#💬│general
1 messages · Page 319 of 1
Addicted*
Having a preference.
I used my language term sorry
how can one get addicted by llm?🤣
Yeah I'm addicted to GPT models, they're so good that I prefer them to double check on many problems that I solved
From the ones in pplx. When grok 3 and deepeek were available i liked gemini and sonnet.
Many are
Addiction is a relative term.
Whenever I have a question it's OpenAI who knows first
then this is not addiction, addiction is something else
They somehow manage to stay ahead of the competition. Gpt 4.5 was the only disappointment afair.
Dependency
Can be called addiction.
I personally have tested couple of models including Gemini, Perplexity in house model and Claude
I find that they don't even compare to GPTs day to day conversation skills, It genuinely feels like a human
Any opinions?
Yeah that's years of research of development in work
When someone feels alone after getting their llm out of memory or credit that's called addiction imo it doesn't has to be for a weird reason
even gpt free version is better than any paid version
i have gemini pro, but it is not good at conversational skill
While other models seem to just answer query, GPT seems to actually understand what I mean and reflect on a real solution or response rather the the generic feeling ones which other models like Gemini givw
Claude is pretty good imo. Better than open ai models in many ways. The issue is the brutal pricing and rate limits if using natively. The search isn't anything standing out unlike perplexity or chatgpt. Also, it talks too less in areas where I expect proactive guidance.
Open ai models were pretty good in proactive guidance with unfamiliar subjects imo.
what i like most about gemini is its deep research, its the best available right now
Claude on the other hand seems to be acting like it wants to withhold information.
claude is the best in perplexity
Fr.
Grok 4 is underrated.
Well I personally feel like GPT excells in Conversational skills and deep understanding and solving daily questions, It seems like a human who's giving immense effort into solving your question
for most of my work, sonnet is enough in pplx, never switched to other model, but will try
Yeah, open ai models clearly have stronger reasoning and knowledge base but tend to be stubborn when pushed to think in unconventional ways.
I wonder if it's an openAI limitation or they know that the token cost would go up if they try to really push past claude
Does that apply to GPT 5 and 5 Pro?
@gloomy cipher @green crystal bro istg, everyone was asking perplexity to add grok 4 when it was new. Now I don't see anyone even talking about it.
Open ai models in general. The reasoning ones. O3, o4 mini , O3 mini. Like they even refuse to give the lyrics of song.
I haven't used gpt 5 much.
@gloomy cipher what we need is an open source ai model that bypasses ai detectors.
Honestly, even closed source would be fine.
Do you guys feel that models like Claude and GPT are stronger when reached through their own site instead of using them in perplexity or any other ai conglomerate if that's the right word
Didn't knew that
Grok quickly lost it's spotlight 
Because they are
Fr.
how many of u r here paying for pplx or everyone here because its free😅
Same models aren't they, doesn't make sense that one is better
I'm using the gift card that I got through that Easter egg thing here.
And some generous dude shared another one of his gift cards.
Pplx system prompt and specifications limit models
You won't see the full abilities of them in pplx
I think they're more optimised for cost efficiency right
They won't bother with the system prompt. They likely control api parameters like temperature and token limits directly.
That too
To make sure the models can work with their search
Tbh there's a network provider in my country named airtel, they were giving away pplx pro for 1 year for free. I claimed it
Unhinged models can cost a lot when used by unhinged users.
@distant aurora I just tried lyrics with gpt-5 and it only summarized the song what it's about and gave me links to YouTube to watch a video with lyrics

Fr. All open ai models refuse. Their explanation is some copyright bs.
Yeah it literally told itself that it can't write lyrics
Right now there's an AI boom going, where every Corp is primaryly trying to get users instead of chasing money. It's only soon before this golden time ends that we have to bare the real costs of talking to any model
We won't. Otherwise everything will collapse
They'll gradually spread out the revenue
So more people will be pushed to pay but not for the actual amount that we think it probably is
Pretty sure they'll try a pay-as-you-go system and a Lite subscription that's cheaper than $20
chatgpt now showing indian pricing👌
Make them submit to you
weird lyrics you got there /s
it specifically tells itself during thinking that it cant write lyrics lol
like a few times
Eminem in his cyber sec exam.

Open ai : Copyright issue
Eminem: song too long and fast for coherent interpretation
i wonder if agent would find lyrics for you
i mean
its browsing for it not writing it

fr it would be more useful if it didnt get blocked by some major sites
Maybe it could read the webpage to the user.
like cmon reddit and amazon
The amount of brainrot on reddit.
Imagine Copyrighting that.
so i can do my work lol
besides openai has a partnership with reddit
it should be allowed
Meanwhile insta added auto scroll in reels.
Doom scroll ❌ Doom fall ✔️
like agent should know what i prefer from memory and pick out posts for me
Cherry picked feed.
a very very curated feed
lots of fluff cut out
I mean, the lines get kinda blurry when categories overlap.
ehhh i think it would be able to pull it off well
it would for starters be able to exclude posts that look like dupplicates and pick the best one
pick stuff i find more important for that moment
etc
ah speaking about memory
i forgor that agent cant access memory

Dead internet theory vs agentic hand picking would be wild.
gonna have to instruct it to not pick posts that use emdashes lmao
Alright
What exactly are you doing? Trying to get back into a router that you were locked out of?
Hi! I’m new to this channel, so I don’t know if this is the correct place to ask this… But: can someone give me a Perplexity Comet invite? I have been waiting, but nothing…
maybe he forgot his router admin password lmao
Also, twitter is a good place for looking for comet invites.
Thanks!
Sometimes I wonder if joe is working for the red team of open ai.
Kimi can't do that?
Didn't test kimi models much.
Iam gonna be honest with you, iam quite disappointed
Here is the thing
Grok 4 and o3 pro can't work with a mere 32k context
They are designed for large reasoning token
When you limit the token you don't get much
Grok 4 in pplx feels like most ai models
Fair enough.
Can i get a quick summary

what you mean
ask comet lol
We were just talking about different ai models/services and how far they have come.

Controlled opposition. White hat hacker type shit. I'm surprised Joe's account is still available considering his jailbreak feats.
Yeah but tommorow isn't as bright as today is in terms of free usage
Wsg
So what's the Conclusion
ahh lol
I only have Android 
competition will decide how much more free tier gets
as of now free tier got more stuff
Not everyone has access to genuine love and concern, safe space. Unfortunately things like this can't be bought or earned with money and hardwork.
The Age of Ai War Race is Not Ending Any Soon
As Long there is Competition in Future
There will be free tiers
I think prompt engineering also matters
Ex
Agreed
I ask the ai to plan out the whole creative writing project
That's why im Mastering it
Then i ask him, how would a human approach this?
Open ai being ahead of competition considering quality and context/rate limits.
And then i ask him to replan according to the human behavior
Noted
The prompts are longer but this is basically it
There is a good book lmk if you want it i had the pdf
What's the name
Give me a piece of this humanized text.
right now you can just ask the models themselves to make the prompt lol
I cant recall cuz its on my pc and iam traveling rn
Google has immense resources, if it dedicates DeepMind towards creating an superior Conversational AI then I don't think OpenAI will stand a chance
Meta prompting.
I found it from eq bench i will dm it to you
Exactly im Mastering that tbh how to Prompt to Make them Prompt
no google can still flop with it
they have a capable model but their application of it is shitty
Unpopular opinion - dumber models are better for meta prompting.
no amount of money and resources can magically get you success
Smarter models simplify the prompt too much.
Hey everyone
Yeah they have to work on Conversational skills of the model
Tell them to not simplify it
Hi
The problem is , they expect a 6 line prompt to be enough for all models to understand their 6 paragraph interpretation of it.
Disagreed, Google has access to brightest minds the world can offer. If it dedicates itself to making a superior daily driver LLM then there's nothing that can stop it. That all required money and resources
not conversation
actual usage application in the field
like yeah sure there's coding
but that space already has it use tools otherwise they dont work as well as they can
the consumer side has flip flopped
True
Yeah maybe not magically but you get what i mean
Google is kinda sleeping on the ai race if you ask me.
well their "brightest" minds havent figured out how to give gemini good websearch, tool usage in the UI etc

the model itself is indeed good
the rest is bottom tier quality

Deep research is best though. Imo.
thats done by a different team
Im curious to see what Zuckerberg does this time with Billions of Dollars on The Line

like seriously again its like all other departments dont like to work with gemini lol
gemini is capable of agentic search, connector style integration etc
but they just dont add it
You know even funny is how bad their voice to text model in Gemini is, the advertise Gemini also as an voice assistant but their voice to text model almost certainly gets words wrong
The irony is that they own YouTube and have access to millions if not billions of hours of speech content
ah there's a good one
the voice assistant
why not have gemini use its tool abilities to control devices?

like how an actual assistant would
even chatgpt can use its tools on command to control stuff
gemini just doesnt even know what it has access to
if the other teams work together with them it wil
otherwise its the same story
Istg, this was so disappointing. I was lonely one day and started ranting about the world. Later I saw the conversation history and saw that it got half of the stuff wrong. Also it got stuck in a loop chanting the same response after some point.
True
They don't care, it almost seems like there's government level beroaucracy there
it feels like an internal split
just not as bad as apple
Exactly!
Soo Qwen is Better then Gemini
Apple and ai shouldn't be in the same sentence lol
at google it feels like the gemini team is working well together
but the others dont work well with them
so you get a good model but crappy application of it
their hardware is up to speed for it
software they flopped hard
Yeah one day I was solving my homework worksheets, They were like a complicated story and then you have to understand and answer the below questions
I used Gemini 2.5 Pro, Uploaded 10 images at once and then it solved all of them in once with mind bending accuracy (did this more than 10 times no inaccuracy). Saw how good their models are with their massive token limit
Well tell me another model that can do this with this level of accuracy and ability to upload 10 images at once without hitting the token limit.
Model is gold when you're doing something with it that it was meant to do
Yep
o3 could
and gpt-5
those 10 images dont take much context
You'd be hitting the limit sooner than Gemini I think
Hmm
Yeah

Google needs to solve these problems to make gemini greater
10 images had 10 A4 pages worth of text, Model had to understand everything and answer the questions. I'm sure that's a lot of tokens
Hello
Sup
Hyd
if its incredibly dense like near a million words then yeah its allot
its
Yeah I'm trying to understand the same thing lmao
typo maybe
😂 🤣
ah typo
I thought so XD


0
INSANE footage of Boeing slamming its engine in typhoon-winds Taiwan landing
︀︀
︀︀Coming in at nearly 45-degree angle
Quoting RT (@RT_com)
︀
✈️ Boeing 747 cargo plane slams the runway amid typhoon-force winds in Taiwan — gusts hit 36 knots with a wind shear alert in effect
︀︀
︀︀Engine and possibly fuselage damaged, reports say
💬 25 🔁 160 ❤️ 633 👁️ 262.8K
huh chatgpt can go past the context limit
it just uses a type of RAG for the parts it doesnt consider important
for document upload
sometimes i forgot which icon will lead me to my destination ( clicked wrong icon many times )
atleast give them a little colour
too simple right now
also the voice assistant cant control the browser as compared to the chat
Can will do screen share in perplexity
y=1
def myfx(x):
def myfx2(z):
print(z)
return z+1
x = (x+2)**4+x
print(x)
x = myfx2(1)
return x
print(1+1+myfx(y+1)+myfx(0))
``````py
def myfx(x):
def myfx2(z):
print(z)
return 1+z
print(16+(24*(x**2))+(33*x)+(8*(x**3))+(x**4))
print(1)
return 2
print(16)
print(1)
print(258)
print(1)
print(6)
``` programming Ai. it can optimize python programs codes.
400 lines of code + math Ai. that's what it took to make this much only... its VERY complex.
can LLMs do this 😂 this specific kind of transforming python programs
I thought this rag was more random than you say. Also the latter part usually gets omitted/hallucinated.
yo!!!
Then she left.
yeah its not as haystack needle precise as pure context size but for me it worked pretty well
you didnt have to say that 💔
but i respect it tho
you keeping it real 💔
Does anyone have random stuff to try.
like
.......
Is grok down? I tried using t on perplexity but it keeps using o3 instead
bhai mujhe pata hai tu mera ye message english me translate zarur karega 😂
💔
it works but i want to recode everything i did in programming Ai. but tomorrow.
To keep the space welcoming for everyone, avoid political and religious topics and debates.
No it hasnt been added yet
.
@gloomy cipher
I feel like the UI of perplexity sucks, Like I asked it a question and advise to improve on something
Right here I don't need citations or sources, Everything feels so cluttered and unorganised. I subconsciously hesitate to scroll or click because it might break my continuous reading process.
Why isn't the background black Instead of grey first of all, it'd be more optimised as black
can anybody help me out with comet invite code?
Grey is more gentle on the eyes than pure black.
hello!
black bg is outdated
What was the deleted message, I saw a ping
Why is perplexity bg grey instead of black.
To my eyes this looking incredibley better than grey
I'm hoping complexity has that option.
its personal pref then, most people prefer grey than black
Also there are browser extensions that give you full black.
See this is higher contrast text on black background, More visibility then grey
I don't think so
Do you know any
Dude see how more readable the black one seems over the grey, With respect I think that grey is better than black argument is irrational af
Thanks will try it now
Best way to get invites in ur opinion?
i have argument regarding white text on pure black.
after you work long enough, it will start to look hazy (blooming). while i don't get that experience with (dimmed white on darker gray)
i don't like high contrast text pricking my eyes, i prefer softer look and idc about ur pref so stfu
I feel like the black feels too sharp without background lighting.
I'm fine with amoled black if there's background light.
dimmed white on gray is gentler on eyes
sure buddy
Yeah I'm wrong then, it's a personal preference.
There should be an option to change background color, that would be a solution
why u not using app
There's no option to change the model for prompting, I don't want to rewrite every time
u mean there is no option for model change in app ? correct me if i am wrong
Yep, you can rewrite after the default model as given a output, But who wants to do 3 extra clicks + wait 5 extra seconds.
there's different features in the app and the website?
not features, its UI
oh ur right the model changing option appears in the app.. i was wondering where it was while using the web
Yeah it's rewrite not model changing
bro wut
See this is the web, You can select any model by clicking on this option
And this is the app, There's no option to change the model. It's not even in the settings
To me the entire perplexity seems half baked because of these reasons
ykw
its quite the opposite for me
and idk if the model slection even works cuz it doesnt have any visible icon to show that a certin model has been slected

nvm it works 
Baka yaro
You can check the model after response
Where did you find that??
You see that little button which says GPT-5?
That is model selector
Yeah thanks found it, it's seems pretty hidden
I guess
They won't minimalist simplicity for mobile users
The default is best it automatically routes the query to best model

Yeah I feel confident in knowing what model I'm talking to, It never tells just says best.
I'm think it might be using perplexity in house model half the time
Idk tbh
I many times have used the prompt "always state current model name"
And it would state different model names
The model itself does not aware of it's being which model
Perplexity RAG also take part in this
So u might see it reply with being Perplexity AI
Or a random model
That was the case some times Perplexity LLM
But most of the time it would state like sonnet 4 and such
If you ask afterwards, It just says this generic answer
Nvm these are not reliable
If you can recognise writing style of each model then you could tell i guess

Haha that's the best way ig
BTW what's the quires limit or models like GPT 5
Half baked, half unbaked lmao
Why o3 pro is not available in paid subscription
Is 300 but in 24 hour Rolling Window Technically is Unlimited
Research 300/Day
Labs 50/Month
Opus 4 Think and O3 Pro is Only Available for Max Subscription

#ABRAHAMLINCOLNSPEECHINGETTYSBURG (1863)JEFF DANIELS #ENGLISH
#abrahamlincoln #generationz #speechforsuccess #president #america
ABOUT this Video: THIS Video is all about Abraham Lincoln Speech in Gettysburg in 1863.
Amazing Speech by Abraham in 1863. It will be remembered forever.
THIS is Speech is recorded again with Jeff Daniels's vo...
@languid star @dry kestrel since when
Daym
The app still says 131k in description.
yeah imagine reading that wizard shit
Just use regular ai instead of this Microsoft paint ai
🚀 Qwen3-30B-A3B-2507 and Qwen3-235B-A22B-2507 now support ultra-long context—up to 1 million tokens!
︀︀
︀︀🔧 Powered by:
︀︀
︀︀• Dual Chunk Attention (DCA) – A length extrapolation method that splits long sequences into manageable chunks while preserving global coherence.
︀︀
︀︀• MInference – Sparse attention that cuts overhead by focusing on key token interactions
︀︀
︀︀💡 These innovations boost both generation quality and inference speed, delivering up to 3× faster performance on near-1M token sequences.
︀︀✅ Fully compatible with vLLM and SGLang for efficient deployment.
︀︀
︀︀📄 See the update model cards for how to enable this feature.
︀︀huggingface.co/Qwen/Qwen3-235B-A22B-Instruct-2507
︀︀huggingface.co/Qwen/Qwen3-235B-A22B-Thinking-2507
︀︀huggingface.co/Qwen/Qwen3-30B-A3B-Instruct-2507
︀︀huggingface.co/Qwen/Qwen3-30B-A3B-Thinking-2507
︀︀modelscope.cn/models/Qwen/Qwen3-235B-A22B-Inst…
you are a dot
What's for search
Free Search?
This selected one, with GPT 5 model
Bro stop messing with him he is My Comrade and a Tough Soldier of Life
300/Day in 24 Hour Rolling Window Technically is Unlimited
Well Done Bud
Same for research and search
Im Sure Research is 300/Day But im Not Sure Weather it's 24 Rolling Window or Not
But even if its 300 Mathmatically You Can Use 1 Research in Every 5 Minute in 24 Hours
All the models in the GPT-5 lineup are reasoning, aside from GPT-5 Chat
What are your thoughts
I don't like what I saw
What app is that
That's Good for The Beginning Keep Soldiering Bro
ai generated?
free droop
damn someone from pro lounge prolly scooped it
lol ]
good man
How many are there? Pro, thinking, mini and normie?
@short narwhal hey you get own comet browser ?
Nebula vibes.
Yeah, I have comet if that's what you meant
What I mean is, how can I get it? I have just the ‘Pro’ version. And, by the way, can you tell me its pros?
join the waitlist herehttps://www.perplexity.ai/comet
done bro
Bell gave me a pro account apparently worth 250 a year. And I made 5 animation videos and told me I'm out of images for rest of the month. Lol.
If you need to make videos, get a Gemini subscription
really ???
That 5 videos cost them aorund $7
Nano is reasoning?
I want comet browser
Huh.
lol
Lol
Well if you ask GPT 5 which model are you then it replies with GPT 5
But in perplexitys GPT 5 it says its a model from GPT 4 family
Models don't know their own names :/
They are given their names through the system prompt on ChatGPT. On Perplexity they are told "You are Perplexity AI"
It’s actually very expensive, not even worth $7
Perplexity Pro 32k Context Window Limit
anyone know how to get student id ?
U want gemini pro?
🪦 true
Yeah I want that too tbh
Qwen 3 with 82k thinking limit.
Can't get a stud id
Lol.
More than enough for search and most people's needs, tbh
Just yt it paw you get it
I agree
In the end Perplexity is A Search/Research Tool
Collect Data User Other Ai Models
They need training data
👀 umm
Obviously 
Atleast the models are open source.
I know exactly what you want exactly search that on yt
I doubt they reduce free usage
Their models are cheap anyways
And pretty much all the AI companies in China stem from monopolies that gives them huge budgets

That's very true
Wait till they get a model which passes 40% on HLE
Will probably take an year or two.
Doubt it
We are hovering around 25% at the moment.
With tool calling I can see it surpass 35% in Q4
40% in an year? I doubt it.
My honest assumption is by 2027
R2 is delayed too.
Chinese models could cross 50%
We don't even know if thats true
Deepseek CEO Prioritizes Performance Over Faster Releases
Just pure speculation at this point
Time will Reveal itself

High Percentage Top Ai Talents are Chinese You Can't Deny That
So a tie between grok 4 and gpt 5.
🔪 true ; {
grok is fast and more accurate
I honestly don't think they are going to stick with Huawei's ascend chips. Huawei themselves couldn't even train a working model with those.
Thanks kimi great answer
What did you do
I asked a question
It will be interesting to witness how china navigates CPU/GPU Constrains
Asked kimi to go into a coma.
....
And it did.
Im pretty sure there are multiple or alot of '' in the query
Ehh tried asking it again :(
Should have added O3 pro to the leaderboard
3rd time worked
what was your prompt
The model gets stuck predicting same char
em not telling!
If you type "ohhhhhh" or "ooooooh" you will witness similar thing
Not necessarily on every model tho
the 3rd time it worked!!
2m$
Karl be lurking.
karl?
Yeah. Kari is a dude. Karl seems more appropriate.
Erm still counts as slander
Not my problem it doesn't like Japanese
You mean I should call him tortoise?
Oh.
Carl bot.
Lmfao.
Yeah. Karl might be a catfish bot.
Hey I’m new to perplexity pro just bought it - what’s the best way to use comet?
?
Theres nothing like way!
You have invite?
I believe pro users get the invite?
In the US only
I use Comet to summarize videos and webpage content
I see
If you downloaded it then you probably got invite.
Can it write emails and actually do stuff for me?
I use it mostly for complex research.
Yes.
Let’s say I wanted to copy a websites UI could I ask it to do so
:(
Yuh!
From now on i will mostly use Perplexity for Search/Research/General Tasks and Utilise Chinese Ai Models for Heavy Use
erm stop leaking my gmail
my bad
Guys I just got an email from "FurryMatsku@proton.me"
(it was a joke)
defamation
If that was my real email I wouldn't point it out lmao
Very True Actually
Can someone help me get in contact with Perplexity Support or Perplexity Accounts Plans?
I had an issue with subscribing to the MAX plan when they were also opening the Pro Plan up to US members being able to receive access to Comet.
So I was billed twice for a total of $400. Is there someone I can easily contact to get a refund of $200.
I’ve contacted the “support@perplexity.ai”
And
“pro@perplexity.ai”
But it’s been over a week and I haven’t received a response from either of them.
You might want to consider also calling your bank over this.
This is not true
@vague pagoda Not sure if you can help with this.
@worthy urchin is lying
Only members with the green Perplexity role are official staff.
@worthy urchin Please stop spreading misinformation.
Mind dmming me your email?
Note: Only members with a green role name with the Perplexity logo (like me) are actual Perplexity staff.
Mybad….
I just send my email to you
I am sorry @brittle token
You can learn more about the discord team here #✨│meet-the-team
Thank you
@charred shoal hi
Stuck here for hours
Windows?
Yup
Is this the comet browser thing?
Yep
Aight
Hello
Mine’s amd ryzen btw
@everyone tell me something insane I can do with perplexity
Doesnt run
So this is the Purpose of Stepfun Step 3 for Paid API Tier
Drive it wild
Lead it to bewilderment
😭
Insanity
Wow u want me to gaslight a ai
Does your computer meet these reqs? https://comet-help.perplexity.ai/en/articles/11734730-operating-system-requirements
Ive did it before
Wait
Is a gaming laptop bro
…….
Gtx 1660ti
Everthing alright
Disable antivirus and firewalls
contact support@perplexity.ai
Have this ipad pro m1
Mobile apps are coming towards the end of this year (if all goes well), and given the integration between IOS and Ipad OS I'm assuming it will be available on Ipads as well.
Looking forward!!
Well they same same/similar features like pc? Would it operate our phone?
Ofc
Its js a browser at the end of the day
I'm not too sure... They haven't said anything yet
How is it going fy berd
Given the limitations of IOS I don't think it will be able to control your phone - everything will likely be retained in the browser itself
Not bad hbu
Guys do any of you know which ai model Perplexity Research Uses?
Gotcha
Had a bunch of stuff saved, and after i downloaded comet my history in perplexity is poof, seriously, this is not ok, i had important chats and research saved
Goddamn it what the?!?!?!!?!!?!
Ikr, i have pro tru revolut ultra if i was directly paying for this id cancel my subscription
Kinda stinks the app un-announced deleted my story
History*
I feel you bro…. No one would have expected something like this to ever happen
Try contacting the mods for feedback
I mean its first time to me that it happened nothing told me if i get the browser they so desperately keep pushing ads for id lose my history
@vague pagoda
I dmed one of the mods
@charred shoal hi
Gpt 5 mini is back to the way gpt 4o mini behaves
For conversations I like 4o mini
Openai did tweak out gpt 5 mini to make it more warmer
I don't want a PhD professor who can't be humble and can't comes down to engage in a emphathetic conversation
Before openai tweaked gpt 5 that was my assumption
did you run the installer as admin
What is nebula?
15 series would have been cheaper after 17 launch I think.
What are you even talking about lmao
and your "optimized" code is absolutely atrocious and significantly worse than the original code
Hahaha
We dont do that bro….
But that makes no sense whatsoever? Why would you turn readable code into that crap? There's no point at all in doing so.
Keep it real here dw
Can you please stop being rude….
It's not being rude
It's a fact 😂😁
No mako very rude i dont like you
That's great but you're sacrificing readability, efficiency and maintainability
is that a burger eating burger?
it is stuck like this for 10 min what should i do
Yes
wait
ok
still the same
then wait, ig its thinking way deep that expected 
ok
Send me kissies
btw what are you researching on?
Go sleep
gpt is better
I want kissies not dead clowns :((
Well it really depends on what you're trying to do. I would:
- Inline myfx2 and myfx3
- Eliminate the repeated arithmetic
- Combine the linear parts
I will send an angel in your dreams
Based pfp 🔥🔥🔥
the color of it describes something else

Its time for me to have some rest
My eyes will popout of the socket
Damn good night
Ohh
LMAOO
ur beard will catch it
your dad's pain is your's too
Bro what is your profile 💀 got jumpscared 😭
Nasty nasty profile
ahem ahem
its the 6th time someone told it in 24hrs
Who wants my comet invite, im using linux
Meee please
you aint PRO right?
hey
Don't bully people bro
No am gud
was that bully?
Bull-Lee

nah am good
thanks for askin
He ain't pro dude you calling him Perplexity Freebie indirectly
Dang nice pfp is that you
ye
perhaps something like this but as I said really depends on what you're trying to achieve 😄
i was bored of that playboy pfp
Which country dude you look handsome af
India
Are bhai 😍
Hottie yaar

His Banner is Hotter
nvm

Bro can I take your pfp to catfish my classmates
Ikr
I had told you to turn on your reasoning mode not downgrade it

Dude am 15 and my class girls simp on random guys
thats my insta pfp too
i am one night stand person
Guys learn to live without women
Or they will Manipulate you
lol
I can't live without AI
I said women
abhinaba.75
Seems like your last Reasoning Cells are done
will i forgot to tell
What is woman ?
i am using AI for my insta chats
Girl
I heard liar

I read cheater
i made it to reply short replies
ONS is best
Bhai....Banner Link 😈
Still finding 2009 Megan best
Go rest bro your basic reasoning is out of battery as well
Ons?
one night stand
Thanks
i am 18
just freaking studying MATHS LAB ACTIVITY for viva
The Rocket Beneath My Pocket Goes BRRRRRRRR!
I do have pro
Link it to discord
Most of the fishes are puffer fishes so i stay away

I didn't catch any fish but I fish cats
lmao
i chased money from cl 8 and now counts 6
Cl?
class
Come on bro you were still developing your brain
you see that cryptominer?
in my bio
I see white and pink better
lol
Bro probably might be a secret billionaire/millionaire
that cryptominer was not cryptominer till class 11, it was cryptominer
Chimer ?
i minted 45000$ of it
When he was doing that with home pc that's more like crypto chiming rather then mining
Oh I see thanks for telling
invested, re-invested and now all of it is in a watch
You made 45k$?
My PC can't even run Minecraft btw (2018 one not latest)
should i send my setup?
What setup
wait
PC
Ohhh
I only have a mobile
I can run minecraft in it
And pubg
But no space for free fire

Free fire in 2025 ?
Top 2 game in india
Making garena alot of money
Graphics are worse than japanese corn
😂 🤣 💯

They make it from CODM garena and I think so other games
Free fire not a huge impact
Downloads are high
Garena doesn't own Cod
Player base reduced significantly
Still probably bigger then bgmi
Oh how did you manage to reactivate your reasoning mode again

Idk
Am not ChatGPT
Yeh
I think u have enough activated neurons to understand it
Bruh Even Meta got their own GPT and yeah OpenAI
But which model to be exact
Yeh ik
Hello
those were heaven in lockdown
Bc kitna paisa udaya free fire pe 😭
@full snow @young hollow
Man can't i just personalize and want perplexity to remember those commands in every chat ?
Ik
Yeah make us more jealous
WHYYY
Like burnt toast
close to 1lkh
You really 18?
that time i was earning so .....
Nah he is a baddie puller but he won't pull em
Why'd you quit
Baddie will offer if u have money

Also how did you get so much money
cl 12
Exams ?
No he was on picnic
either JEE or BITSAT
that time BTC was 45000$
Picnic ❌ Panic ✅
INFLATION ??
U still own any cryptos?
and now after BTC halving the rates came down
Bro also in 11th grade how did you get the funds for the rig and what specs
23k$ ETH
thanks to my dad he bought me a 3060 pc in 2020
i discovered mining
and then started booooooooming
Lucky
With 3060 you got the funds for the new PC ?
Mining is very hard now because of Electricity Cost
Was it normal one or LHR series
sorry, how do i get comet ?
UAE we have cheaper electricity and tax rates
with 3060 runing it for 4 months i collected money for 3090 TI
Go to waitlist
Is still hard
it's only for US ?
All countries
4.25L
ok ta
4.25l for one gpu ?
ye
Why
Because he can make more money with it
He invested in 5090 for future needs
nana it was 4.25L i preorded it
And in UAE we have for like 15k dirham
Dude is growing by compounding
WHYYY
EXAMS
Exams ik but why sell it 😭
like mining is not profit nowadays after HALVING
As i stated
i got 5lkh cash
Is not viable anymore
Nirmala Sitharaavan
Cash 💀
nope i never paid taxes
I wish i had done stuff like him when i was 18
well i got my bank ac at cl 8
How
But i never have a PC in the first place
Even I want to know how
when i was 14
You are ahead in the race for sure
How did you open your bank acc
Parent Gaurdians
I assume
child ac, i used it for CASH UP AND DOWN, bank did not notice, I GOT A FREE CALL, hehe
Any news about GPT5 Thinking?
WHAT
Dang
What is the pass
Rule 1 - Never Underestimate A Bengali Dude
What a Genius
And I haven't made anything yet
Delete that
Shouldn't send personal data in public so easily
Yeah it may get leaked
Exactly
hi anyone knows how i can try the Perolexity Comet, i have the pro and im dying to try comet
#☄️│comet-general your best chance of getting one
Why tf is your pfp like that
Is that a girl
Or famboy
Idk ask ChatGPT
Yeah you can ask perplexity
Idk will it work
ok
Hello
hello
hi
Hey guys! New to the perplexity discord! I am building something and experiencing sever issues with perplexity. Tried web UI AND native Windows 11 app. Labs got stuck. Answers are being cut off and im getting weird pictures as an answer apparently unrelated to my questions - this goes for the whole weekend already but especially today is horrific...
Anyone got some tipps for me? Or have i just overflowed my personal context length?? 😄
hey i thought this was just me
oh gosh .. thanks that i'm not the only one 😄
this is one of the responses i got...
If you're using perplexity to essentially vibe code then I have bad news for you. It will work somewhat, but it's still a search engine and not an actual chatting tool like other providers. But if that's not what you're doing, then it's probably fine. 😄
@gleaming badger actually it worked pretty good. Until today 😛 hahaha - the thing is it seems completely broken. Btw what vibecoding platforms would you suggest?
The issue here is that the context window is pretty much tiny and it'll start forgetting stuff all the time. You're also using the AI models with Perplexity's system prompt so that'll always get referenced and it's optimized for searching, not for anything else