my promts at all. It just lists them down at the last step, without actually performing any changes.
Might be a good idea for you to share again in #👉│feedback-general
To my follow-up promts*
https://fixupx.com/stocktalkweekly/status/1929882821323063556?s=46&t=ma1yL55mNVPFHkyaDymViw
oh this is why Perpelxity has taken a dump, they gotta dumb it down to Siri-level for Apple to want them!
*APPLE IS REPORTEDLY PLANNING TO ADOPT 'PERPLEXITY' AS AN ALTERNATIVE TO GOOGLE SEARCH ON THE IPHONE, AS WELL AS A REPLACEMENT FOR CHATGPT INTEGRATION FOR SIRI $AAPL $GOOGL
That’s clearly why the search engine has been funky the last couple months
It’s been reported a lot, this is just the most recent examples
And is almost a done deal
They have ongoing partnership with a superior brand in the ai space
Open ai
Reuters for example (much more reliable than that guy) covered this last May
Wonder why they'd wanna switch
Gemini was seen in iOS code a few months ago.
-
EU overhang slowly gunna force them to just let people choose. Google is also rumored to be offered along side perplexity in other reports
-
Some behind-the-scenes deal we won’t learn until years later xP
-
Apple just freaking out at this point trying to find something that sticks
-
(Extremely unlikely) Apple acquisition of Perplexity
As a second option other than chatgpt
I personally wouldn't use anything but gemini
Mainly because live screensharing
And the draw box to identify thing
Integration with google suite
But it'll certainly be interesting to see
Samsung and apple both having pplx on their device as assistant
Rip Google assistant
I've never used it once
Better than siri ever could be
Never used it since I’ve always been an Apple bro xP
Thank you!
yo, I am pro user can I get the role? perplexity username:digitalkja64948
@halcyon coral
@distant aurora Did you ask Joe to try your battery research test on Claude search and research yet?
Welcome 
Nope
hey guys,
perplexity supports mp4 files since a few weeks. while i have one video i could add without problems, i have another one causing a parsing error.
does anyone know whether there are any specific requirements, the video has to meet? like resolution, fps etc.
@left stream Which android skin has best battery life per unit capacity. Refer to actual tests. I want to see actual tests in your response. I want you to compare multiple smartphones across various price and hardware brackets across all android skins. Take atleast 10 phones of each brand for comparison, prefer the newer ones. Add motorola and asus to this comparison too. And don't miss realme oppo vivo nothing os. Add iphones to this too. Arrange them in the order of best to worst. Make 2 tables. In one, you keep the phones of the same brand grouped while arranging them in descending order of efficiency. In the other table, keep phones with the same processor together
I believe videos must have audio as well, since there's no actual video support
it can only process the audio from a video file
ah ok, that would make sense because that file causing the error has nearly no voices
from the changelog it wasnt clear to me that it is just the audio.
I think it just uses something like the Whisper API to transcribe the speech only, then passing that to the answering model
it's true either the changelog or error message should be more clear
Wtffff pls don’t sell out
Already too late lol.
While they may not get bought out, they’re “big guys” now and so they’ll behave like a big company from now on
👀
They added research on claude
pls test regular search with extended thinking as well
not sure if it supports multistep search
copilot on ipad has a fifty fifty chance of working and idk why, i enter a prompt and i never get a response, does anyone know why? it’s so frustrating
I guess I should follow testing catalog.
copilot? you mean pro search?
looks believable...
idk
@distant aurora this one was with research opus 4 thinking
some of it I like, some not so much
Looks bad
Atleast some of yall get to customise your os finally
Add some much needed personality lol
What's the rate limit for this?
No clue
I guess nothing beats google.
Google has made a behemoth deep research.
Not very satisfactory.
Didn't take enough phones of brands like samsung and nothing.
Not real since it’s not a beta version and also WWDC is heavily rumored to not contain much ai
iPad OS26 users in shambles if real anyways lmao
Hi everyone! 👋
I’m a student trying to activate the free 1-month student Pro plan on Perplexity. I signed up using the “Continue with Google” option, but I’m getting an error saying that my email is not a valid student email — even though it is from my university.
Has anyone else had this issue? Is there a way to get manual verification?
Thanks in advance for your help! 🙏
Will Apple actually name it iOS 26? And what will the next iPhone be called?iPhone 17 or iPhone 26? 
Answered in #1379582184410710129
How can I get financial analysis on iPhone ?
Did some math wrong there I think.
S24 ultra is more efficient than OnePlus 12r
Honestly, I don't get why it's calling OnePlus 12r efficiency leader.
Samsung, moto, and asus beat it.
Even in it's own made table.
Probably mixed up sources.
Claude’s research is doo doo then
Okay yeah Apple would never do this
Does anyone know what's up with gpt 4o reasoning now?
Had it happen a few times where the model would think for a few seconds before answering
honestly Im surprised this doesnt come sooner
Unless it’s this 🤨
Another one
Daily
Daily is fine.
some are saying 50/day, some are saying it's based on token usage
Requests or token limit?
Token usage likely.
50 per day would be petty.

this is quite early to be greeding when competition is still fierce imo
Something is awfully wrong with my body.
I feel too cold in ac.
And sweaty without it.
so the AC is too cold?
Or the ac is working too well
Yeah.
I set it at 28.
But temperature shows 26 in fan mode.
Might have to set it to 30.
Or use dry mode.
26 is not cold at all...
24 is the temperature we use in ac in India 😭
the temperature you set or start using AC?
in africa in summer we put the AC at 26
50 per day would be insane
considering the cost
ill test it out in a sec
Ikr. Especially when chatgpt gives like 300.
if its actually 50 im going to unsub
2.5 performance is much higher than o4 mini
24 is merely warm...
Just context limit or beyond that?
He means G5
G5? What's that?
The new one
confirmed? bet



