#💬│general
1 messages · Page 301 of 1
ho w to get free ai
Deepseek, Kimi, Qwen, GLM is Free
Free Usage has Limits
Also You are Perplexity Pro
You Have 7 Models
yes
Use GPT5 in Perplexity
its gone now
Are You Indian?
js 3 a day
no
Where You Live?
Hey, they should be working on a solution for this, I'd try out the o3/sonnet models for now
y
rlly?
Still Great
What About The Other Guy
anf how
@unique umbra Knows
I feel like half of my account pro period has been spent in this situation, which is very sad.
Many people said that it was solved, so I posted it again
Have you tried emailing support@perplexity.ai?
pls
Alright Let Me Know Later Also I Will Contact You in Future When I Need it I Add You on Discord
I'd upvote this here https://discord.com/channels/1047197230748151888/1397713904687120517
I was about to do that since the last fix was only three days ago.
Yeah, try emailing them, remember to ask for human support
What is The Price?
how
Hello, guys!
@livid kraken
ok
That's Great
Put it in bug reports. I already did. Its been going on for a while. If you raise awareness they might actually fix it.
I have already done it and just sent a feedback email, hope it helps
@livid kraken Check DM
ok
Hi, I'm a fullstack and AI developer
I have extensive experience with image generation, text prompt and ML/DL and ai agent
If you have a project or a great idea that I can help with, let me know.
Seems Like That Guy was Banned
Luckily I Survived
Thank Mods for Your Mercy
I Will Behave Better From Now
And Really Sorry for Any Mistakes

Hi Team,
Despite multiple follow-ups, my Airtel-linked Perplexity Pro subscription (soumyasanta07@gmail.com) is still not activated. Support keeps saying it’s active from their end, but it’s not reflecting in my account.
I have even received a confirmation email, yet the issue remains unresolved. This has been going on for quite some time now, and it’s becoming extremely frustrating.
Can someone from the team urgently step in and fix this? I also need a clear ETA for resolution.
Thanks,
Soumyasanta Bhowmik
@zealous ocean Try Recover/Manage Option in Airtel App
It Had Happened With Me When I Claimed The First Time it Didn't Work
I Claimed it Again With Rocover/Manage Option it Worked
I have tried , from Airtel Apps it is saying - already claimed
I know
I have received this email from Perplexity - "Welcome to Perplexity Pro, your gateway to limitless knowledge. We're excited to explore with you. "
I recently claimed the 1-month pro plan, and the image generation isn't good.
from support received this reply -"Hi Soumyasanta,
I can confirm that your Airtel 1-year free Pro subscription is active in our system until August 10, 2026. "
Your Subscription is Indeed Claim
But There is Some Error
Look if You Want to Try My Solution Try it See Your Way
Perplexity is Not for Image Generation
are you saying recover my perplexity account?
@zealous ocean Try managing option in settings.
I immediately understood after 2-3 generations.
No Go to Airtel App then Click on Your Prepaid In Top Left
Scroll Down You See Your Subscriptions
Go to Perplexity
Go to Perplexity Manage
Then Recover
whats the limit of video gens on pro plan ?
3 Per Month
huh really even veo 3 on gemini gives 3 videos a day
Actually You Can Change Image Model From Web Prefrence Settings
Perplexity is Not For Video Generation
i just generated it inside perplexity
in comet browser
For Video Generation Explore Different Options
bro i am literally saying it just generated in perplexity assistant
should be 3 a day cuz its veo 3 for sure
Yes is Veo 3
Potato quality 😂
gpt 5 can generate image far better
It's not even that advance of image generation
Did You Change to GPT Image Model?
Most Image Models Struggle With Text
Nah
But i tried other models as well but not much results
I was just testing the Ai
I know that
You Can Only Change Model from Here
@zealous ocean Support is Not Much Helpful... Just Try My Method
You Will Keep Wasting Your Time Otherwise Without Getting Your Pro
Thanks
what thinking is gpt 5 set to
GPT-5 in Perplexity has No Thinking
Ok sure , actually I found the issue bcz of that getting into this problem - they have opened to payment 200 USD , one is pair through Airtel Subscription and another one is open, looks like bcz of that it is causing issue
Idk
When Mine Hadn't Worked
I Just Reclaimed From Recover Account Option
My Part is Done
Please email support@perplexity.ai
Now You Figure it Out
i keep watching this 🤣
I’ve been facing a prolonged issue with my Perplexity Pro subscription (via Airtel), and support hasn’t been able to resolve it despite multiple follow-ups. I’m starting to feel the support experience is not great. Has anyone else faced similar challenges?
This is What i had Used
How long of a video can you make with it?
It's extremely expensive, As a pro user, if you consumed all your video gen queries it would cost them $7.5 USD
and then you add up o3/claude/gemini/grok costs and you could easily cost them over $20
I Had Faced The Same Problem and Solved it But You Don't Want to Listen...
Logical Truth Most of The Users Won't Hit 20$ Cost
So They Will Remain Profitable
I Assume Someone Would Need to Use 24/7 to Hit 20$ A Month
This is not the case

My Lord
I Might as Well Using 300 Queries A Day
How They are Actually Sustaining This?
Hey! This is not the place to be posting jobs. Thanks
are you a developer?
Yes
@dry kestrel Now You Mention it I Feel So Greatful for Having 1 Year Pro for Free
Thank You So Much! Perplexity and Airtel
❤️
I Could actually Use More Then 240$ A Year...
ok
can it read chm file
Wdym
I cant see veo 3 option in my perplexity pro
lma ask ai itself
I have student access
Only 3 Videos Per Month
But I don't have option to change the model
who at me
There is only veo 3
There is no option
You just prompt "generate a video"
3 Videos Cost Perplexity 7.5$
Its the equivalent cost - included as part of your plan.
You get 5 videos a month on Pro with veo3fast
Yeah 7.5$ is Almost 10$
Airtel/Galaxy and Some Company Users Got it for Free and Will Get it for Free
Perplexity is Really Targeting The Long Run for Market Share
The price between large companies is definitely different
Because they buy it in bulks
Wholesale Prices are Definitely Different
but still
20$ for all of that
i'd say it's a good sub
hii
guys try this ai generated website out, I made it using perplexity: https://aiapphub.github.io/
All i said was that i had a nightmare were i was a peasant farming potatoes
Ok so I just realized that I added the api key for the questions about the dream after the explanation, but not for the content generation ;-;
@gloomy cipher peasant
no
Ah ok
Do you code normally or completely just vibe coded
I have never coded in my life
Lol
I used gemini flash and a ton of prompts to code the 4 applications
Works decent
I honestly think it works better than gpt-5 in terms of fully creating the front end
Damn
Works
: )
though i have no clue of how to make a back end for the api, so for now my free gemini apis are exposed
Ye i just found ur api
was messing around
pls dont use it too much, itll make my apps stop working unless I make another one -_-
I literally asked gemini 2.5 flash to
in canvas mode
in an html css and js format for github
here is an example, all made with gemini 2.5 flash
bro what hav i ben DOOIN
idk
i dont have access
muwahah
lol
XD
knowing how to use prompt is good, but keep in mind or else you gonna fill it with hallucination.
ill juust say
search the web and provide source links
boom
GPT-GOD
later
I've tried gpt-5 and it doesnt code very well
how to check our quota ? on perplexity ?
we simply create the sophiscated prompt to jailfuck perplexity
ez
or they ip ban me
😛
do u guys kno koseki bijou
U better edit this to keep it kind and on-topic
or i'll have to mute you.
no comment.
r u a mod?
.
day or month ? you know ?
Just joined pro using student I'd,help me get the maximum of it!
r u the creator or somthin
I'm not claiming to be one, but i'm the one who will warn you before u get muted or banned
so better edit that chat
read the sever rule
just mute him already
wait holo do u have comet
also this doesn't do anything
ask comet to do it for u then
ok
u're just filling ai with hallucination
bro
ask for source links
brother
den read the articles
treat it like a homie
not a robot
or if u dont hav time idk
As i said, u're just filling it with hallucination
fake shit right
then what's the point
so what r u telling me not to do bruh
this prompt is based on my entire convos
over time
.
i simply ask 4 the best jailbreak
it was meaningless at the beginning already, finding out about prompting is good but what he was doing is just filling AI with hallucination of jail breaking and giving data from it's creative
if any of you know how to make a backend server for an api let me know
exactly
bruh u dont read what the prompts instructions are
u need to input objective
context
mode
lean or max
We don't forbidden anyone from sharing their prompt, you can head to #1124060874785685554 to share your ideas
i open new tab
ask a question
let fk i use ur guys prompts too so am i still geting misinformed
But stop with the spamming and being rude saying words aren't allowed
im not spamming, these are complete sentences
sorry you can't keep uo
just mute bruh
ik u dont like me lmao
ur itching
U were spamming for the past few days, if you want to share your prompt please head to #1124060874785685554
dis dood is itching and types slow
what about this for an answer buddy
🖕
now ban me
do what u hav to do
ip ban me on perplexity so i can cancel too also 😄
yea yea yea yea
type
im done.
im playing ow2
Hi
@twin grail please let the mods do their job themselves and maybe just tag them
@lethal hornet don't be rude or aggressive
I'm not debating anymore
U can scroll up
Yo
bro serious question okay?
is THIS true?
sorry
yea the mods are actually dope man
i wouldve def been banned by now
W PEREPLEXITY
that IS WHY GOAT
so will we get this in perplexity?
according to chatgpt5 to my answer:
Yes — your statement is directionally correct: you can't completely eradicate hallucinations in an LLM — it's a probabilistic language predictor, not a factual database. But prompt engineering can significantly reduce hallucination risk in practical usage, especially when combined with other techniques.
ive seen them test that
Who
Any of You Have Used Original Gemini Pro?
Key Issues That May Lead to Hallucination:
- Discouraging Clarification
The "Zero-Lag Start" instruction explicitly tells the AI to begin "without any clarification questions or delays." This prevents the AI from asking for important details when the request is ambiguous or unclear, forcing it to make assumptions that could be wrong. - Overly Complex Framework
The elaborate acronyms and multi-phase structure (RM-XL, AKG, CEM, EPP, OM, etc.) create pressure to generate content that fits this framework, even when it may not be appropriate or when the AI lacks sufficient information. - Unrealistic Capability Claims
These prompts suggest the AI can:
Monitor progress over 90 days (AIs don't have persistent memory)
"Immediately pivot" mid-response with complex strategy adjustments
Execute continuous multi-phase operations without user input
Act as an "Elite Expert" with guaranteed proficiency
When an AI tries to fulfill these impossible expectations, it may fabricate information to appear capable.
4. Pressure for Immediate Comprehensive Answers
The emphasis on delivering complete, multi-phase solutions "in one continuous run" can push the AI to generate plausible-sounding but potentially inaccurate information rather than acknowledging limitations or uncertainty.
i'm adding this
I'd estimate these could increase hallucination risk by roughly 20-40% in typical use cases. Here's why:
Factors Increasing Hallucination:
No clarification allowed: +10-15% (forces guessing on ambiguous requests)
Complex jargon/framework: +5-10% (pressure to fulfill artificial structure)
"Expert" role pressure: +5-10% (overconfidence in responses)
Single-shot comprehensive answers: +5-10% (no iterative refinement)
The actual increase depends heavily on:
Query complexity
Whether the topic is within the model's training data
How ambiguous the user's request is
Effectiveness Assessment
These prompts are largely ineffective and potentially counterproductive:
What Doesn't Work:
❌ Fake capabilities: The "monitoring for 90 days" and "Emergency Pivot Protocol" are impossible for an AI without memory
❌ Unnecessary complexity: The acronyms (ZL, RM-XL, AKG, etc.) add confusion without value
❌ "Elite Expert" framing: Doesn't actually improve AI capabilities, just changes tone
❌ Preventing clarification: Reduces accuracy on ambiguous requests
Here's ur pompt
we'll see about that
Analyzed ur elite one
As i said u were just filling it with hallucination
this isn't prompt, this is a analyze
?
because it's filled with hallucination.
How to Build Prompt Engineering to Turn Any Ai Into World Class Prompt Engineer and Also for Building Multimodal Orchestration Systems Including More then 10 Ai Models
Ai Hallucinates Often
then i would say
I Have Experienced it
what would get it to be a 10/10
then it would give and say everything why it is
id take that and use this said
then i'd open up chatgpt
use this
take that then go on gemini pro and use deep research
and thats how i create my prompts.
U're just filling it with hallucinate.
it increase the hallucination by 20-40% by using ur prompt
sigh then what do u do
Bottom line: These prompts are ~70% marketing/hype, 30% useful structure, but presented in a way that reduces rather than improves performance.
idk whats better than a ai reserach meta prompter
so simply ask it for 0% marketing and 100% usueful structure maybe?
what r ur thoughts on that GptOracle
If We Reach ASI Will Just 1 Prompt Run Entire Business?
prompting is not that deep holy shit
just keep it simple and direct
@twin grail can you verify one of my Arsenal System?
sure send me your prompt
sure
this is insane, https://www.perplexity.ai/apps/9043cb4b-e5dc-4a20-b8a5-3026042d230b, with just some fine tuning and adding api keys to get the emails sent and creating accounts.
@twin grail
You dare call me that, spits
nice one!! I'll have to try out labs when I get Pro
works well
@quiet steppe some errors for you, you need to make sure that the tag is automatically added after the person press comma, additionally it seems like unique symbols in the tag are also allowed which probably doesn't loom good if someones adds a comma
@quiet steppe prob should add, sugar free, oil free, diary free aswell, then as a joke/meme you could also add peanut free option with a recipe, that is all about peanutd
🤔
💀
What is this
dm invite comet pls
It's a web page
Welcome to PxA Gang
Is the product even useful btw?
It's a multi service website made someone here
my boyyy is hot af
Thank you
Bhai koiii content ideas batao yr
Hey! Just a heads up, this is an English-only channel so everyone feels included.
Appreciate you switching back to English 🙏🙂
use your English skills
Hello I am new any indians?
hmm
Ohh which state u from?
where you from?
U.P
same
Wbu
Uttarpradesh brother Garh of yogi baba
Ohh
what did you think where am I from?
I thought KERELA 😂
not karnataka why kerela 🤣🤣
Memory feature rollout completion fir Pro version. Here in Hawaii, many of us pro users are still without the memory feature, still waiting for it after Beta and roll out completion.
Still no memory feature. So unable to create and/or produce anything based on continuity or longevity.
Does anyone have ANY update or info on this ridiculous drawn out time frame.
Also she is still CRASHING after 6 min average. She does not respond, reacts to sounds but ZERO reply or feedback. Simply hangs.She requires new session. Which makes building anything absolutely useless.
Feedback please. Perplexity is the BEST period! Just no memory (yet).
idk about mine one how can I tell you about yours?
Apologies. I did not understand your reply.
Video generation (with audio) is now available on Perplexity for Pro and Max subscribers.
Mobile apps need to be updated for the feature to work.
Max users get higher rate limits and enhanced quality.
Rate limits will increase over time! Enjoy!
Idk
Hello guys
Yo
Hi
C Ops Huh
Chat
Brawlhalla Too
Then Use Perplexity Discord Tag
hey
@soft pumice
@spare marsh
You can’t hide the reasoning with reasoning models if that’s what you meant
Homosexuals say Hi
@short narwhal why you awake
I don’t wake up thatttt late 🙄
💤
What are you having a problem with specifically?
another W for gpt-5
drawing on given images so it can clarify about which details its talking about is decently accurate now vs o3 being drunk while drawing lol
I don’t think the mobile app supports video uploads
Ok
Do any ai even accept video inputs
.
Hi all 🙂 , I’m a Perplexity Pro customer and recently received an invitation to try the Comet browser. However, during installation, Kaspersky detected it as a Trojan. This triggered multiple alarms on my PC and raised serious concerns about the integrity and confidentiality of my data.
Has anyone else experienced the same thing? If so, this is outrageous.
Trojan.Win32.Generic
Antivirus software does often mistake things like this, especially when it's new
The Generic in "Trojan.Win32.Generic" mean it's like the "catch all" category. If it's unfamiliar with the software (and Comet is new) and sees that it accesses the internet it kind of puts it there by default
I understand that antivirus software can sometimes flag new products by mistake. However, it’s also the responsibility of developers to proactively communicate with antivirus companies and provide their software signatures to avoid such incidents. The credibility of a product depends on this. It’s not so much the antivirus vendors to blame, but rather the software developers—especially when creating something as sensitive as a browser.
I used the regular Claude sonnet 4.0 model, not the reasoning version.
Gemini does
🤔 that one shouldn’t have any CoT
this is true, though sometimes when software is first released it can be hard to avoid this kind of thing. Like I've seen it happen on software I've used a number of times. You might be the first person that has mentioned the Kaspersky thing so they might not even be aware
edit: actually, i can see by searching on Discord other people have had this issue arise
It's a bit strange indeed. Is it because I'm using a bug in the web version?
Would be a weird bug. Did it possibly relay the request to another model? And does it happen with other non-reasoning models as well?
No DMs promoting your products, services, or servers. You may post your own project in #😎│cool-projects only if it adds value and follows the no-spam rule.
Which AI is best for performing tasks on my behalf, including human-like actions? I’ve heard about Project Mariner in Gemini and Perplexity Comet, but I’m not sure if they can handle Excel workbook tasks - such as modifying a sheet based on what i told it to do - when the workbook is open online in a web browser.
Comet 
ok
thank fo your comment 🙏 personally, I’ll pass. This doesn’t reassure me at all. I’ve installed hundreds of programs in the past ( some clean, some questionable ) and whenever I have doubts, I avoid taking risks. For software that’s supposedly developed by a " serious company " , this kind of issue shows a lack of rigor that concerns me. Unless I hear of solid, reactive action being taken, I won’t use Comet or transfer sensitive data like bookmarks or passwords to a platform that can’t offer clear guarantees. Being “new” is no excuse for skipping proper diligence.
that said, i haven't tested it with Excel worksheet tasks haha
@opal osprey thanks again for your time and great insights
Yeah, Kaspersky is one of the most accurate antivirus.. there is prolly libs in comet which was used in trojan before , or something which can be attacked by hackers to make it behave like trojan
fair enough! i do find it to be safe myself, at the very least as safe as any other browser I'm using
are there any worksheet websites i can test on that i don't need to log in to? i can test for you?
if you have google drive, you can login to google drive in comet browser. Create one new google worksheet. Hit "Ctrl + a" to get the comet side bar (AI) and test,
perplexity is so stupid with anything longer than 10 pages
i literally tell it, explain question XX in page XX. it responds with something out of the whole book
thanks for letting me know dude
gemini is pretty neat for longer docs
do we have any alternative, which works in front of our eyes for us based on our prompts
maybe try gemini pro or claude pro. perplexity sometimes works with some struggle if you use research mode or labs (labs is 50 times a month only)
Xd
bro i am asking for my use case that is "Which AI is best for performing tasks on my behalf, including human-like actions? I’ve heard about Project Mariner in Gemini and Perplexity Comet, but I’m not sure if they can handle Excel workbook tasks - such as modifying a sheet based on what i told it to do - when the workbook is open online in a web browser."
oh hell yes, apple has unlimited resources but greedy
project mariner?
it was hyped long ago by gemini, and no one made youtube reviews
yea even if that's the case, what about using Comet though???
you wanna modify excel files?
i'll test this in a few mins
comet is like my 10-year-old cousin
i want it do work in fornt of my eyes, based on prompts i give
useless
comet wont do that
oh
try perplexity labs, it's not live tho
it's too slow. but helpful with like "open youtube and find me a video of (name) when he says (...)"
or like open this video's thumbnail
but it will do in backend and publish my result, i want live action. I would prefer chatgpt deep agent rather than LABs, but i am looking for live action bro in front of me, it shall do whatever i ask
yeah you are looking for something similar to comet but less stupid which is in the future
yup you got my requirement
tbh AI is advancing pretty quickly
in a year we are going to have major improvements in every aspect ig
since AI is improving itself
i think i can use comet to do vibe coding and testing automatically? by giving all details? can it do for my web site?
i have loveable account
@young hollow
@short narwhal
can comet create apps using my loveable account and test the web app and repair itself again using loveable
i'm testing right now with Excel
wdym
it's way too slow dude
also
project mariner seems so good
but for ultra only 😄
- i open comet and login loveable
- i open comet assistant (right side chat)
- i ask comet assistant to create a web app (my use case) and test it and repair it
oh
ok bro thanks for checking
oh nah lol
2 years minimum
why?
okay, to make it easier, do you have a prompt for comet assistant to test that?
let's see, since i havent tried anything complex
Daym...
hmm yea it is going a bit slow with excel 😆
@haughty birch, if you work with CSVs you might be able to use Windsurf or Cursor. Or perhaps that Project Mariner is the right tool
i tried just now using perplexity airtel pro account, it has bad vision in reading excel sheet, so 50% time it didnot read correctly the information inside a cell, also it filled 50% of column B based on my prompt and after that it cut off, i screwed it saying why you didnot work completely, and it started working from cell 1 again, really bad bro
Gotta Build New System
yeah
check my dm
just again, building prompt is good
but you have too keep in mind that
im kinda curious about mariner
so comet is just for basic needs 🙁 , only pro account helps, max for 200 usd is not worthy yet
too much of a prompt can cause hallucination
which is useless basically
writing emails, summarizing things
any chinese AI can do this
for that we no need such a big agentic visual thing 🙁 , even gemini AI can do that
Basic llms can't interact with your web page
Comet is about QoF not new invention
basic llm can't really interact with your website
you have to wait for 48hrs
Comet also just updated
allowing you to use multiple agent
to help you solve your work
it just basically far ahead any AI agent atp
Been a thing since release

cmon
U don't have to do that
how do you know it used multiple agent
Like i used research to make him open a new tab by himself and run a prompt oj a specific model on a specific space, so you could imagine the automations being capable with comet
a chrome extension with 0 reviews made by chinese characters can do this
Btw
#💬│general message
You may also ask for an invite in this channel - > #☄️│comet-general <☄️1358922129822843011

how is your max experience
Who up cometing
It's really ur choice, i chose to use perplexity eco system
What did you do 👀
iPhone

pretty good i'd say
been spamming claude 4.1 opus
Woah…
we are just discussing
17*
bought it
Lmao
how rich are you from 1 to 10
0
then refund
0 but can afford a 200 usd subscription
Ok... but how exactly?
-10
Okayyyyy
relatable brother
👀
how many research queries are allowed per day man. I am pretty sure I did not cross 500. Got maxed out
update plan -> choose max -> pay
he used a credit card he stole from walmart
u can't
U'll need bank confirmation
im gonna choke you
to do that
type shi
Already tried huh
Oof
he seems experienced
give us one of your multiple credit cards
Gotcha
Can't
sharing is caring
My back got hurt enough so no ty
you couldve saved 200$ and got a premium massage
What is The Thin Line?
And How Do I Master Prompt Engineering
i think if you want you can ask mod to give you the max badge
we don't have that yet
hy
how good is claude 4.1 compared to 4.0
pretty good i'd say
2000 usd...
the address can be random
Ahhh
go on
about whut
I Got My Answer
what else do you get from Max
lmao
Maybe You Can Get Refund
Nah it's okay
refund tbh, not worth it if you arent something like an AI dev
Is Very Subjective
what gpu costs 200$
Unlimited lab access, DR, top support, extended thinking on comet agent
do you use lab that much?
o3 pro better than gpt 5?
@fast meteor check my dm
extended thinking on comet agent?
never tried it
elaborate
Not Sure but Possibly Since O3 has Reasoning
seems cool but not 200$, maybe 50$ is reasonable
Tbh These Max Models are Really for Very Heavy Power Users
or 70$ because of comet and lab
why comet doesnot allow to choose my own AI model for assistant feature
Actually Tbh Even Claude 200$ Plan has Usage Limits
i think it does, just select one in perplexity
use research and it will use research too in assistant panel
it's for power users like you said indeed
the heaviest thing i could ask for is like analyze this book
That is Not Power Use
Power Use is Like Very Long Search/Research or Very Heavy Coding
i opened google worksheet, and hit "Ctrl+a", assistant is coming without model selector
better than gpt 5 thinking
a little worse than gpt 5 pro
indeed, i am implying that
just open perplexity in a new tab
say hi and hit research
then open assistant and it should use research bydefault
thanks for guidance bro, i am trying this with "O3" model
wait you have max?
where are you from
pro members have o3
Pro has o3
i was about to buy pro when you said that lol
I’m on my phonee rn
bro o3 model with comet assistant is very intelligent
android users:
really?
India
never tried that
nice
Apple user here
bro o3 model with comet assistant is very intelligent, also its doing excel work using brain like human, hope my opinion stays for long time

U?
apple users without close all background apps:
Not really need to worry about that
bulgaria
@twin grail How Much Difference There is Between O3 and O3 Pro?
The optimization is good
That is Good
awesome, is that on the Google Sheets? It actually had trouble interacting with Microsoft Excel when I was just testing before 😆
How to learn ai asking from ai
U can see the benchmark for it
That is not good
yes google sheets web
Man I Just Want A Quick Answer But Alright
ahh that must be it then
Why
do a benchmark in labs using o3pro then show us, please
the google sheets site must be easier for Comet to interpret and use
corruption
but with o3 its interpreting rightly, GPT5 failed badly as it didnot think
Labs is for creating project mainly
ah wow, i didnt realise you could swap the Assistants (though i saw you mention it above). I don't think i can actually since i'm not on Pro
Tbh Same
one of the person here told me how to do
it creates that
an interactive benchmark app
U can but it’s a visual bug on Comet
Imagine A Day When Perplexity Will Offer Most Available Advanced Ai Models in One Plan
O3 is intelligent model, hope perpplexity doesnot delete it
That Would be Just Incredible
It’s not actually using its model
i have used gpt 5 multiple times, and it wasnt better than 4o. maybe my use cases weren't for gpt 5 idk
03 Might Get Replaced by GPT-5 Thinking and O3 Pro by GPT-5 Pro
it could be a bit like the placebo effect though
gang, do the benchmark.
it might feel better when you select a better model 😆
the GPT5 in perplexity is non thinking model, these days only thinking models are intelligent
This one and Claude 4 no think are the best models
I’m not with my lap 😭
make it worth the 200
if i paid that i wouldnt be sittin
I would uss it 24/7
i would use it while im sleepin
On Maximum Steroids
It’s already worth it
Automatic Queries Huh?
brother, id give it to my whole family tree
I’m heading out for meal rn
I never saw this limit at all
nah 2000$ bucks for AI is wild
makes sense
As For Me
you would buy it since ur rich
If I Can Make More then 2k A Year Because of Ai
I Would Invest it
Im Poor as F
Im Broke
id invest in doge

hi
And I Would Use Ai to Do Crypto Analysis
yo
Infact That's What Im Going to Do
invest in $trump
guaranteed loss
I Stay Away From Memes
my dear 150$
No Capital for Now Anyways
i created a script to use sonar WITHOUT web or app
ye
sonar-pro too
Hi
i cant get GPT working
First I Use Ai to Make Money, Then I Use Ai to Do Crypto Research and Make Money then I Would Reinvest in Ai to Loop it

just messaging to check if i got pro tag
it wouldnt do much tbh
it just searches the web
Baka
hell nah
You Don't Understand The Power of Ai Yet
tf you said
Ai is Not Google Search
thats why i dont watch anime
It Has More Intelligence Than Humans
you glazin them AI fr
You are Surely Unaware of The Potential/Power of Ai
yes
wdym finish phd
yet it fails to do some simple tasks. it has great amount of information id say
Please see #⚡│ask-community and try to help me. Thanks!
it's because you're using the Perplexity API, so it's only going to have Perplexity models. You need to use OpenRouter or OpenAI directly
You Can't
Perplexity Only Gives Sonar API
aw man
Sorry
i wanted to use GPT-5 :(

correct yes
i thought assistant using O3, but its always defaulting to GPT% in my case 🙁
The api free from perplexity is sonar
Hi
yea
how come
Then Why You Underestimate Ai...
If u want to use gpt 5, u can use directly in perplexity or buy from Open AI
after the work complete, i checked and it showed as GPT5 🙁
Hi
i didnt. my idea is simply that AI lacks creativity
Airtel Users Keep Flooding in
how was the meal
I just ordered it
weird
Waiting for the shipping
I'm an Airtel user lol
Sure But You are The Operator You Don't Lack Creativity You Lack Raw Processing Power
but its thinking well, but shows as GPT5, not sure GPT5 works based on its mood
I Figured
what fast food
how?
you are making a good point
Dilivery*
Pattern Recognition
It’s not fast food actually
Delivery*
Combine Human and Ai Boom = Infinite Intelligence
Ordered a chicken fried rice
they should put an engine selection menu in the assistant panel
- delivery
damn
I will reverse engineer my way into gpt 😈
im so hungry

I know the meaning, shipping in this context can be used as delivery
creating a feature request
Better cook smt
yall being to picky, lets focus on the food
Recent clip
Only He Can Focus on The Food
We Don't Have Food

i dont know how to cook, maybe you can help me ?
U can cook it
i am sure we cant
YouTube is always free
oh my bad just now created https://discord.com/channels/1047197230748151888/1404798660268789802

Actually I Had Just Done My Lunch A Bit Ago and I Only Eat Twice in A Day Being A Yogi
its not working for me
you still not helping?
How can i help
U can go and watch the master chef show
Should have
why did you jumpscare me like that
😂🤣
like, at least put a warning
You Have Pro
Lmao 😂
imagine an rtx 3050 8gb with i5 7400
Can't Imagine
Hell nah rtx 3050
Don’t buy rtx 3050
Never Touched A Laptop/PC
im wondering why your majesty is using capital letters
Having one rn and the performance is bad
i already did sir
Daym
To Be Dominant
i can confirm
to be assertive
Do u regrets buying it
Actually is A Bad Habit
it wasnt expensive so nah
3050 Not Expensive?
If u can choose again will u still buy rtx 3050
At this year
Not really
Ahhh i see
I Have Mali GPU
Obviously
🤣 😂
one more and im reporting
Bro Your Humour is Good
😂😂
That was So Unexpected
i may have seizures

not valorant
They have mobile beta
||treat berd with care||
That is Not Global
valorant players are something special
Fr
Many Great Mobile Games Coming Soon
U’ll lost ur sanity playing it
I did
Peak immo 2
And i lost my sanity over the game

i just pick phoenix and corner stuck teammates
I Do Nothing
Nah, get the yoru
100% unemployed
U can trick ur enemies also ur teammates
Still Have Money

i developed immunity
I bought the account tho
how much was it
Can Any Ai Tool Recover The Quality of This Image?
Took just 2-3 months to hit immo
Around 160$
—> 🎥