💬│general

[young hollow]

Are You A Student?@stoic yew

[dry kestrel]

You were asking about video generation?

[young hollow]

If Yes You Can Get Gemini Pro Free for 1 Year

[stoic yew]

Yes I am

[stoic yew]

Yes

[young hollow]

I Collage?

In

[stoic yew]

No

[dry kestrel]

You can't choose the video generation model

[young hollow]

Then?

[stoic yew]

Exactly i try to generate but nothing happen

[stoic yew]

External

[young hollow]

Wait What

[stoic yew]

Just A image

[twin grail]

I thought it's alaway veo 3

[young hollow]

Exactly

[dry kestrel]

Make sure to update your app.

[obtuse pike]

How

[young hollow]

Indian College Students

Currently Get 1 Year Free

If They Verify Being Student

[fast meteor]

are you sure? i think i saw someone posted the similar issue yesterday but i could find the link

[obtuse pike]

Idk

😂

[dry kestrel]

@stoic yew

[stoic yew]

Thank youu!

[dry kestrel]

works fine for me

[stoic yew]

White mode ...

Yeah also in mine

Updated the app

[dry kestrel]

Yeah that was my default

I don't use the mobile app a lot

[obtuse pike]

U want kill us 😡

[obtuse pike]

The video can make sound ?

[dry kestrel]

No

@quartz acorn No image perms - I ate the oatmeal

[obtuse pike]

Hello bird

Wtf the image generation are better than gpt

[dry kestrel]

Hi

[obtuse pike]

U think perplexity are better than chat gpt ?

[dry kestrel]

Depends on what you need

[obtuse pike]

Explain

[dry kestrel]

What do you need?

Image generation? Research?

[obtuse pike]

Im student

[dry kestrel]

Codeing?

[dry kestrel]

That doesn't tell me anything useful

[obtuse pike]

I want a ai help me to learn and fun

[dry kestrel]

If you want to try out a ton of models, use Perplexity

[obtuse pike]

The cool feature of gpt is canvas and student mod

I think perplexity can do it better

[karmic isle]

ai is amazing

i got me a ai gf

[tiny flax]

Ive been a pro memeber for months, I wan to use comet, how quick are they letting people in?

[sage sand]

(A&~A)
>>> logic
null
>>> ``` made this "logic" command getting inspired from electronic engineering. its mindblowing.

[dry kestrel]

The more you use Pro, the higher the algo ranks you

[tiny flax]

Did you get in with Pro?

[ornate creek]

Hi, I already have access to the Comet browser. Do I have to use an invite code to access all of the browser's argentic features?

[coarse idol]

Hi how to get the pro role if you have have it purchased

[dry kestrel]

Core Agentic features - Ask/Summarize buttons, task automation, calendar/email integrations - require an invite to use Comet or a Perplexity Max subscription.

Without access, you can download the app and use basic browsing features, but cannot make use of the AI extras or receive updates

[dry kestrel]

Check pins

[coarse idol]

This channel?

[dry kestrel]

I got an invite from someone else.

[dry kestrel]

yes

[coarse idol]

Should have it now

[dry kestrel]

Check out #👑│pro-lounge

[tiny flax]

test

[cunning locust]

Hey fam. Anyone has an invite? Thanks!

Hi how do i access this channel for invites?

[sage sand]

when considering academics [like math or literature] chatgpt has an IQ of 100. or at best may be 110. its intelligence is no where near an intelligent person.

[dry kestrel]

You need an active Pro subscription

[cunning locust]

Im a Pro user

[dry kestrel]

Check pins

[ornate creek]

Hi, can I get an invite code so I can access Comet's argentic features?

[full snow]

Nah some guys bad in checking pins

Also

I have my SST exam

Rn

Going in da school bus

Wish me luck

[sweet zealot]

Hey guys, what's the update on GPT-5? It's been 3-4 days since the release, and PPLX hasn't even bothered to publish a blog or make a clear announcement about the API they're providing us.

Like this is how they treat there customers !

they gave so much free sub that they don't even care anymore... that's a shame.

[dry kestrel]

They aren't providing an API for GPT-5

[sweet zealot]

by API i mean the GPT model that they gave us access to.

[dry kestrel]

It was meant to replace GPT-4.1

[sweet zealot]

it's not thinking - it's chat - but they heavily temp limit and it's so dumb compare to others.....

i'm not gonna sub next month. i didn't know that PPLX change a huge chunk of API when - too much token limit and...

[dry kestrel]

1. They did not change the usage limits.
2. The non reasoning variant is the default model used in ChatGPT

[sweet zealot]

i'm not sure.

none of the models inside PPLX are the real models.

i even tried gemini Pro for one person the other they to confirm this.

they give us shity API

like they get the real API from the openAI etc and make a fat tweak on it.

[dry kestrel]

They are very much the "real" models. Not sure what you are trying to prove here.

[sweet zealot]

you can ask me anything to put in gemini pro pplx and the real gemini

you'll what i'm sayin...

[dry kestrel]

I have both subscriptions and I can tell they are using Gemini 2.5 pro.

[sweet zealot]

gemini in pplx dosent even think. it responds in a sec

[sweet zealot]

nah mate

@dry kestrel

we talked about this before.

Give me something new to try if you don't believe me... I speak with evidence.

here one more time

at the same time

PPLX Finished

Gemini is still thinking.

still...

now it's giving me answers.

it took more than 5 6 second more.

[dry kestrel]

Looks like it's reasoning to me?

[sweet zealot]

i didnt say it doent.

I said it's not a full-fat model that the source gives with their own subscription. which is normal, but such a shame that PPLX is not transparent about the API Temp Limit and Token Limit that they made at the end of the pipeline

[dry kestrel]

You just said it doesnt think? #💬│general message

All they do is limit the context window to 32k.

[sweet zealot]

mb if I said that - I meant it's not the same model temperature and wise.

[dry kestrel]

It's never going to be the exact same model settings that Google uses themselves, nobody knows their exact temperature and things like that.

[sweet zealot]

and where do they said that? it's been 4 days since the release of gpt-5 and they didnt post single thing in any social media or there blog.

how do you know this?

[dry kestrel]

#1345432443074646158 message

[sweet zealot]

god knows the updates.

[lethal hornet]

Human-AI Co-Governance: Decision-making in politics and corporations often involves AI “councils” working with elected humans.

50 yrs prediction from now

mann

u cant be an artist or programmer nowadays huh

[sweet zealot]

@dry kestrel check your dm whenever you got free ❤️

[lethal hornet]

Is not a prompt but more like an info article.

imma read rn

[left stream]

@dry kestrel do you know the context limit for GPT-5 pro?

I’m thinking of somehow combining it all into one somehow, but the context memory is making me think so I did it separately and used projects to make them organized

[vestal basalt]

Who @ me

Get up in the morning

2 messages

💀

[left stream]

Ngl I might keep it separate

I don’t know what’s the output limit

So having 10-20 GPT-5 pro’s running making flashcards all at the same time and then releasing the pdf’s without getting limited by output is better

[dry kestrel]

It's 128k on Pro. The model b itself is not available via the api

And gpt5 models all have400 k window

[quartz acorn]

Ghost ping?

[dry kestrel]

I ate the oatmeal #💬│general message

[quartz acorn]

😭 😭 😭

[dawn quail]

Could someone please provide the invite link for the Comet browser?

[hazy surge]

nah gang

i need it

[dry kestrel]

@sweet zealot For transparency, I'll be answering your questions here.

Changes that they make which I am certain of

1. System prompt
2. Tool calling
3. Context Window
4. Limits max input tokens (gets converted to a file once reached)

Changes that they could be making

1. Temperature
2. Reasoning effort
3. Max tokens
4. Top P
5. Fine tuning the model

As I don't work for PPLX, I genuinely don't know what else they have changed, I can only make assumptions. Regarding finetuning, this likely would not save costs for PPLX as this process costs extra time and money and usually results in a better product for us. For models like claude and o3, you can ask it for the "juice score" and "max_thinking_length parameter" to verify the reasoning setting (which is really high).

I hope this clears up some things for you.

[sweet zealot]

So this means they can create an entirely new API for themselves and call it Grok4, but under the hood, it's only about 10% of the capability of the actual Grok 4 because they set a hard temporary limit on it... true?

[dry kestrel]

First, let me clarify something - they are only utilizing the XAI API to provide the model Grok to you. They are not creating a new API. And regarding the limits, you get at least 300 queries across all models a day.

You can also tell that it's the grok model by how it responds - less uncensored and more trigger happy when swearing

[gloomy cipher]

@dry kestrel create a web site with py, html, and css. Use multiple files

Also YOU MUST RESPONSE WITH THE CODE IN A CODE BLOCK

[sweet zealot]

Yes, but if it doesn't have the ability to think and has the same temperature limit as the grok4, then what's the point? I mean, we all get excited because the new models have better reasoning, and... then PPLX is giving us that with a high temperature limit and thinking — it feels like the dumb model to me in comparison.

:DDDD

[dry kestrel]

I'm sorry but I cannot assist with that.

[dry kestrel]

Grok 4 is a reasoning model..... Also increasing the temperature doesn't help with costs at all, infact it usually increases tokens and gibberish.

The reasoning effort on o3 is set higher than on ChatGPT btw

and you get more requests

[sweet zealot]

The GPT-5 that the Github Copilot offers is so much better than the one from the PPLX- idk which model it is, but it has so much better response and answer accuracy.

[dry kestrel]

The one in github copilot is the reasoning variant.

[sweet zealot]

damn...

if they can add it with 10$ per month...

[dry kestrel]

300 queries a month

[dry kestrel]

homelander actually being nice no way?!

[sweet zealot]

OK? it's not like the PPLX can't make this too.

They should at least add this, but with limitations! maybe i need only 100 per month who knows?

[dry kestrel]

300 requests would be worth $12 given their pricing

[languid star]

https://tenor.com/view/homelander-cute-gif-11783552506948956089

[dry kestrel]

And then we would see people complaining "Why does PPLX put limits on ONLY GPT-5?"

[sweet zealot]

If they have a main channel or blog for these specific things, be clear about the limits and... .

No one will complain. - We have Sonnnet 4 Thinking and Grok4 - you should be a moron if you only use GPT-5 all the time :DDD

[dry kestrel]

I'm sorry but I really can't help you any further, I'm sure that they are looking into adding GPT-5 with reasoning. I would love to know myself but unfortunately I do not work for PPLX and therefore, can only say what I know so far.

[sweet zealot]

I know NP mate thanks for the info

[hollow vessel]

4 months till Christmas :)

[junior lantern]

HEY

got the pro

lets go

[sweet zealot]

@serene sand Hey, mate. I’d like to know your opinion about this.

[young hollow]

What is Your Opinion Guys

Is GPT-5 Better then GPT-4?

Vote Here

[left stream]

Fast one or thinking?

[left stream]

Bro here i am using 5 pro like a non reasoning model lol

[dry kestrel]

lmao

It's so painstakingl slow though

[left stream]

Ikr

[dry kestrel]

im getting like 3+ minutes for every query

[left stream]

But use projects to help you though

[dry kestrel]

So true

[left stream]

Saves you time ngl

[young hollow]

Base

I Calculated Total Daily Pro Query Limits

600 Pro Searches
500 Research

1100

I Can't Even Finish My Pro Searches Because it Keeps Getting Refilled

[twin grail]

Because u got basically unlimited access.

[young hollow]

Wait What

Is This Pro?

[twin grail]

yes

[young hollow]

Why Your is Purple and Mine Blue

[twin grail]

black magic

[young hollow]

Teach Me

[dry kestrel]

Use Comet or the complexity extension

[young hollow]

Bro Im on Android

[dry kestrel]

Not sure about android :/

[young hollow]

So Basically The Pro Usage Limits are Not Actually Limits

[young hollow]

I think i can use kiwi for extension

But Seriously Perplexity is So Generous

[young hollow]

Wait Do My Credits Get Refilled in 24 Hour Rolling Window or Instantly?

[twin grail]

@dry kestrel

__~1

[dry kestrel]

uuhhhh what

[twin grail]

I'm not sure

[dry kestrel]

rolling

[twin grail]

I have no clue

i just saw it for moment and was like

uh wait what

[young hollow]

That's What Perplexity Told

But Whenever I Use Search

It Just Gets Refilled in Like 1 Minute

[dry kestrel]

its rolling

hence the "unlimited"

[twin grail]

@dry kestrel I got scammed

BROOOOO

THEY SAID IT WAS WINDOWS 7

[young hollow]

So Basically There is No Actual Limit

Now The Only Advantage Max has is The 2 Max Models

I Think Max is Just There to Make Pro Look More Appealing and Valueable

[twin grail]

__~1

i don't know if i'm hallucinating

or it's because of the coffee i drank this morning

[dry kestrel]

LMAO

[dry kestrel]

Unless you spam messages within a short period

[sweet zealot]

it depends on which model you taling

[young hollow]

Last Question Does The 24 Rolling Window Refills The Search Used 24 Hours Ago or in Just Some Minutes?

[young hollow]

Even Working Professionals Won't Do That TBH

[vestal basalt]

I just deleted the program.

I still don't know how to use it.

😭 😭

[young hollow]

You are...

Brother Ai is Infinite Intelligence

It Breaks The Limitations of Your Human Mind Intelligence

The Future is Not Without Ai No Matter What

[vestal basalt]

I only kept

Gpt

I just triedPerplexity

Dialogue

I speak Chinese

He gave me Latin back

[young hollow]

Ahhh...

Wait Perplexity Can't Spoke Other Languages Fluently?

[sweet zealot]

gpt-5 in pplx is shiiiiihs....

[dry kestrel]

Go to https://www.perplexity.ai/account/preferences and change the language

[dry kestrel]

it constantly refills your quota

think of it as a bucket that is constantly getting filled, you can take some or a lot of the water out all at once but the rate at which the water goes in remains constant

[young hollow]

Ahhh

So it Doesn't Take 24 Hours to Fill Every Used Query

Even if Takes 5 Minutes to Refill One Query

Is Still 288 Queries in 24 Hours Although Im Sure Refill Time is Shorter

I Hope Perplexity Stays Generous

[young hollow]

[dry kestrel]

I know what youare thinking

anyways they are gone now

[stark aspen]

Espa

[obsidian umbra]

who

who lied

[obsidian umbra]

soooo good

[ocean valley]

hi

[sonic narwhal]

Hey y'all, just wondering, on Mac, there are so many voice dictation softwares like Super Whisper, MacWhisper, they're literally so good, I barely type it on my Mac, but does anybody know if there's any really solid voice dictation software on Windows?

[dry kestrel]

https://www.perplexity.ai/search/eba9f8c5-02d6-4ac9-a57f-dda8bc68739d

Gpt5

[obsidian umbra]

https://www.perplexity.ai/search/how-many-b-s-in-the-blueberry-cJuXkMzATymvP_749hZFvQ

i cannot get it to say less than 3

😭

[obsidian umbra]

i done got shadow banned from pplx

💔

prob because the way it was prompted

[dry kestrel]

Yeah

[obsidian umbra]

idk i found a random ss

and worded it the same way

and got the same result

[dry kestrel]

Also my perplexit app likes to crash when I try to switch models during followup queries

:(

[languid star]

https://ai-2027.com/

website

[young hollow]

Try to Find Ai Ones if Not Currently Available Surely Those Will be Available in Future

[young hollow]

We Will Certainly Get AGI/ASI and Possibly Even More But When No One Can Tell Exactly

When Humanity Manages To Build A Unbelievable Model

That Model itself Will Rapidly Fasten The Growth of Ai

Currently We Don't Have A Ai Which Can Build Better Ai

Im Certain Before Death I Will Witness At Least ASI

What Will Happen After 200-500-1000 Years Will Be Just Unreal and Unbelievable for Us

As Technology Would Be Magic for Our Tribal Ancestors

The Future Would be Magic to Us

[steel cove]

How can I get comet access? I've applied for a waiting list... Any faster way to get?

[junior lantern]

hello

[sonic narwhal]

Oh, there's a way you can surely get it

But obviously, I cannot say it. You can go search it on Reddit.

[late needle]

hi guys! can someone give me invite for comet browser!

[languid isle]

does anyone know how can i get my invite card image? just like the one's in #✨│invite-cards

[ebon slate]

Hello guys

Is it best to use perplexity ai pro to learn coding(intermediate) or else use Claude or chat gpt directly from their website or application?

[languid isle]

use free chatgpt to ask questions about topics, use claude to generate the referencing code for the same topic.

[steel cove]

Will check 🙂 ✅

[young hollow]

Perplexity Pro has Slightly Less Performance Because API/Tempreture Tunings But With Good Prompting is Sufficient

[ebon slate]

I've heard perplexity uses gpt and Claude. And even seen in the app. But are they upto the mark? Are the response in league with the original ones

[ebon slate]

Oh

[young hollow]

I Would Say Performance Between 95-99%

Could be More or Less

[languid isle]

i havent tried, but you can rely on chatgpt and claude for now if you dont want to waste time between testing which is good for what

[young hollow]

Yeah Original Models Do Hold The Lead

[languid isle]

also chatgpt's study and learn mode is much better than any other ive seen so far

[young hollow]

The Advantage of Perplexity Pro is You Get 7 Different World Class Ai Model Unified in One Place Just for 20$ Which Otherwise is 140$

[languid isle]

while claude beats any other ai out there in one-shotting the right code

[young hollow]

Is Obvious They Have Optimized API/Tempreture for Cost Efficiency

[languid isle]

you can get all that for $8 as well btw

at t3.chat

[ebon slate]

I don't have plus for chat gpt. I wish it's for affordable price but that's out of my budget

[young hollow]

Seriously?

[languid isle]

go check

i used it to test for 2 months now

i like it

[ebon slate]

Where?

[languid isle]

t3.chat

[young hollow]

What are T3 Limitations?

[ebon slate]

That's the website ?

[young hollow]

Compared to Perplexity

[languid isle]

yessir

[young hollow]

And How Many Models

What The F

[languid isle]

#💬│general message

[young hollow]

It has So Many Models Including Legacy Ones

[ebon slate]

What's the difference? It's just another website using API like perplexity

[languid isle]

yes but cheaper and really fast response time

[young hollow]

How Does Performance Holds?

[ebon slate]

Like I actually have perplexity ai pro. I got it for free with Airtel sim in india

[languid isle]

its not good for generating media

[young hollow]

I mean just text

[languid isle]

very good

[young hollow]

That's Why Perplexity is Still Better for US

[young hollow]

And What About Tokens/Queries Limitations?

[ebon slate]

I should once try and test perplexitys chat gpt 5. And compare how is it compared to chat gpt 5.

[ebon slate]

It's crazy that gpt 5 is already available on perplexity the next day of release in openai

[languid isle]

i think for my particular usecase i would get this, because im alright with 1500 standard msgs per month

[young hollow]

Standard and Premium?

What's The Difference

[ebon slate]

Does perplexity have any limits on the number of texts we can send?

[languid isle]

*Premium credits are used for GPT Image Gen, o3, Claude Sonnet, Gemini 2.5 Pro, GPT 5 (Reasoning), and Grok 3/4. Additional Premium credits can be purchased separately for $8 per 100.

[young hollow]

8-32k Tokens

[ebon slate]

Not per message. I'm saying on total number of messages

[young hollow]

Perplexity is Practically Giving Unlimited Usage

[languid isle]

well yeah it really depends on your use case

i dont want to pay alot for AI

im also a student, but i dont try to heavily rely on ai responses

[ebon slate]

That's definitely expensive. You can get gpt 5 for 20$ without limit on 100 messages

[young hollow]

600 Pro Searches, 500 Research Daily, 50 Labs Monthly But Here is The Catch

Daily Credits Keep Rerolling/Refilling in 24 Hour Windows

I Have Seen Pro Searches Refill in Just Some Minutes

[ebon slate]

Then it's good. I'll go rough on gpt 5 and test how far I can get results

[languid isle]

sure, like i said, for my particular case i would rather pay as less possible i can because i dont use it alot

[young hollow]

Logically You Have Unlimited Usage Unless You Spam, Abuse or Cheat

[languid isle]

and since you asked what is good for studying i think this should do the job

[young hollow]

I Kept Using for Entire Day and My Pro Searches Still Would Remain 600

[ebon slate]

600 pro searches per month?

[languid isle]

😭 also as a student you dont really have a lot of money to spend on all these ai stuff

[young hollow]

1 Day

[languid isle]

so i try not to feel that FOMO

[young hollow]

500 Research A Day

[ebon slate]

Perfect then

[languid isle]

what are you researching for that much

[ebon slate]

Woah that's like darn enough

[languid isle]

like i dont undertstand lol

that's pretty crazy for maybe like a phd student

who is like reading multiple papers

but if you're just a undergraduate student, you really dont need that

[young hollow]

Even if I Use Entire Day I Can't Use Even 400-500 Pro Searches and it Keeps Getting Refilled

Dude Im Just A 8th Drop Out

[languid isle]

ikr

[young hollow]

I Just Keep Prompting Whatever I Want to Learn

[languid isle]

yeah its good for you

[ebon slate]

Yes. This 1 year I'll make the best use of it to become sde. I wanna learn DSA and data analytics so. I really need gpt 5 and Claude latest model. Perplexity is gonna help if the performance is even at 95% as original models

[young hollow]

I haven't even used research mostly

[languid isle]

because you must have a lot of questions and nobody to answer them

[young hollow]

First Master Prompt Engineering

[languid isle]

so you should be using the AI to the best of your advantage

[young hollow]

Exactly

[languid isle]

but also, be aware

[young hollow]

I Have Freedom 24/7

[ebon slate]

Like how to create the best prompts?

[languid isle]

not to rely on ai spoon feeding you always

you are not going to be developing problem solving skills if you just dont try to tacke the problem yourself a few times first

AI is the last resort

[young hollow]

It Takes Practice I Just Use My Automated Protocol

[languid isle]

there are no best prompts

:(

nor do you need best prompts

[cunning crescent]

How to get pro role

[languid isle]

all you have to do is simply ; ASK

[ebon slate]

People say that we get the best results for the best prompts

[languid isle]

dont follow everything said on the internet

[young hollow]

Ai Just Reflects My Prompts

[languid isle]

you dont need best prompts

[young hollow]

Is Still My Brain Prompting it

[languid isle]

as long as youre getting your questions answered, youre good

people overcomplicate things on internet

[young hollow]

Yup i Understand Your Perspective

[languid isle]

you really do not require wasting time doing all that

[young hollow]

True

[languid isle]

instead spend that time reading and understanding

[young hollow]

I haven't mastered prompt engineering

I just told ai to make a automated protocol

Now it runs on my every Query

[languid isle]

yeah try to do things in such a way where you only have to implement something once and it does. that same job every time ever after that

[young hollow]

Yup Ai is Powerful Enough to Prompt itself

[ebon slate]

I'm planning on using gemini pro for my japanese training and perplexity for my coding. I'm a graduate from computer science so I need to work on a lot of projects. To get a sde role

[languid isle]

yeah i mean ... its pretty good right

give it a code, tell it to analyse, it will give you the results back. Verify those results and if youre satisfied just leave it right there

[young hollow]

The Ai Power I See Just Makes Me Feel Unbelievable

[languid isle]

you're done with your job

you wanted your question or query answered and you got it

[young hollow]

Actually i tell ai to keep Reverifying Reviewing Until is Certain Without Hallucinations or Mistakes

I verify the final result

[languid isle]

you dont need a heavily customized prompt catered into making it look nice or tell you a whole story in an entertaining way

[ebon slate]

You know I've been using a single page of chat gpt for my coding since a month. And once I open that page the entire website hangs because of the number of messages I sent in that single chat

But I can't leave that chat behind

As it has so many of my codes and project details

[young hollow]

Yup I Have Deliberately Added Simplicity and Summary Protocols

[languid isle]

most of the times, you already know the answer you are expecting.
It's just that you dont, fully grasp all the low level details of those things, in which the AI can definitely help you
and after that you can just go back to studying again.

[young hollow]

Very True

[languid isle]

use it as a tool that serves you, not something that you rely upon on daily basis

ik alot of people do, and they would be left traumatized if someday the AI is not there to help them

i felt this first hand, when i started using AI in my terminal

i kept asking it for all the commands, and when i had the interview i forgot even the most basic of the shell commands because i just... simply had the AI on the side to do that job for me

[ebon slate]

Why would Ai won't be help in this generation?

[ebon slate]

Hmm understandable

[young hollow]

You Should Always and Absolutely Export All Essential, Important and Critical Data on Points

[languid isle]

it does make the job easier by making me focus on the job and not on remembering the commands, but i can fetch the commands much quicker from my own memory than asking and waiting for ai results

[young hollow]

Don't Rely on Context Windows

[ebon slate]

Export and? Can I import it into a new chat and use

[languid isle]

dont let your brain die haha, i hope im making sense here.

but you really dont need so much AI for everything you do

[young hollow]

Exactly That is My Fresh Context Technique

[languid isle]

😅

[ebon slate]

how can I export my chat and import it into new one?

[young hollow]

In Any Case No Matter What Your Brain Must Be Absolutely Active

[young hollow]

Simple Form Copy/Paste or Use Platform Export Option if Available

[ebon slate]

I've clicked on share and copied link

And how do I import into a new app or new chat

[young hollow]

There Are Probably Tools to Turn Text Into PDF/Markdown

[young hollow]

Link Might Won't Work

[languid isle]

this is my most recent query, that i did.

[young hollow]

Just Copy Paste The Response in Any Ai

[languid isle]

this was my tangent thought, where i didnt wanted to get stuck on.

[young hollow]

Additionally You Can Paste The Text Into Any Note App and Export .txt File

[ebon slate]

That's gonna be huge

[languid isle]

so i quickly got it to write me a bash script and i was done with the job within few minutes

[young hollow]

There are Probably Tools That Turn Text Into PDF/Markdown/.txt and Other File Formats

[young hollow]

.txt files takes lowest storage

I recommend markdown since is coding

Perplexity has PDF/Markdown/xdocs Export Option

[languid isle]

markdown is generally the best option here, also because it is globally well supported on all platforms

[young hollow]

Agreed

[rigid rivet]

Do you think it could be worth asking the model to summarise the whole chat first and thus reducing the size that way?

I suppose you lose a lot of important detail doing that though?

[young hollow]

Perfect

[ebon slate]

How about import option? Or just sending it into the chat normally?

[young hollow]

You Can Just Get The Summary of All Essential, Important Critical Data Without Losing Depth of Details/Data

[young hollow]

You Can Send Files Nowadays on Any Ai

That is Data Import option

[rigid rivet]

Yeah I hear you, make sure you instruct the model clearly on why you are asking it to complete that task. This way it will probably do a better job at keeping relevant detail.

[languid isle]

dont rely on AI with analyzing your data with lots of numbers and complex expressions, and if you still do then make sure to verify manually.

[young hollow]

It hasn't Even Been Around 7 Days Since I Started Using Perplexity Pro Properly

And I Learned All These Skills

I Have Bypassed All Toke Limits

[languid isle]

that's great, but make sure you're implementing what you learn

or you'll forget all this

[young hollow]

Certainly

[ebon slate]

You're doing the best job then

[young hollow]

The Use it or Lose it Rule

[languid isle]

yep :)

[young hollow]

For Input/Output Token Limits I Just Use Continuation Cycles

[ebon slate]

I just want my entire python learning journey be in a single page

But the page gets hanged just after 1 month

Where was it takes atleast 6 months

[young hollow]

Then You Need Claude Subscription

[languid isle]

not possible, im learning python currently as well because my university wants me to

I was more comfortable with using C, C++

[young hollow]

Try Kimi K2

It has 200k Tokens Context Limit

[languid isle]

dont learn python from AI

[young hollow]

I have heard is good at coding

[ebon slate]

I don't have enough budget to spend on ai subscriptions so 🙃

[young hollow]

Is Free...

[languid isle]

Kimi is free if im correct

[young hollow]

Daily 200-300 Queries in 24 Hour Rolling Window

[languid isle]

more than enough

[young hollow]

Or At Least 24 Hour Reset

Actually

If Someone Uses Maximum Everyday

The 30 Days Cost is 5-6$

Absolutely for Free

Kimi is Just Competiting with Deepseek, Qwen and Others

[ebon slate]

So kimi k2 is free

[young hollow]

Yup

And there is no Pro Subscription

You Need API if You Want More

[ebon slate]

I've heard Claude is so good at coding. What's the premium costs for that?

[young hollow]

Minimum 20$

or 2000₹

[ebon slate]

Any websites selling them for cheap? 🤣

[young hollow]

Idk tbh

There is Perplexity and T3 I Know

You need to research

But if is Cheaper There is Probably Constraints

[ebon slate]

Claude 4 is available on perplexity

[young hollow]

Yeah But

[ebon slate]

Not sure about perfomance though

[young hollow]

All Perplexity Models Have Token Limits

Context 32k

Input 8k-16k-32k

Output 4-8k

[ebon slate]

But I guess that's enough? Like 32k tokens will be around 10k lines of code approximately

[ebon slate]

What? Output just 8k?

[young hollow]

But You Will Run Out Eventually

[young hollow]

4k Average

You Can Use Continuation Cycles

Claude is 20$ Yes

But With More Token Limits

[ebon slate]

You mean we can ask the ai to continue from where it left last time?

[young hollow]

You Can Just Prompt (Use Maximum Input Token Limit Before Hitting Token Limit Ask Me for Continuous and I Will Provide Fresh Token Limit Keep Repeating Till All Data Collected)

Your Queries are Technically Unlimited Anyways

32k Context Limit is Your Biggest Problem Since You Want Everything in One Conversation

Are You A Student?

[ebon slate]

NO I'm a CS grad

[young hollow]

Ahhh

There is Gemini Pro for 1 Year Free For Indian Students

[timid finch]

How can I get the pro rank on discord?

[ebon slate]

Yes I've claimed it

I've even claimed perplexity

[young hollow]

Then You Can Use Gemini

It has Longer Context Window

[fathom stratus]

Oh that changes everything

[young hollow]

But Not Best Coding Model

[fathom stratus]

that's why a bit more text just fricking fails/attachments

[young hollow]

Just Use Continuation Cycles

[ebon slate]

Yeah it's not good, that's why I again choosed perplexity pro

[young hollow]

You Will Keep Getting Data in 4k Tokens Per Response Until Complete

[young hollow]

Core Perplexity Pro Models (Available 24/7):
✅ Sonar (Real-time data gathering)
✅ Claude Sonnet 4.0 (Analytical powerhouse)
✅ GPT-5 (Flagship reasoning)
✅ Gemini 2.5 Pro (Multimodal specialist)
✅ Claude Sonnet 4.0 Thinking (Enhanced reasoning)
✅ Grok 4 (Creative breakthrough)
✅ O3 (Mathematical precision)

Here Strengths of Each Available Model

[ebon slate]

For coding it's sonnet 4.0 i guess

[young hollow]

Sonnet
GPT5
Grok 4

Tier 1 Models for Coding

[ebon slate]

Gpt 5 just got released Yesterday. Not sure how is it so different from gpt 4.5

[young hollow]

Idk Neither

[ebon slate]

People on the internet/reddit arguing that gpt 5 is just released to keep up with other model updates without anything big changes made into it

[young hollow]

Yeah There is Big Community Backlash

[sweet zealot]

The actual model for the OpenAI subscribers has a thinking mode, which makes the GPT-5 amazing. The one we have in PPLX is the GPT-5 chat that does not have thinking, so it spits out. also the gpt-5 in PPLX has limited token and ....

[full sinew]

.

[ebon slate]

That's bad

[ebon slate]

Atleast PPLX Claude sonnet 4 is same as the one in the original Claude?

[sweet zealot]

I have the same question, but the PPLX isn't even clear about the API they provide from other models. The only thing that has been somewhat clear after two years of groundwork is that they limit input tokens to 32K (I think), and it's not the same as the actual model. For example, GPT-5 has 400K, but PPLX is 32K... So do the math.

[young hollow]

As I Say

[young hollow]

Perplexity Has API/Tempreture Tuning for All Models

[ebon slate]

Hmm. I bought chat gpt plus on telegram for 7$ and it is running since 1 year. But i forgot the microsoft accounts password and lost access but it was still running in my sister's mobile.

As I don't have the password i can't login to my mobile though

Would have been nice if someone sold chat gpt plus at such prices

🤣😂

[ebon slate]

That's sick. Less than 10% of tokens

[young hollow]

How Did it Run for 1 Year

[ebon slate]

I have no fkn idea

He said 1 month

[gentle siren]

whats the best model rn for coding?

[ebon slate]

And sold to me

[ebon slate]

Ofcourse gpt 5 if you can afford gpt plus

But I can't

So looking for alternatives

[gentle siren]

its comes with perplexity pro right

[sweet zealot]

yeah.... check your dm when you got free ❤️

[gentle siren]

isnt grok or claude better than gpt5 still?

[ebon slate]

Nope not in terms of coding atleast

[ebon slate]

It's restricted one. The one with so much less capabilities

[ebon slate]

I asked gpt and it says like this.

[young hollow]

Daym

Makes Sense Otherwise How They Can Give 7 Models at The Cost of 1

[ebon slate]

It's just renewing every month

[ebon slate]

Yes

But still

10% of the capability is so bad

Atleast 50? Would make sense

[young hollow]

Maye He Forgot to Turn of Auto Renewal

[ebon slate]

I really wish I could get his contact, I lost it.

[young hollow]

That Would Still Cost 70$

[ebon slate]

Id have known password

[young hollow]

But How is He Selling 20$ for 7$

[ebon slate]

Hmm. Ai has become a business now

So my question is. What's the big difference in use gpt 5 free model by open Ai and the premium one by PPLX?

[distant aurora]

https://cunnyx.com/theo/status/1954710820606407035

Theo - t3.gg (@theo)

The EU's "Chat Control" proposal is horrifying. There's no way to implement this safely. It will destroy private communications online entirely.
︀︀
︀︀If you're in the EU, please fight this.

💬 89 🔁 246 ❤️ 1.8K 👁️ 87.5K

CunnyX · Aug 11, 2025, 1:05 AM

W Poland https://cunnyx.com/7N7/status/1954711478495322518

. (@7N7)

@theo WE LOVE POLAND 🦅🦅🦅🦅🇵🇱🇵🇱🇵🇱🇵🇱

💬 9 🔁 1 ❤️ 182 👁️ 7.5K

CunnyX · Aug 11, 2025, 1:08 AM

[young hollow]

Perplexity GPT5 is Not Premium

It Just Has Unlimited Usage

[distant aurora]

More like 300-400 per day.

[ebon slate]

Okay. Unlike the gpt 5 on openai which runs out quick. We can use it as much as we want to

[serene sand]

#💬│general message

please don’t tag if I am not online..

#👑│pro-lounge message

[young hollow]

Exactly

[distant aurora]

What's stopping perplexity from offering thinking variant of gpt 5?

Or does the normal variant automatically decide whether or not to think?

[sweet zealot]

No support, no blog post about it, no announcement. They just rolled out something and left it there.

[distant aurora]

On chatgpt, it seemed to trigger thinking automatically when it felt like it.

[sweet zealot]

Ys

[young hollow]

Costings

[young hollow]

If GPT itself is Giving Limited Usage for GPT-5 Thinking

How Can Perplexity Possibly Give Unlimited Usage

[distant aurora]

Cheaper than O3 on api afaik.

[ebon slate]

Free version of open ai has token limit of 8-16k. Which is even worse compared to PPLX pro version.
So this isn't half bad

[young hollow]

Perplexity Probably Needs Official API Keys From OpenAi

[serene sand]

#💬│general message

[ebon slate]

Ofcourse

[young hollow]

As I Said Perplexity is Giving 7 Latest Advanced Models Costing 140$

Is Obvious There are Limitations

[distant aurora]

@dry kestrel is this supposed to mean that we can't control whether or not it thinks?

[young hollow]

Perplexity is More of A Search Ai Platform

Not A Full Fledged Performance Platform

There are Alternatives Like Poe or T3 But for Those as Well We Don't Know How Much Performance We Would Get

[ebon slate]

Correct

Gpt5 is integrated in search modules for PPLX

[young hollow]

Perplexity Still Does Offer Great Value for 20$

Airtel Users are Getting it for Free

[distant aurora]

It's still decent. Second to only open ai. Makes up for it with the variety of models offered. Especially the generosity with sonnet.

[young hollow]

Exactly

Perplexity has Designed and Implemented Perfect Profit Sustaining System

All Token Limits Can Be Bypassed With Techniques and Efforts

No Query Limitations

I Believe With Great Prompting Better Performance is Inevitable

And Tbh You Don't Need More Tokens for Better Performance

[distant aurora]

Nah

[young hollow]

I have done it

Takes Extra Skills Tho

[serene sand]

deep dive ➤ #💬│general message

[languid star]

🔒 Thousands of ChatGPT chats including personal confessions surfaced on Google

A short-lived ChatGPT sharing feature left over 4,500 conversations, many with sensitive details, publicly searchable on Google before OpenAI pulled the plug.

💬 “Discoverable” share links let search engines index chats without extra safeguards
🛡️ Exposed content included mental health struggles, trauma accounts, and workplace issues
📉 OpenAI has removed the feature and is working to de-index cached results

The leak is a stark reminder that AI chats can be as public as any web page once shared.

GPT-5 Thinking has the lowest recorded IQ of 57. It scored 70 on the offline IQ test.

[young hollow]

I Feel Like

Neither GPT-5 is Fake

Or They Have Made Major Optimization Mistakes

[lament lark]

Yo gng

[sweet zealot]

If the token was the only limit, yes, it would be bypassable, but they mess with the temperature and so the thinking process and logic are weaker than the actual model. (Based on my knowledge, I don't have any source like the PPLX that doesn't have a simple blog page for this kind of stuff D: )

[short wave]

hey, is it possible to change shortcuts in comet? everytime i want to write polish signs it opens me assistant (option+s, option+a) it's so annoying, i can't write

[twin grail]

Setting high temperatures doesn’t have anything to do with giving better output

[sweet zealot]

I know, my point is PPLX is around here for like 2 years... and they don't have a simple blog page for posting updates and... about their own stuff... they charge people.

[twin grail]

A blog about?

[sweet zealot]

https://cursor.com/blog/gpt-5

sign of respect for there customers??🤔🤔 and not have over 10K+ chat over simple gpt-5 API that they are giving? :DDD

[young hollow]

He Meant A Blog Where They Give Absolutely All Details Up-to-date

[twin grail]

Do he even checked out their X ?

We got update log

[sweet zealot]

My simple sites have an auto update that shows a changelog automatically..... it took me 10 min to set it up.

[twin grail]

About what changing, features

[young hollow]

Actually He Want to Know The Exact Details of All Available 7 Pro Models

[twin grail]

I mean about what tho

[young hollow]

Performance

[twin grail]

What details do he need to know

[sweet zealot]

hmmmm, DISCORD, X , instagram , how about tinder?

and they don't Push update to all at same.

[twin grail]

You don’t get it right, it’s 2 different field

And 2 different purposes

[young hollow]

He Want to Know and Me too Are The Performances of Perplexity Pro Models Same as Their Orginal Counterparts?

[twin grail]

Also for cusor

[twin grail]

They are the same API straight from providers

As i said again

Perplexity was meant for RESEARCHING

[young hollow]

But it Doesn't Make Sense How They Can Give 7 Models Worth 140$ Unlimited Just for 20$

[sweet zealot]

this is a simple thing that every single company have - " ** a blog page where they push information there in there [ own website ] no x - t - y - z social media.**

[twin grail]

They have different purposes in giving out information

For cusor, they are for devs

Who knows what’s happening

What to do

And what they actually for their new model

For perplexity, it was meant for research in the first place

They have their own RAG

To provide this

[young hollow]

So Perplexity Can't be Used for Other Purposes Like Coding or Writing?

[twin grail]

U can but it doesn’t do very well for complex project

[young hollow]

Ahhh I See

[twin grail]

It’s a fair trade

[young hollow]

So Not Same Performance Basically

[twin grail]

If u want access various models

For tasks

Perplexity is a good choice

[young hollow]

True The User Can Only Use One Model at A Time Anyways

[sweet zealot]

btw,

[twin grail]

If you prefer something can handle complex thing for your project

[short narwhal]

Not meant for it

[twin grail]

I suggest you buy their own plan straight from chatgpt

Or claude

Why u use a search purposed ai

For coding

[young hollow]

Tbh I Think Perplexity is More then Enough for Most Professional or At Least Individual Tasks

[twin grail]

Then u start to complain about it

It doesn’t make any sense

[green crystal]

not really

you get a dulled version of performance for everything

technically

even for search

chatgpt search has improved massively

[twin grail]

The purpose of it was not meant for other things except for researching.

Are we clear ?

[green crystal]

and with how good gpt-5 can code

canvas can literally make a website for you for every message lol

[young hollow]

Conclusion You Must Get A Official Subscription for Your Requirements

[green crystal]

its like labs in a way

[twin grail]

Yes exactly

[green crystal]

atm that is the case

[young hollow]

Perplexity Might Become Allrounder in Future

But Now for Now

[green crystal]

perplexity's main thing is very fast searches like google

[twin grail]

Just buy claude pro atp to code

[green crystal]

but thats it

[young hollow]

Yup is Just Basically for Search/Research as it Says

Tbh I Can't Complain

Is Free for Me

[twin grail]

It wasn’t meant for other things in the first place

U got an AI that can resolve most of the work for u

If u are a heavy load worker

[young hollow]

Is Still Amazing it Can Do Other Things as Well

[twin grail]

I suggest again you buy their official plan

Instead of using a search based AI

To code.

[young hollow]

Yeah Basically if You Make Money With Ai You Can Buy Ai

[twin grail]

It can do 60-70% of the normal works

I can’t complain about it

[serene sand]

all LLMs without RAG are heavy on hallucinations

[young hollow]

That's Still Incredible

[sweet zealot]

idk man - no one knows the real answer and it's a bad feeling, you know. I paid for this service and I don't know if I'm getting the same API with the same knowledge and power, or if it's just a label.... if the token limit was the only thing i'll be ok with it.

[twin grail]

20$ are still a good trade

[twin grail]

We told you a million times, the API is the same

[young hollow]

20$ for 7 Latest Advanced Models

[young hollow]

And For Airtel Users is Essentially Free

[sweet zealot]

allright xD

[twin grail]

It just the RAG

Just buy chatgpt sub at this point

[young hollow]

What if They Make Perplexity Allrounder Not Just for Research

[twin grail]

If you are afraid of not getting the same api

[twin grail]

They will need to increase the cost

[young hollow]

They Can Introduce New Tiers

[twin grail]

Most likely

We will have to pay more

[sweet zealot]

If that's so, then damn.

All these questions are because of the GPT-5 they released, and it's so much different from the real one — so, yeah.

If they put out a GPT-5 mini or nano, it would be better than the non-thinking one we have now.

[twin grail]

And they have to change their RAG

[young hollow]

No Like Different Limitations for Different Tiers

[twin grail]

Dude, it’s about the RAG

And it’s the non thinking version

What do you mean ?

[sweet zealot]

yeah ik - nano and mini have thinking mode as i know.

[twin grail]

If you really think they can add another model then announce it another way ?

[serene sand]

many of us have given you good information, yet you still persist in asking the same questions over and over.

please stop spamming

[young hollow]

@twin grail Do You Think Someone With Very High Skills Can Get More Tasks Completions then Baseline?

[sweet zealot]

An extra $10 or $20 can't hurt, but going from $20 to $200 is a huge jump 😄

They should offer a middle subscription, like a $50 plan.

[young hollow]

I think it will happen in future

Different Tiers With Different Limitations

[sweet zealot]

good information - all not the same, so the question remains, bro ...

[twin grail]

Uhmm it really depends on you

I use perplexity mostly everyday

For my tasks

[young hollow]

And You are A Working Professional Right?

[sweet zealot]

Thanks for the information

[twin grail]

About to

[young hollow]

Final Conclusion Perplexity is Sufficient for Individual/Personal Use

But May Not Be Sufficient for Specific Professional Use

[twin grail]

It was meant for search

[young hollow]

Bottom Line Buy Official Subscriptions of Your Needs

[twin grail]

Yes u can use it for other purposes

But again it wasn’t meant for that

[young hollow]

That's Still Incredible

[wanton flare]

.

[young hollow]

I Feel Like One Day Perplexity Will Become A Allrounder

[twin grail]

Indeed

Just buy chatgpt at this point

[young hollow]

Different Tiers for Different Needs

[twin grail]

Or claude

[young hollow]

I Don't Need it Lol

Im Just Using Perplexity for Up Skilling

And Learning

[serene sand]

do tell/ask

[young hollow]

Any of You Guys Know Any Perplexity Alternatives? With Multimodal Support and All Task Doable?

[sweet zealot]

ninja tools something like that

[twin grail]

Blackbox AI

[sweet zealot]

but idk about there api and ....

[young hollow]

Gotcha!

Have You Tried it?

[twin grail]

Ye

[young hollow]

Is Great?

[twin grail]

I bought their pro plan also

Uhh

Not really

My experience kinda bad

U can try it

U get very limited uses

[young hollow]

I Trust Your Words so Nvm

[sweet zealot]

8$/m is so cheap and good - are they providing good API?

[young hollow]

I Guess Openrouter is a option

But need to pay for api keys

[young hollow]

Probably Not

[twin grail]

Nothing’s cheap come with good quality

[young hollow]

Exactly

I don't think there is any platform that gives multimodal quality

Without great cost

That's Not Logical

[twin grail]

I bought their service

even for business plan

[young hollow]

Yup

GPT is Best Option for Professionals

A Allrounder Generalist

Claude for Coders

[twin grail]

They don't refill you

So don't waste ur time buying it

[young hollow]

Perplexity is Best for Now

[twin grail]

U'll get a very limited access

[sweet zealot]

damn

[young hollow]

Anything

Cheap

[twin grail]

it's for 15$ a month

[young hollow]

Won't Give Quality

[twin grail]

and it's even not as good as 20$ plan

just 5 more u got everything u need

[young hollow]

For Now Official Models are Best

For Required Needs

[twin grail]

it's not really the case

Perplexity RAG reduce the hallucination

without RAG

the ai tends to hallucinate itself

[young hollow]

Ahhh

They Will Probably Figure Out Solutions in Future

[rose crag]

Hey

[serene sand]

it’s just more “creative” 🙂

[rose crag]

I have pro how do I get pro discord from my pro subscription

[twin grail]

checkout pin !

@rose crag

[rose crag]

Where do I find that

[twin grail]

jump to the message i pinged you

[serene sand]

#💬│general message

[twin grail]

th44

[rose crag]

I think I got it

[twin grail]

welcome onboard

[serene sand]

https://x.com/karpathy/status/1733299213503787018

RAG kills dreams, so to speak

Andrej Karpathy (@karpathy)

# On the "hallucination problem"

I always struggle a bit with I'm asked about the "hallucination problem" in LLMs. Because, in some sense, hallucination is all LLMs do. They are dream machines.

We direct their dreams with prompts. The prompts start the dream, and based on the

X · Dec 9, 2023, 1:35 AM

[zenith lantern]

Hey guys I have a question. I am a pro subscriber. I noticed in comet when I hover over search it says unlimited access for subscribers, for research it says extended and for labs says 50 remaining this months. It is stating the usage limits but Is pro search unlimited? I thought it is like 300 per day

[full snow]

Guys

Exam is finished

Today was my last

[short narwhal]

Woo congrats 🎉

[short narwhal]

300+ = basically unlimited, you shouldn’t reach that ever in a day

[zenith lantern]

I thought so. Btw is the number changing based on models? Like using sonar all time is not as heavy as O3

[short narwhal]

Nope. Same limit for all

[zenith lantern]

Ok thanks appreciate it

[serene sand]

300/day means once every 4.8 minutes if you use it for 24 hours, so it’s basically “unlimited”

[twin grail]

it's time to spam hi

hehehe

[zenith lantern]

Wow I didnt do the math that’s surprising

[covert bluff]

anyone need comet invite

https://perplexity.ai/browser/claim/VTJCOTYF0Y

🙂

enjoy lol, I had one and didnt really have a use for it since I already use comet

[serene sand]

@whole pecan

[lusty swift]

Can we input system prompts?

@serene sand

Into Perplexity

[winged matrix]

Hey guys anyone wanna build a website or any kind of project together? Wanna brainstorm 👀

[covert bluff]

Lol I JUST completed my discord bot and its web dashboard, would u like to test it?

Im deploying it right now on the web

[winged matrix]

Damnn would love to join you

[distant aurora]

@twin grail #💬│general message

[winged matrix]

Learn and earn 🥱 alwayss

[covert bluff]

Thank you! I will send u the link to the dashboard in a sec, once its deployed

right here

[winged matrix]

Could you please dm me directly

If possible

[covert bluff]

sure

[distant aurora]

Unless you are on a 350mcq speed run.

[winged matrix]

Thanks

[lusty swift]

can’t find this in mobile version 😦

[serene sand]

web app only

[distant aurora]

It's only available in web version.

[lusty swift]

what if I input system prompt on web, but put my queries in mobile

[serene sand]

it follows

[lusty swift]

will system prompt exist even on mobile

aight

thanks bro

recarveting thanks for info too

[serene sand]

my pleasure

https://discord.com/channels/1047197230748151888/1124060874785685554

deep dive ➤ #💬│general message

[lusty swift]

useless channels

but thanks goat

[serene sand]

try this prompt #1199176325051723837 message

[lusty swift]

I’m trying the denis prompt

Which fine tuned chatgpt for me

Yeah works as charm

What else I can try that people failed on chatgpt 5 hmm

[distant aurora]

There's a square edfb of 6cm inside a right angled triangle abc (b being the right angle) with hypotenuse ac = 20cm passing through d. Ab passes through e. Find length of ab

[lusty swift]

I'll answer as a world‑famous Euclidean geometry PhD specializing in right‑triangle configurations with the 2020 Fields Medal in geometric analysis.

TL;DR: AB = 12 cm, assuming the standard configuration where a 6 cm square is inscribed with one side on AB and one vertex touching the hypotenuse AC.

Step-by-step reasoning with the key configuration clarified

• Interpretation of the figure: Triangle ABC is right‑angled at B, with hypotenuse AC = 20 cm. A square EDFB of side 6 cm is placed so that side FB lies on AB and vertex D lies on AC (“hypotenuse passes through D”), and AB passes through E (so E is on AB as well). This is the classic “inscribed square with one side on a leg and one vertex on the hypotenuse” setup.

[distant aurora]

Wrong

[lusty swift]

Bruh answer is so long

[distant aurora]

https://www.perplexity.ai/search/there-s-a-square-edfb-of-6cm-i-9FiSKqWDTma6WC7NOFVx5A#0

[lusty swift]

Fact check TL dr

[distant aurora]

17.84 or 9.04

[whole pecan]

already claimed

can anyone sent me a link via dm?

[covert bluff]

damn, im sorry lol I asked but no one replied so I just posted it in general

[serene sand]

comet invites —> https://discord.com/channels/1047197230748151888/1111786888626438245

[distant aurora]

Gpt 5 solved this natively. Without even enabling reasoning switch.

[lusty swift]

This is wrong

[distant aurora]

Lol.

[distant aurora]

Try any other reasoning model.

You'll get same answer.

[lusty swift]

Ok

[whole pecan]

huh what is there?

[distant aurora]

The diagram it generated is wrong but the answer is correct.

[lusty swift]

Same answer lol

With different reasoning model

[distant aurora]

https://chatgpt.com/share/6899caf8-aac0-8013-a3c5-87358c604dff

[lusty swift]

I think yours is wrong

[lusty swift]

Bro have you even read

Yours saying

The hypotenuse length gives x^2+y^2=20^2=400. Substitute y and solve:

Mine is saying that too

And it actually pulls the proportion law

Of the triangle properties

[distant aurora]

Okay. Do one thing. Paste my solution into your ai chat. Then ask your ai if my solution is correct.

[sweet zealot]

We should upvote this so the mods and the main PPLX admins can see it (if they exist in the chat, of course :DD).

https://discord.com/channels/1047197230748151888/1403076559207796927

[lusty swift]

It says mine is correct anyway

lol

But yours too

[fast meteor]

see these crazy stepshttps://www.perplexity.ai/search/there-s-a-square-edfb-of-6cm-i-G6dG2BA1RIOKKVLLc3YdcQ

[lusty swift]

Answers depends

On figure configuration

[lusty swift]

400.07

[tame current]

Hi

[lusty swift]

Not even mathematically correct

lol

Mine is more math oriented cuz it gave precise answers

But at the end configuration matters

[fast meteor]

it used numerical approximation

[lusty swift]

As a physics major that doesn’t satisfy me

[plucky spindle]

Hmm

[lusty swift]

Actually it would satisfy cuz I love approx

But for math it stinks

[meager rover]

https://www.perplexity.ai/page/musk-s-xai-makes-grok-4-free-f-a0YaHLP0SriH4eDrtL1Eug

[cold kernel]

Hey guys!

[tranquil orchid]

@cold kernel

Adam.escobar said:

@spare marsh hey how do I get the pro role?

Please avoid @mentioning members of the Perplexity Team

[compact gust]

[serene sand]

#💬│general message

[lusty swift]

Bro pinged team

😭😭😭

[cold kernel]

Then what they r use of if we can't ping them?

[meager rover]

damn assistant mode is accurately writing down mock theta functions in desmos

[cold kernel]

Ngl but comet browser is cool af 💀 I can use it to trade crypto hell yeah

[lusty swift]

what agent mode

u mean in GPT?

[meager rover]

Sorry assistant in comet

[rain crystal]

agent lol

[meager rover]

Whatever

[frail crane]

Hey Guys, did you also notice this or have seen this.

I am a perplexity pro user and I use mostly research mode, on first to third messege in the same thread, perplexity research the topic more and gather good number of sources but as I want to go deep and have sent 4th message, and so on then it just give me normal and short answer within 45 seconds with either 0 sources of very less number of sources.

Have you also ?

[meager rover]

I think after you’re done your initial researches use normal chat mode

Because context fills up fast

And you’re asking questions for what’s been indexed

Already

[serene sand]

James answers questions in https://discord.com/channels/1047197230748151888/1161802929053909012

[frail crane]

Bro, But it let me select research mode, and it's always even if it's my first thread.
I think it's a context window issue.

[cold kernel]

Ahh ok

[meager rover]

I know just select the normal chat mode after your inital searches.

[frail crane]

So, I should use Research mode for only next two or three messages.
Because after that, it'll not respond with good research because of context window or anything else.

[meager rover]

That's my guess yes

[frail crane]

Got it, Search mode.

[young hollow]

Is Kimi K2 as Power as Sonnet 4

[meager rover]

what aspects

[young hollow]

Like Overall

[young hollow]

There is 32k Context Window

Tokens

[meager rover]

Kimi K2 demonstrates top-tier performance on several key coding benchmarks:
• SWE‑bench Verified: 65.8% single‑attempt accuracy
• LiveCodeBench: 53.7%
• AceBench (English): 80.1%
These scores place it alongside—or even ahead of—Claude Sonnet 4 on many coding and tool-use metrics

[frail crane]

Oh that's the reason.
So, I guess if I have any topic which I know I will go in deep with follow ups, so I should go with Chatgpt or gemini instead perplexity.

[meager rover]

Ye

[young hollow]

32k tokens ≈ 128,000 characters ≈ 24,000–26,000 words.

[young hollow]

And its Free Like DeepSeek Amazing