#š¬āgeneral
1 messages Ā· Page 181 of 1
1 token is 4 characters
1194 words
Including spaces I think.
That's a lot iam not sure if even lab can handle that but that means you need to do it Section by section.
but at the moment it will re-writte it, in the middle it will forget everything
1200 words
^^ with files the context window is 32k
That's like 1 and a half page.
Put it here to see how many tokens it is
A page is approx 750 words in my experience.
what model do i choose
4o
Tokens
1,582
Characters
7434
Yeah that's pretty short
2 pages
so why does it still forget anything if the memory is 10% filled
Idk if you saw this yet haha
Did you put just the intro or the whole book?
I'm telling you, he ain't real.
Sam is Dennis
a chapter (the intro) that i'm trying to be re-writted in a good way so i don't have to find the information acrosse the entier thing
because the entier course will be like this with way bigger texts
@cobalt vale is Gemini flash thinking unlimited for free tier?
Isnāt the rest of the context used for the search results
32k is for the whole chat including the files afaik.
and i don't want to adjust it myself, it's way too long to copy/past/re-writte/etc... everything
i thought ai were able to do this and follow instructions as basics as "don't touch the texte just move portions of it by theme and merge theme by theme in different parts"
Pre prompt is separate?
no idea, almost certainly no
I guess your course is maybe 100k tokens?
No. I meant it's included in the 32k even if files usually have a size based limit.
And non thinking?
i don't want it to re-writte the entier course. just chapter by chapter and keep the global course structure.
just re-writte each chapter structure
2.5 flash is thinking
Yeah for the previous messages, then the rest is saved for search results I think
It's just adaptive
I just wish they explained better
Has a switch in ai studio. But not in Gemini I guess from what you are saying.
So is any model unlimited on Gemini rn? They got rid of the 2.0 models.
Pro is limited to 25 per day iirc
so... my best chance is to try with o4 mini?
Your best chance is Gemini.
I been saying
Don't bother with any other models.
so Gemini 2.5 pro?
@timber niche no you said Claude 4 Sonnet
o4 mini is smart but long context recall it's not the best
That or flash. Use it natively.
Oh on ppxl lol
Or through ai studio.
Only pplx
32k won't fit your book. And splitting the chapters would be too much hassle for you.
but i don't want to spend an other subscription T.T so perplexity was the worst choice for my use?
If you have to process long docs you'll need to process it though gemini on ai studio
o4 mini on chatgpt is pretty good tbh
To get the max tokens
Well to a point that is
32k for plus. 128k max.
He has a big book.
i'm beyond frustrated at this point.
everyone say every model is the best for everything.
i'm totaly lost here.
As someone who uploaded a book recently to make a 500 question MCQ bank, I'll recommend google for this specific use case.
It's 200k
and does google follow the instruction without doing its own thing / hallucinate / lost its memory randomly?
2.5 Flash thinking is not available on Gemini app actually
Sorry, you were right.
No way O3 has lower context than o4 mini.
Didn't expect that.
why is it called mini while it's better? it's SO confusing!!
not better
Cheaper. Less resource consuming, more cost effective.
uh
Bio weapon vs nuke scenario.
Ordinary user won't notice much difference, don't worry about it.
what
Between O3 and o4 mini high.
i'm 100% lost right now
so it's the same models across different platforms/websites but with different numbers????
Accuracy is nearly same.
Well itās certainly noticeable, the quality is about the same for most average users tho
O3 just talks more.
like gemini with google, gemini with perplexity
GPT with openAI, GPT with perplexity
Fair
Gemini with google, and perplexity shouldnāt be super different
He needs context limit for big book.
but i'd read everywhere that gemini struggle to stay on track and hallucinate a lot
What's the difference between this and learn lm
Not sure
Iāve personally never used either, but my friends gf said it āsucked assā after I suggested it
it's like everytone say different things about every plateform/models/good and bads things
You should be fine with theoretical stuff.
Lucky girl.
Every model struggles with something
And most of what youāre hearing is opinions
yeah, yeah
i was told ChatGPT website was good for this
i'd cancelled my subscription because it sucked so much at keep on tracks. hallucinate all the time
You cancled ur gpt sub?
Every model will hallucinate.
like it removed and added things to the course each time
"In ChatGPT and the API, o3 and o4āmini both have a 200k token context window, and max output tokens of 100k. "
Welcome to chatgpt naming lol
it mixed the course + its old memories/training and random web sources.
ended in a super ugly chimeria course.
you don't want to see this.
again my head hurt rn.
I should try gemino or noteboot or earlier i was told to use claud
yeah everything is so perfect
everything is so amazing
yeah? ask Hero, he told me it was the best
I told to use claude for best search results in pplx
Not for uploaded stuff
Just to be perfectly clear
A model perfectly editing an entire large document without much supervision is quite difficult
Doesnāt antropic use pplx for search
and it suck ass each time
still randomly merge random sources
Not sure but for now i just told them to use claude 4 sonnet thinking to get the up-to-date results which is put together well
You're still gonna have to use something like canvas in either chatgpt or gemini
Now their requirement is different
To selectively pick parts and tell the model what to do with those parts
@rugged idol i would suggest you to learn about the models first
And then make your decision
so
Or donāt
Just try shit
Til it does what want
And be pissy and hate life eh
I don't think they have time for that
No, hero is right.
i need to buy a subscription for each thing?
Eh
First understand your requirement
If they donāt have time to bruteforcr trying models they donāt have time to learn about them all
and spend like thousands of ⬠to find something that finaly work?
What you need
I'm not sure what the actual deadline is, but it seems like it's for a course they need to provide pretty soon
Gemini, and notebookLLM are both free for you to try
What you will need ro achieve and expectations
Find the best tools to do that
Aka llms or ai
Which model is best for that use case
Sadge
Gemini canvas is free
Overview is enough
Bro calm down. Head to Google ai studio, or Gemini, try it and let us know how it goes. Google has the best memory but open ai and Deepseek are better when it comes to engineering or math problems.
Try some editing there to get a feel for it
What sort of content are they working with anyways?
ok so i need to do google studio Notebook geminiLLM ai sadge ???
Context isn't that large there but enough to see if you can work with that type of editing and decide on what to subscribe to
You don't need to pay for anything till you're concrete on it meeting your needs
seriously i'm totaly lost you ask me to try 657135873852741528 different things
Bro is just lost
you can ask ai to give you recomendation on which ai you may find useful.
yes i am...
Goodluck soldier
Guys, let him try google.
whgat?
In any case, forget perplexity for large context editing
That's already out the window

Google āGemini AI studioā, try it. Itās completely free
Leave this server, go open ai studio
Upload your document
Ask your stuff
It will do it
@rugged idol
So now it's either Chatgpt or Gemini
I never mocked you.
i have NO IDEA when it sarcastic or serious. no clue.
if you dont want any headache use VOE 3 it will definatly do what you want and its free i think
Dawg š
Someone send him the link
i see memes and all i have no idea what to do
it's like random things al allong i don't know
oh and now an other thing to try
gemini
@rugged idol https://aistudio.google.com/
Nah most are serious here
But try the canvas writing tools
There's free options to see if you can work with it
oh and now canva
What model were you using?
Large context editing without supervision is very difficult for now
yeah i'm going to try 6138763571357163574158 shit thing at the same time
Don't come back before you try ai studio. Seriously.
I don't think you will find a single model out there that will retain most of the structure of the document while doing specific edits
@rugged idol
@rugged idol you're just annoying now man, we are trying to help you
while trying to keep the track of who talk about this or than
if only this server had a non-pro vc š„
imagin 5 person talking to you while saying different things
If you don't want help then you can leave for other sources
we all saying the exact same thing tho š
its usually dead so iam forced to chat here
Literally
Vote for ai studio
yeah? i'm told to try like at least 5 differents things
Nope
please name those 5 things
Is Apple in the works on buying perplexity?
š
Samsung might
samsung is betting higher rn but who knows mark usually wins
Nope
Maybe
True
i'd love for apple to get pplx š
Gemini is mid beyond mid
Perplexity is mainly colabing with moto rn.
wild take
Google sucks bro
you cant use the word better when apple doesnt have any
-Gemini
-Google Studio
-O3 mini
-chatgpt
-Gemini Canva
-Notebook LLM
you're looking too far back in chat
I meant better than what perplexity gives in moto.
Whatās the difference between o3 and ChatGPT?
Canvas is a feature of gemini
i don't knoiw
we all gave up on everything other than ai studio...
i do'nt know either
o3 is a model on chatgpt
š
someone talked about something in chatgpt i don't remember what
Ik.
glhf ā¤ļø
I like perplexity slightly better ngl.
Not for native assistant on device
Native must be nice ngl
Google is the only saviour in this paid ai masterrace
To have answers in your fingertips is helpful
Offers most shit for free
Google lens is invaluable to me
More than ai sometimes
Easiest free option to try is Gemini
Set model to Gemini 2.5 pro and click canvas
You'll see a large input and you can put a chapter there
See how it works
Yes you still have to do some stuff manually but that's just how it is with AI for now
you forgot voe 3
Not the consumer side, only the dev side
....
Yeet
He's already confused as it is, stop trolling lol
thanks...
and sorry it's so hard to keep up with like a constant flow of messages
Ai ain't for you.
ok but jokes aside i think canvas is the most trustable here
The free version is limited to 25 per day and then you can use flash version which is lower in accuracy but you can still work with it
its not troll canvas is literally made for text so try google canvas
well
From there on out things get a bit more specific for models
Bro just misspelled veo
Voe might be a different model.
hey the man wasnt listening in the first place i thought it would go overhead just like that
sh*t it exist
video generation lmao
Anyway
pleast stop i already struggle to follow
Back to regular chat please
sorry dyslexic
i mean TECHNICALLY you can ask it to generate avideo without long term explanation of short term phrases
š
jk
so it's the 2.5 pro (preview) that i use?
Yeah
They said all the steps
And canvas right at bottom
Once that runs out you can switch to the flash version to experiment more
okay done!
and thene, i put a prompt with my course and the instructions?
You'll see something along the line of ask Gemini and then you can give it instructions
You can also highlight a piece of text and you'll see that you can add a prompt or it'll say you can ask Gemini something
like, i can copy my instruction from perplexity and use theme here? or does it work differently than chatgpt/sonet/etc?
The way models interpret prompts will always be a bit different between all of them
So you'll have to experiment a bit
ehm... but how does Canvas work?
If it's just a simple prompt and nothing complex setup then it should just work out fine
It's like Microsoft word but very AI focused
like i can give it a picture of my course? would it work?
Like the name says, it's a canvas for many things you want to do. Like writing documents, editing stuff, coding
yeah but if it remove something, can it put it back?
let's say it remove a definition, can it be restaured?
I haven't tested that out on Gemini yet
Chatgpt works very well with images though
Yeah, there's an undo button there
So you don't have to copy paste everything again if an edit is incorrect
Omg what an amazing question and what an intellectual you are for asking it
okay it totaly re-wrotte the course. i don't even recognize it at all.
Bro just get a human copywriter
or a functional AI?
Is it this complicated? i don't get why it's Einstein for complicated tasks but Dorie for simple ones
Im serious if you don't want any change in phrasing and context and everything
What you're asking isn't just the use case for ai
I actually find this sometimes an issue , so i include " critsize me logically and technically so we can have the best output "
Cause i might have a flaw in my request and it would be just " yeah sure"
Even when you selected a part of it and used the "ask Gemini" field?
as if the limits weren't lowered yesterday in the first place
My love for qwen each day š
Google solving the problem they created
And patting themselves in the back for it
So apple of them
How so
Free, no limits, strong reasoning, has a 1m context model too. Are probably working on their deep research to make it much better.
Reasoning model has 130k context.
Happens the moment Google is seen as the best for one second
Pretty bad habit tbh
even better than deepseek
Deepseek still gets server busy issues.
actually nvm
Saw someone rant on reddit few days ago.
yea thats an issue but thats why u have perplexity
Free through you.com?
Qwen Chat offers comprehensive functionality spanning chatbot, image and video understanding, image generation, document processing, web search integration, tool utilization, and artifacts.
I'm specifically talking about free options. Besides, the R1 on perplexity is based on the old version.
Deepseek on perplexity is pretty nerfed
Qwen is natively free.
Thank you
No it's fine, just outdated wrt the new update.
New update pushes it to O3 level.
Old model was o1 level.
Idk why qwen defaults to Spanish for me
r1?
Yeah. R1 1776 is perplexity's custom R1.
Wait.., o1 in PPLX or somewhere else?
Yes
R1 1776 on perplexity ~ o1 level
R1 0528?
What's this
Thinking token limit.
pplx did have o1 for a moment
It can spend 36000 tokens in thinking alone.
Which model? Qwen?
is this a daily limit or a response limit?
More than perplexity's whole context limit.
Yes.
Cheers
i asked it to just extract the text.
it failed
qwen is alibaba right?
Yep
they have been cooking quite a lot they even made wan2,1
What's wan 2.1 ?
open source video model
video gen by ali
Oh.
i think its doing better than klin
Didn't try yet.
perplexity not working now?
claude thinking
Works fine
No Opus or O3?
Papa Lex City Pro
iam on latest update and no
What's the site that benchmarks llms
Private, domain-specific benchmarks in legal, tax, and finance.
You're the best!
does this show image gen and their variants?
why do u have upside down ? mark
I don't think so.
Spanish in Australia
Maybe it's just a Spanish thing idk lol
Watch memories think im Spanish
ok gemini isn't doing anything i ask it
Get a human:)
i asked it 5 times to not add anything, it still re-writte it
Yeah. Ai is a bitch.
My bitch
...
What havoc did you cause
how you guys get the IA to not ignore the instruction randomly??
.........
you told me like at least 5 times that gemini canvas was perfect and now you make fun of me because it doesn't work
Because nothing will satisfy you
What you need isn't possible with ai
We're not even sure what you're looking for
so why you told me to try many things if it's so obvious why i want to do isn't possible?
Why are you being so self righteous?
here i'm trying Gemini pro (trial)
You always just blame and not say the exact issue or your requirement for it to do
So that they don't end up clueless or dumb when they grow up.
Rip
"Sorry, something went wrong"
i was doing a project, now i don't have time to send it š¢
is this ur api?
What an ass lol
No. T3 chat.
Waste of multiple prompts outta 1500
I'm thinking of trying open router.
i dont remember his name the blonde dickhead
Bro be like you're a waste of energy utilised running your compute
it was something like landlord
i already said it
-i have course with big chapters
-each capter has many notions but the essentials are scattered in the course.
The said course said a "story" but all the notions are scattered inside.
-there is a tone of technical words that i don't understand and i want to have a short definition next to each word
In other words, a readable course, not someone telling why he is so great at something and telling me his life.
but the issue is when i ask an ai to re-writte the course (or when it does it by itself) it will remove big parts of it without me asking. like an important definition, important notion, etc...
sometime entier paragraph or sub chapter
Open the course
yea becuase u are asking to rewrite
Look up notions on google
how about you start with ''remove'
Don't use ai
It will summarise your stuff and rewrite
Based on it's model
That's how ai works
You essentially will have to do your reading
In the course itself
Don't expect ai to rewrite as you call paraphrasing
Good day
i ask it to like solve a puzzle where notions are scattered here and there in diffrent parts of the chapters
Exactly
Literally
use ai as a tool, not as an employee š
slave*
Let's not get into there
What one personally deems as a notion
How does the ai know
ah yes, sorry
employs get paid
the issue is i'm learning, i can't re-writte a course i don't even understand yet because it's wrote with ass
Unless you yourself make an ai of yourself that is you in computer form
Is anyone having issues with chat sync? In the app it retains all the history, but on the PC it simply forgets everything that was said after a while. (literally erases the prompt and goes back to the old one)
No one is asking you to rewrite
Just open and read your course
Like a person before 2020
i don't know how to make you understand, that it's so confusing the way it's wrote
imagine you have notion at the end to understand the begening
Pplx is having issues for multiple people rn
Maybe because of it
iam so confused when we started u said u understand the short term phrases such as AI CPU so you dont like them being there and now you are saying its hard to understand yet you want all of these to be done without rewritting?? do you perhaps want noting and listing?
i know what CPU is, it was an example if you refer about what i wrotte like way earlier
are u allowed to send the book you're using
honestly
or tell us what it's called
i may finish it before the ai ''rewrites'' it
I understand, I just wanted to know how to proceed, because on the cell phone the history is retained, outside of it it is not. I don't know if I should keep the tab or change to a completely new one, because there are more than months of memory progress in this current one.
Just wait, and hope its an temp issue š¤·āāļø
Open a ticket and provide your details in #šāfeedback-general
It's good if it's mostly unrelated waste.
Would be so sad to lose it
Easier to clean up.
I mean, it's easier to continue in a single tab than to explain everything from scratch again and again.
after like 40 prompts in a single thread pplx be using like 3 gigs of ram
alone
meanwhile i have a different space and a thread for each conversation
Bro living rich
Bro crashing pplx
missing out
dont ask
Holy
It's just folders with shared context and custom instructions.
idk, i'd always tryed to have a "folder'" (space in perplexity or project in chatgpt)
Roleplay
I don't do separate topics, I keep them in the same topic and just improve them, I do this with each tab I use, and from time to time I export it to notebooklm to help me review everything.
but i don't know if perplexity can find informations in different threads from the same space?
Ikr
Was going to ask about the same thing
I only started using pplx last week tbf
no, it will not.
what the point of spaces then?
wdym
Theoretically it's supposed to.
Custom instructions
being organized
š¤·āāļø
u used to be able to force a model in spaces, but not anymore?
ig it's not required anymore soo
hm... could i put my important instructions in the one inside a space description? would it follow it without lost it?
Is there any way to export information from one thread to another while retaining memory?
no
Nope
its an ''opensource'' role player i attach the chatacter via pdf and its job is to decide which one i wana talk to
Export the thread as pdf and upload it in the new thread
hm... i try to see why perplexity is better than anything else then ._. could someone enlight me?
Well, I have the PDF of everything and it's about 1200 pages, would it handle that well?
No. 1200 pages. That is unlikely to fit in 32000 tokens.
its all about personal opinion it need it for the space futures(if u cant tell) some others need it for page some for lab we arent all the same
- i also use it for news
So I just have to sit down and pray that I don't lose the 2-day history of the tab I'm in.

1200 pages
save it a page?
bro is saying hi every min
Lol
ig he might be using ''manual'' api š
Writing books, I am a ghost writer and proofreader, it is my primary source of income.
Nice
hahah
š
This current prompt is 1200 pages of a book, I will publish it at the end of this month.
title?
i think i misenderstood how basicly ai work
it was advertised like an assistant that does things from A to Z but it's more like an advanced tool (search engine, generate, etc...)
it's not something you can ask to work on something that already exist?
I can't provide information due to an NDA, I'm paid to write, my own is being done too heh
my man is here printing books and we have issues with reading a course
it most certainly can
its alright can i dm u tho i need some advice if u are okay with it
when prompted right
As long as your books aren't just ai generated you're cool
or given something realistic to go off
i just wonder
is the fact that it's full french that confuse it so much?
can anyone give me a good prompt for PPL labs
Nope
or my prompts? i did theme in french too
thats an ai agent not ai
what's the difference?
ai agent is closer to an assistance than ai is
I have currently made about 42 books since November 2023, in 3 languages, Portuguese, English and Spanish, most take about 1-2 weeks, some done at the same time, others take a little longer. I work writing the Raw material, I use Gemini/pplx and some others to help me, something that would take 3 months I do in less than 2 weeks.
also ai can still asssist you just because it cannot assist you in anything doesnt mean they cannot assist
like they cannot assist you in a crime duh
So ai writes your stuff?
Or do you mean the ones you've proofread
i think he still needs writer skill to check and verify
Tell me what I should change, improve, then I'll redo it and send it again for review, it works like a beta reader.
That's alright
basically a free supervisor that wont get mad
hahaha
43 is still impressive congratulations
its not that creative sadly
not yet at least
I don't care if it's stem.
I just hate the concept of ai creativity replacing humans
Then, obviously, I pass it on to human review, I receive around 60/$ for every 100 pages, I pay around 20/$ for beta reading and the rest I use to invest in other tools, voice assistant/images/video/review etc...
Stem isn't creativity so.
its prob gonna happen tho
Sounds great
not much we can do
Will stay a hater
It's sustainable, it doesn't cost me much and I can do it while doing my own work, but losing a tab that has over 1200 pages of prompts frustrates me.
how do you like perplexity since you have been using it alot
Ai doesn't love you either, don't worry.

It's essentially hurting your earnings
Pplx better help you
I mean, it's good, I've recommended it to most people, but it's not the best by far, but it's definitely the one I use the most.
Ai don't have feelings bud
i mean with 1200 pages idk if perplex wouldnt geta hedache
Prompting to get the desired results from an ai image generator takes some effort too.
I pay for the pro plan for myself, my wife and 2 of my assistants, I renew every year, but I also pay for Google One and a few others.
Any other ai services you use?
Like chat gpt pro etc
Because gpt is the goat for creative stuff
Gemini/ImageFX/MidJourney/GPTPro/Cloude and Grok.
Sheesh the entire suite
You paid for all of this?
š
I mean, Cloude is better than GPT Pro.
The money invested in this doesn't hurt how much I earn using all of this.
Nice
I mean, could I just use one? Yes, but the flexibility of having 5 AI assistants is much more practical than just one.
For me ai is more like my life assistant.
Yes.
Like a secondary parent.
Especially since they all keep one upping each other
And you have the flexibility of everything
I used to think so, but after seeing the difference in productivity, I was able to monetize almost all of my creative work and some extra things just with that, even more so receiving in any currency and distributing in any language.
welcome to the team
hi, i just hopped in. could you explain what is it that you do? something related to books?
You could start by saying something like: I recommended my actual assistantānowadays he earns around $1,200 to $1,500 a month selling fantasy books to the Chinese market.
Dang
I have no idea what you meant.
I usually sell ghostwriting services, do revisions, write books, analyses, and other related work. I was just sharing a bit of my experience because PPLX lost about two days of progress on a tab I was using.
Istg, getting rid of this itch in my ear will require an exorcism.
Sorry, English is my fourth language, so I might get a bit mixed up when speaking.
Unemployed William

4th language
It's alright. I appreciate the effort.
well that's nice (i mean your work, not the pplx thing lol)
š
That's wild bro scorched earth no wonder
So European
Why don't you just save the pages in your computer?
pplx also lost 2 of my threads. but the thing is i can't tell which 2 lol. yesterday there were 6, now there are 4. and i don't even know which ones are gone š„²
I save my work, that's why I have 1,200 pages of a single bookābut I lost the most recent revisions. I can redo everything manually, yes, and I'm already working on the manual revision. But itās going to take time away from other projects.
Istg, it feels like something is biting me in the ear.
PPLX kept the history on my phone, but not on my PCāso now Iām praying to the AI gods to help me out with this one.

Of course

Will need censored video proof
Reborn
Mewborn
You noticed the changes on the link when typing o3 pro?

@distant aurora will have to join too
This ritual will summon o3 pro
Yes
yep
haha
i noticed that there was no way that was gpt4o
or 4.1
My prompts from the last 5 days just disappeared. For the godsā sake, Iām gonna lose my mind 
Bet
F
Just on web or both mobile and web
Iām afraid to check on mobile and find out I lost it there tooāitās like Schrƶdingerās prompt

Did you visit an ENT yet
I can't afford it.
There's some hospital in Mumbai that does it for cheap.
Will have to ask my dad about it.
hey there
YOOO
HOLY SHIT
I DID IT
I KNOW HOW TO DO IT
@distant aurora TOOOO
YOOO
GO ON PC AND IF YOU HAVE ORO
TYPE THE o3 lin
Link
I don't have pro
Bro it lets you access fully integrated o3 pro
At least this forced me to start a new book. I found a new helper here on the server, and Iām going to start working on something with him so he can make some money. Some bad things come for the good
BRO
š„¹
@dry kestrel

BRO GET ON PC AND THEN CHECK ON THE MODEL SELECTOR
It says o3 pro and it responds @dry kestrel
O3 pro
I was considering to switch from the pro plan of gemini to the one on perplexity is it worth it?
that is unfortunately NOT o3 pro
its gpt3.5
O3.5 pro
?
.
@left stream
Context length is limited to 32k on perplexity

So it's like 2.5 pro but with 32k tokens?

There's more variety
4o mini?
no
nano...?
@dry kestrel i went on windows app chatgpt and went to the chat, it still shows o3 pro
look at the sys prompt
yeah its a bug
Do you know what tools it has access to on the app?
None at the moment im using the Macos app
here
i clicked on the web search and it didnt search anything
in the app
the web icon is grayed out
Doesnt use the canvas tool either
what are they doing at openAI
its prob just a bug
where it accidentally redirects to gpt3.5 if an invalid model parameter is set
lol
forget GPT 5, GPT 9999 is where it's at
they prob just forgot to update their backend
have some fun with gpt3.5 while at it
howso?
working fine, check if its your device
clear cache, cookies and update it
OK I will test that ty
not just u
pplx having issues td
did all of it and still have this, i'll just wait for tomorrow i think
someone got their threads deleted today or at least part of them?
My paper was just published. Any thoughts appreciated! "Evolving LLMs Through Text-Based Self-Play: Achieving Emergent Performance" https://ai.vixra.org/abs/2506.0018
Ikr
So o1 pro is gonna go bye bye soon @dry kestrel šš
yesterday i was getting clear ass images like this with chat gpt image gen selected
now today i get this bullshit . same model selected
why
well its not like o1 pro was here anyways, it got replaced a while ago @dry kestrel
Dang! Last I tried I couldnāt generate images. Glad I stumbled on this discord
yeah lol
should contact perplexity support then
so 24hr cooldown?
This actually looks good.
Highlighted this earlier because it was 100% earlier even tho most people had issues
Mission impossible rip off vibes.
Down detector and down for everyone are more reliable imo.
Dd best
Some guy is alone in vc.
Does anybody here know where we can give feedback about PPLX to them officially? PPLX Labs is awesome, but I can see some areas for improvement.
unclear
There was an AMA today on Reddit about labs
But since it's over you can just go over to #šāfeedback-general
You can scroll chat to see some of the answers or check reddit yourself
Gemini is an elite prompter
Wait. Lmfao.
@languid star you think that bio got you restricted? š
š
š¤
NO WAY
@distant aurora
Share a link if you can
1 dollar
Will there be a code in the stream or something?
disbelief
active right now
some people said there are region restrictions though
Dang
LMAO
How long will it be valid for though?
thats crazy
Just 1 month?
So useless for me.
it's $1 no matter how many seats you pick
So 2 usd
$1 USD
Oh
Lmao
lmfao
So all just for a month tho
Bro
looks like us deal hunters won again
Theo be giving away t3 chat codes on twitter.
that they are doing to cancel this promo
surely they wouldn't be surprised
Wasn't eligible for the promotion
Fr
can't use a second code on the same account though
thats where the magic of 2 accounts come in handy
I'll just make an account with disposable email then.
Not like t3 chat has proper chat share or export anyways.
I thought t3 chat only supports google accounts
Will make a dummy google account then.
I'm wasting money for no reason. I already have you.com and perplexity pro for now.
š
I have lost interest in grok lately.
I only open it when I need deep search.
grok has been dead for a long time, until new releases
Something promo something.
You've been putting it off way too long
Ik
i've only ever used grok on x itself on a post i want to learn more about, it's a great tool for that but going on grok's website is crazy. i don't think anybody does that
Both go through like 30 twitter posts parallel to web search.
I have the app. I don't bother with the website either.
I stopped using website since they removed grok 2 from there.
i just don't really understand what their goal with grok was. it's so x/twitter involved (whatever u want to call it) it doesn't feel like an AI tool you can use such as perplexity, gpt, claude etc
I'm not gonna bother to open a tab to get rate limited after like 10-15 prompts.
I'll go back to the app anyways to continue the conversation with grok 2 there.
Uncensored ai.
Mostly.
i guess so but once again that's so x invovled it just feels a bit bias in a sense
Well yes
They gave it a little bit of freedom.
Before it was a separate company it was an in app feature
So now it bites back if you kick it on the internet.
grok?
Yes
Truthgpt
what was the company prior?
Was introduced as a feature for people who paid for twitter
Then elon made it a different company
XAI
.
Truth gpt was a failure.
Grok however is pretty successful.
give me grok's history, when was it introduced onto X, give me past company info and etc.
It boomed with grok 3.
pretty interesting tbh, i never dived that deep into grok
like i said i've only ever used it on x on/under a post i want to learn more about, never opened the website until just now to post that link and never downloaded the app onto my phone š
Anyone else having trouble with perplexity? Today every single prompt I submit ends up giving this error after minutes of the AI doing research.
It's down for many @fathom garden
Thanks šš¼ at least I know now Iām not going crazy
Maybe I am going crazy =\
It's not correct
dang. didn't know
Use down detector
ive been too scared to try labs yet š
Im thinking it can replace an AI IDE perhaps, for a small project in python
OpenAI will release GPT-5 next month and xAI will keep working on Grok-3.5
For GeminiApp Pro plan members, we've just doubled your 2.5 Pro limit, from 50 to 100 queries per day.
They reduced from 100 to 50, back to 100
it was never 100 before these stupid limits
it was much much higher
Was a true golden era then
100 is chump change still
Especially if you're trying to get something done
This Fall, we may also get the Comet browser for Android! Running Comet agents on the go will be a big win.
Would be so op
Google is testing a new model, "Kingfall", on AI Studio. It is labelled "Confidential" and likely was made available accidentally.
Thinking is supported, and it seems like it allocates quite a lot of compute time for simple prompts
Gemini 2.5 Pro full is near
you can turn ON/OFF thinking and mention budget for thinking.
Hammy out here going ham
deepseek r1 is better in terms of intelligence
and o4 mini high
Anyone still remember the Cursor perk?
What would be the best way to access the recent 0528 model on mobile
It's been well over a week
People complaining about it daily
Removed because of resellers abusing it ig
both of them just waiting for everyone to forget I see
@left stream reached my limit on pro lol
for free? just use the Deepseek app
and the mobile app
btw this new UI is so annoying
wait they removed that?
It doesn't use the 0528 model
it should use the newest model
Or does it not answer the question what model are you incorrectly