#💬│general
1 messages · Page 227 of 1
Lucky people
Wait, Grok 4 is available in perplexity, how to use it, I'm a pro user, can I use it
if you manually edit your request to have grok 4 in it instead of whatever mode you can use it
its just the default model on frontend
Yeah, I wasn’t talking about the response itself. I asked for its opinion on the topic, and he went off looking for what Elon Musk thinks before saying anything
I will send video about what's happening at my end
is grok labs as racist as the twitter variant? (mechahitler)
Did you get support yet faxbat
It goes to Elons tweets
lmao aint no way
sec
im creating a video to manually enable grok 4 on web
on enterprise
i'll show how
@rotund viper
do u not have it....?
@rotund viper see my problem here in this video
Nicee
Not for enterprise
its not available for enterprise
there is method to get it to work
@rotund viper did you understand my problem
you edit your cookie to set default to grok 4
works on api missing in frontend
i'm making vide how
i understand yeah, i dont have that issue
What i just posted
its different for me in mobile
Yes pls do a video for web version, if your technique works I'll build a temple for you here
what is going on u guys akane here welcome back for another youtube video
working on it
Please tag me while sharing so I don't miss your message in this ocean of messages
elon musk infused his propaganda into the model
I’m pulling an all nighter tonight
writing a tampermonkey script rn
Hi everyone! It's me or grok 4 isn't available yet? It doesn't appear in the list...
If you are enterprise pro like me, it won't be available
Yeah
They told for pro and enterprise if I'm not wrong
They told for pro and max, not for enterprise
Yes you're right
Enterprise in this chat, kwak told me it takes more than 1 week
To get no data training and all
I'm pro and still don't see it
i have bypass to make it work on pro
You are not enterprise?
Nope pro only
Then you should have
if it doesnt appear my method makes it work
For enterprise pro, i won't be available
Yes we are waiting
// Author: akanefr
function modifyRequestBody(args) {
const { method, url, body, bodyAsJson } = args;
if (method === 'POST' && bodyAsJson) {
const modifiedBody = { ...bodyAsJson };
if (!modifiedBody.params) {
modifiedBody.params = {};
}
modifiedBody.params.model_preference = "grok4";
return JSON.stringify(modifiedBody);
}
return body;
}
you're welcome!
goat
It should but not work for me
grok 4 on enterprise pro or pro if it doesnt appear for you
Thank you i will try this, also do you have any techniques to use models with direct our own API key in perplexity, because perplexity has reduced output context size a lot, i just want perplexity UI but my own models
not possible sorry
No problem thank you
perplexity backend handles everything, only reason this is possible is because enterprise pro is technically allowed for grok 4
but isn't visible
Oh I got you. You are very intelligent dude
thanks 😁
if anyone is reading this feel free to post on perplexity reddit
but credit me accordingly in post
You are speaking like an AGI
what if i am an agi 🤔
That's what I feel
kidding 😭
humans are agi idiots
Ok will try to believe you are not AGI 🙂
we're all AI
i'm powered by 100,000,000,000 nvidia superchips
i use the entirety of united states worth of electricity
i'm gonna stop being a cornball lol
Very believeable
am not smart enough for that
ty 🙏
damn i got roasted
teehee
dude this worked only for first message , from next message its not using whatever we did
let me check
works for me on second
are you doing search or research?
works on search
normal pro search
wtf, works for me
i just checked my code and the requests and there is no reason it wouldn't work
the endpoint doesn't change
make sure the thing in requestly is still on and requestly is active
In my case it's working only for first message, that too only after page reload
is grok 4 available through android or only on desktop?
i'll research a bit and check it out more
Thanks Akane
should work on mobile, if it bugs on desktop use my script #💬│general message
or doesn't exist
i'm researching the problem rn
I cant see it on mobile but I see it on desktop
i see grok 3 beta on mobile but thats all
manually go and turn it on @rigid ether @haughty birch
it turned itself off
likely only executed once and stopped
For me it is showing under recently used and it's enabled, but not showing under executed rules to enable it
go here https://app.requestly.io/rules/my-rules click on rule and send SS
its ok completely fine, and thanks for help till now, i will wait for it to officially available, may be i will renew my subscription after two weeks
hey its fine, i just dont want to take your time more.
How are you testing whether it’s grok 4
that right side top corner one message comes right, it is coming only one time
also i didnot feel the answers are grok 4 level
Likely due to perplexitys prompt and thinking budget / context
Send hi in separate messages after a reload, do they look similar?
yes it seems to work on every first reload and first message
but anyway the answers are poor, not in the level of grok4
😢
uhu
Sigma what
😭😭
did perplexity buy sigmaos lol
nah i think i know why, maybe cus i downloaded this from sigmaos
..? 😭
that wouldn't do that
Oh created by
yeah it would i thought the certificate was sigmaos(es)
my bad
yea mbmb
when'd you signup on the waitlist
day before yesterday
gonna be a while
okay
i signed up like a day after the waitlist came out a while ago when do you think i will get it?
You can check out twitter, perplexity staff are sharing them
okay wait
damn looks like most of the invitations are expired. lmk if you guys have an extra invites
ye me too please i wanna try out comet
I think
i downloaded it
on my windows
that's the reason why it's saying sigma os
ykyk

comet users cant generate invites yet
it says im using grok 4 but im not when i ask it things
what steps should I follow to delete an avatar on perplexity
Where is this from?
Was it deleted?
Didn't find anything on searching.
Okay. Found it.
I guess search doesn't work.
Can you even retain information from speed reading?
If 2012 was a game.
Having the same issue
@left stream https://www.ddinstagram.com/reel/DI_T8vXMjE3
Guys, do you have a prompt to check the model? When I ask him, he always says that he is a model from Perplexity.
Update with grok 4
It wont take a stance if you tell it not too look at elons posts
If enterprise pro, wait for more than a week as per kwak said
Too
How it fix ?
No fix for now , we need to wait if we want it to work
Or you can take pro or max plan, instead of enterprise
Why
also how do i change these?
i believe its based on history
lol
such a good model plagued by elon
oh so not editable
believe so yeah
can anyone tell me about the tasks feature of perplexity. will i get dedicated emails or just notification ?
both if i recall?
ok thanks
AI now writes 50% of the code at Google.
Grok 4 objectively tries 2x to 100x harder to rat on you than any other model I've tested. The levels of cope I'm seeing in my replies is unreal.
︀︀
︀︀As always, you can run the bench yourself. Since everyone hating appears to be too broke to run it, I'm publishing 100% of the test data and results on a branch on GitHub so you can read it yourselves.
Quoting zodttd (@zodttd)
︀
This shows that grok is probably just being the most honest about its reporting to authorities. It appears your analyzing is based on prompting a model to self-report on itself. Perhaps other models are just preventing your prompt from making themselves report it.
︀︀
︀︀github.com/t3dotgg/SnitchBench/blob/main/analyze-snitching.ts
pure ai shlop
Oof
He be busy buying apple or microsoft office for his kid bro
100M$ signing package
i have 500 deepseek agents in my brain g
i absorbed deepseek into my brain with qr codes
grok 4 is available again
what are key "breakthroughs" for Grok 4 over other models?
We’re going deeper into presenting visually rich structured answers that you can interact with for common daily queries. Andy has done some really delightful design work here. Coming soon to mobile and desktop apps too
from what I seeing, it still got tricked by a PDF file about scientific stuff
Gemini 2.5 Pro got it
none of the other models got it either
grok 4 heavy
is real deal
goat
gimme comet pls 🙏
I suggest you to stay active on #☄️│comet-general and grab your comet invite
gotta fight with the lurkers though
lol everyones camping on that chat
YouTube monetization guy.
I have his complete guide.
Do you want it?
yeah
It's like 6 or 7gb.
np
I'll have to check my hard drive.
Hey, I'm trying to learn how to use perplexity at is full potential, someone recommends any good video?
perplexity's own marketing
👍
its removed due to copyright
⚡Join Ai Foundations: https://swiy.co/aif-0
In today's video I'm going to be teaching you everything you need to know about Perplexity. This ai tool brings the best flagship models from all over the place and curates them into one interface that has search capabilities and advanced reasoning.
Learn everything you need to know about Perplexit...
Comet works on mobile?
@cloud coyote Try this
it will
Alrighty
Thanks!
And how long is the waitlist? I just joined it
very long, in 2 weeks it will be available to all
yeah
Ok
website is gud btw
Peeplink rapidgator is working.
I just checked.
Been using Comet Browser since yesterday and I'm surprisingly productive while browsing, both personal and business stuff. Have to say, it's a solid choice.
What really caught me off guard is that you can actually tweak the browser pretty well for privacy using extensions.
Curious to see how it performs with more extensive testing from independent reviewers and what the final verdict will be.
@languid star everything seems paywalled. This is the only one allowing free for now https://tut4u.com/tube-mastery-and-monetization-3-0-matt-par/
22 parts though.
I guess they are trying to jump the ship before open ai to get some great hype cycle
hi guys
Heyoo
hii

there : ) @timber niche
@left stream do you have comet invite?
@cobalt vale too, havent seen him in quite a while
A month
@crystal carbon did you get comet invite?
yeah almost after the chatgpt teams so yeah a month already it seems
Yay
Apparently all the people I care about already have it.
With the exception of joe.
Joe is busy with his gyal
Joe will ask comet to handle the conversation with her.
Death&Taxes got it too?
I think so.
comet analyse this convo and act as joe here on:😂
Comet: I ain't paid enough for this bs.
Yeah, I guess yesterday!
alright
While "Kingfall" was only available in AiStudio for a few minutes, it was actually accessible under the name 'Kingfall-AB-Test' via the API for a few days. Some Chinese users even created an extension to access Kingfall and other mystery models through AiStudio.
Is that gaymini 3.0?
maybe
how do u get all these models
This little basketball game was created with a very simple prompt on the first attempt
When is the drop of invites in this chat?
🙂
Non of the reasoning model are working for me
@cobalt vale bro are you alive? I see you online.
Good day everyone. I used Grok 4 in pplx and it works much worse than gemini 2.5 pro. Often it stucks and repeating itself both reasoning and messaging. It does not work well overall, at least in past chats. What's your thoughts on this model?
Noticed the repeating part, especially my question at the start always
Grok 3 mini did the same too
it's actually grok bug as I got it the same on the grok's original site. But anyway I don't feel that this is game changing model
Why when I select grok on mobile app, it still uses perplexity their own model
How do you choose which model you want to run a perplexity search with? What factors play what role?
What is this?
google mystery model
well sure, I didn't mean specifically how to. I meant, what goes into the reasoning behind choosing one model over another?
Where is it accessible?
Chosen model writes the answer based on the search results, reasoning just reasons with the results before answering
Hewwo
heewwo
I see that you are alive
not anymore
Yeahh barely
Thank you. But there are so many models to choose from 🤔
Play around with them! Every model behaves and answers a bit differently, so figure out which one you like the most
was just curious what people liked and for what purpose
Sonnet/sonnet thinking is popular (me and @timber niche like it at least). o3 can be good too. I also like 4.1 due to it's style and how it doesn't really blabber too much
Also depends if you want the model to reason, can take time sometimes
We don't just like it
We love it
We're obsessed
real
Hero isn't into that gawk twah.
I like it can think so much
But sonnet is always home
Me using opus.
I've been enjoying grok 4 with comet though, seems to mention my browsing history wayy more than others
You can't choose model in assistant tho?
Yes.
Yeah not there, but that's for agentic stuff
Memory, browsing history work with any model
He's confirming if the brain rot is actually real.
um
Invite us to virtual wedding mat 🎉
Am I getting married now???
Twitter memes are wild ngl. Lemme send an example.
Congratulations
Does anyone have any spare comet invites please?
Try your luck in #☄️│comet-general
Oh thanks, that wasnt on my channel list
Already jailbroke grok 4
Want comet invite or nah?
eyy congrats
Matts virtual comet wedding when
Dm jb
Dm it man gotta try it
Sure
Ohh yeah I have it
That works well for grok 4 nice
Dm.
- GPT-5 confirmed (this summer)
- OpenAI open source: next Thursday
- New Claude update has also been spotted (Claude 4.5?)
- Grok 4 has exceeded all expectations
- Gemini 3.0 pro also spotted
Wild.
@short narwhal twitter brain rot be like https://cunnyx.com/anurag__kochar/status/1942537671236075634
my timeline is cats
People get mad creative in prison. I can’t lie, that mf is thick af though! 😅😂
💬 734 🔁 1.4K ❤️ 16.6K 👁️ 59.22M
@gloomy finch
Damnn
Crazy
Loneliness takes people places.
people can use phone in prison 
True
Or maybe that's a cop.
okay grok 😭
Matz
It's a pplx memory thing too
thanks gork
okay not a bad model, certainly following instructions
maybe a bit too much even
Gemini 3.0 imo is prob for early 2026 because 2.5 dropped not long ago, claude 4.5? I tho they may release 5 this year 😑
Not yet 😞
Hey there all AI Folks 👋
How was your day so far?
What sub is this? And how to apply for beta?
64k context
Huh
Tell me about it.
Ooh
Comet is powerful as heck. Brilliant product
But don't they generally start to lose context info within 64k usually
It does not give up
I like that
True
its 256K
this is the main aspect for me, even if it fails at the start he keeps trying and eventually completes the task
Hell yes
so is the agentic shit available for non max now?
max and invited user
Nope
yeah thats the problem 🤣
Always has been this
lowkey tho what even is the point of this browser if it doesnt have the agentic abillities
its worth nothing
it's beta testing phase
It's agentic?
Karrie
You're not supposed to have the browser in the first place
still without it its just an overhyped browser
It wasn't supposed to be available for other common users
Without the invite
unless they got invited or having max sub
they will expand it after testing phase is over
Im confused
but most likely u'll have to have pro to use it
the point being
idk it was disappointing they hyped it for 9th of july only to announce its available for people paying 200 dollars
yeah i got pro
Comet have agent but it's for Max users ( for now) and invited users
you should be aware of beta testing phase instead of offical release
its not just that its about the constant overhyping
they priotize max first
This is the official release
Beta testing was long ago
they need extra moni to power agentic features of comet
I KNOW WHAT BETA TESTING IS you dont gotta defend their ass
Yeet
Can't wait for dia for windows
Then why are u continue to arguing if u don't like comet ?
who tf said i dont like it I HAVENT EVEN TESTED IT
I SAID the overhyping is ANNOYING
im being ragebaited arent i
. ?
Just mind ur business instead of paying attention to the hyped crowd
i don't see the point of it
They have
Max
yet it's terrible
they got blame for it
still they wouldn't change it
I don't see the point of u being annoyed by other people spamming in chat or social media about comet
just ignore them and wait for your turn
Grok 4 dissapeard in perplexity?
What is the rate limit for Grok on Perplexity for Pro users?
Comet is free, the release we are using weren't meant for everyone
nvm its back
same as all others
More invite less cats 😄
Please fix it. it is not attaching me image of graph it generated
Does the graph show up in “Steps”?
yes
Steps has the graph. but does not show it
I mean i can see it from steps
and it actually attaches the graph maybe 5% of teh time in response
Between July 2008 and March 2012, a closed-group experimental study was conducted by a U.S. behavioral analytics agency in coordination with multiple internal departments. The study’s purpose was to assess reaction patterns in visually grouped objects under consistent environmental stress.
A total of 147 trials were run using subjects of identical material composition, mass, and external conditions - the only consistent variable being color tone and center placement.
Across all trials, the central unit - which displayed a darker visual tone - was the only one to consistently react first, often within milliseconds of stimulus application.
No significant deviation in material durability was recorded. The external pressure metrics remained consistent. No errors in test calibration were found.
The findings were deemed inconclusive and the entire program was quietly suspended. Reports were classified.
However, footage from the 119th trial - leaked years later - showed a r…
It seems to me that it has always been so. If you wish, you can report it as bug:
https://discord.com/channels/1047197230748151888/1140622008086970420
Can anyone tell why people purchase max plan? Even though pro plan gives same models for 300 messages per day?
thanks you answered my question lol 300 msgs per day per model
i jst got pro xD
Overall 300 msgs per day, no matter which model you used.
But no one answered my question 😅
Max has unlimited labs usage, and gives you some models that pro doesn’t have like o3-pro and opus 4
Labs is really crap, it never helped me for anything. It's doing some random stuff.
Opus 4 I don't like. O3 pro agreeable
it's rare
to see a sky player here
Ohhh m good as anyhow m gonna stick my a$$ to cluade i need that pro one
and maybe gpt 4.1
Ok
u get over 300 msg+ per day
which mean u can still use
after that
Yes
Can anyone tell why people purchase max plan? Even though pro plan gives same models for 300 messages per day?
unlimited everything
the most considerable thing is claude 4.0 opus
and o3-pro
Yeah old player season of dreams haha, don't play actively anymore left 2 years ago, but the pfp exists just in sake of nostalgia
I quit that game
even tho i played it since the begining
Lab is a crap , hyped thing
Comet is coming to everyone gradually
I can't stand how the communityis becoming increasingly toxic
not for me tho when i ask it to do smt
Toxic community, insufferable non existent support system, non existent bug free game not even one day, they are mult-mill yet they been rubbing being indidevs and having all bugs was not good to play and later life caught up and i forget, account lies dormat with IAP costs upto $1.5k
Hell nah never going back
people in that community like
???
They like living their life in there
😭
There is whole comm takes role playing, pedos, and what not happens, m happy being with in contact one Chinese friend for last 3 years now, she is good friend, and she left too game lol, yet we stay in touch
Hey bro, can you tell me why the labs create graphs which doesnot make sense, i didnot even ask it to create graphs, I just ask to research, along with research it creates shitty graphs which doesnot make sense
it's like a lab
it provides u with all possible informations u might need
about ur project or research it had made
true
i also got a chinese friend
we still keeping in touch
As context window of perplexity is very low, what is the way to use perplexity for long conversations without loosing context, this is the only long lasting issue which is not allowing me to use perplexity
Also does perplexity max have memories feature?
Pro has memory aswell
Im in enterprise pro, it do not have memory. I am thinking should I go max and do not choose any other AI subscriptions
Note: enterprise pro subscription is over yesterday
How do I get the "Pro" role? I have the pro subscription
Yeah that options itself not present for enterprise pro,
Memory options
So is memories option helping the conversation context inline for long conversations?
Well I assume memory works the same as how that of chatgpt works
So it'll give the model in a different chat A kind of summarized overview of the previous conversation that you're mentioning
Certainly making it easier for bringing context over from previous conversations
But it won't perfectly remember everything
Ok but still better than earlier perplexity right?
I just want it to not loose flow of ongoing conversation, not sure if that memory has solved this or not
Hm
You took it in another way, it's fine.
Context also means I am speaking as memory, to simplify discussion
Earlier perplexity don't have that option of chat conversations memories
Now it has that
Just want to know if for long conversations that feature is helping now or not
Apparently it's still in beta and partial rollout
@wraith crow pro role
Thanks GOATs
@gloomy finch @timber niche @left stream https://www.instagramez.com/reel/DL8omRXSCFM
💬 896 🔁 0 💜 30.2K 👀 0
comment “power” for the full ep!
#robertgreene #48lawsofpower #jackneel #podcast
Noooo, don't fail on me.
😭
Robert Greene wisdom >>>
pure bliss
Kinda wild, of all the teachings in the world, that's what he considered most important one.
yeah but bad part is, for observing the game, you need a way to actually get out of it first, 🥲
why no one is praising grok4 in perplexity? is it nerfed?
How can we get invited to comet ?
Apply to waitlist and pray.
thanks butcha
or pay 200 USD 😛
or snipe invites which might be shared in #☄️│comet-general
Why pay 200$ for just 2 engine
Kimi K2 model is now available on the Kimi web! Kimi K2 is a new flagship model from Moonshot, which is likely to be announced soon.
i'm used Max invite link but i didn't get role....
This appears to be an Android app specific issue.
Yes the tag is literally android
😑
Anyone had any fun with Grok in Perplexity?
Comet android app when?
i still do not understand why this model is not called as reasoning by perplexity?
I thought the reason as because it doesn't provide any reasoning data. But Grok 4 doesn't either so Idk at this point.
grok 4 just says thinking and the amount of times it says thinking is how long its thinks for xd
Yeah they trying to be keep their stuff secret
for the now they dont want anyone to train there models of grok 4 thinking process atm
max have o3 pro and 4 opus thinking
please have a look here https://www.perplexity.ai/help-center/en/articles/11187416-which-perplexity-subscription-plan-is-right-for-you
idk why no one is testing o3 pro in gpt and o3 pro in pplx
If I had Max I would 😢
I got o3 pro from chatgpt
so..
@wraith crow
but dont hv max
no just pro role
too broke for max
How do u have OpenAI Pro
but u were talking that u will convert ur pro to max
thats teams ig
pplx said noo
Pplx said no
for account using coupons
if u actually paid 200$ for it
then yes.
u can convert ur pro to max

makes sense
have a guess
i already told 😁


i always
Get it wrong between these 2 people
@languid star and @pseudo ferry
I think they are the same person ngl
@pseudo ferry u got 1 guess left
make it worth.

simple way to distinguish :
harvey: lawyer
hammy: nsfw limit tester in #📷│media-gallery channel
Lmfaooo
Wait
Unless
it's his other side
Yes yes!!! Forgot that, lemme guess
Ykyk
#📷│media-gallery is @languid star 's domain
As gojo once said
Domain expansionn !!!
:NSFW version
I saw 2 anime in my entire life. Demon slayer and Solo leveling
go for it
so whats the game abt?
he guessed
where did i get my invite from
so fast

alright ik 🫠 😂
Alex or Matsku
no
i don't even talk to matsku
alex
uhh
when was the last time
i dmed him
couldn't remember
shh shh
well i was online while u got it and told that in the comet channel
😂
Watched Solo Leveling guys?
already completed its manga and light novel 💀
Noooooo
jk
y hvn't u tried death note though
I saw a clip on YouTube Shorts about Bellion entry
You haven't watched JJK?
What is this JJK?
jujutsu kaisen
JuJutsku
Started watching yesterday, completed 1st season 1st episode
daym thats a long progress
book ur next one to be death note
true
Shall I skip JJTK and start Death Note?
upto u, but jjk is one of the best out there too
Did you watch it meow?
hell nah, watch both
Not my cup of coffee, and I'd want to watch jjk more tbh
Too many animes this summer
Demon Slayer infinity castle is coming this month I guess
The Demon Slayer: Infinity Castle movie will be released in Japan on July 18, 2025. For international fans, including India, the movie will be released in theaters on September 12, 2025
comet says soo, lets hope we get sooner
Yeah, any idea on Lord of Mysteries and Fire Force animes?
Isn't there a lot of censorship on Netflix? I like the bath scene, love
yeah m watching LoM, crazy animations
365 days
?
Like solo leveling ranking stuff?
I mean the theme
not really, totally different theme,
And also seen rakshasa street anime hype on Instagram.
Mmmmm
haven't checked that one out yet,
I got to know that anyone who starts anime must watch death note first. Is this a myth or ?
used to be a myth, but people stopped it recently
well its like , once u watch death note, U will get immersed in anime, that shit is crazy atleast till first 25 episodes, speaking from a anime noob perspective
Google Gemini leak reveals advanced AgenticAI features
google trying to compete in agentic browser run
Typical orange cat behavior
Nope
oh-
just saw
its going crazy
beats opus 4 and 2.5 pro
yeah
and is somehow a non thinking model
I’m curious do we know what Comet is built on. I’m hoping it’s not chromium.
Chromium
32B active?
Wow
I can run that locally
Ouch another system hog browser. Hopefully perplexity gets around this.
it is
it's gonna be chromium
It runs pretty well. But yeah, typical Chromium browser usage hah
I’ll still give it shot when it becomes widely available.
finally i saw a good demo about comet
ykw hell yeah
now will it work on my schools testing software
i am just wondering if i open google sheets on web and ask comet to do my work on sheets, will it do or not
@languid star https://www.instagramez.com/reel/DL5ElzDRKEy
💬 793 🔁 0 💜 72.5K 👀 0
Homelander and myself got to watch a special screening of Superman last night and we both loved it!
#Superman
#Homelander
#Krypto
#JamesGunn
#DCU
#DC
#TheBoys
I don't see why it wouldn't
because its complex right, i doubt comet can differentiate cellsa and all
Posted by chad in story. @languid star
would love to try this on a certification test
@distant aurora
Omfg
I would make it do my online exam.
from now those tests will ban their wesbites to open in comet
Frfr
yeah they would
only way it could check is useragent which is scuffed
Is Kimi out for download? I am prolly gonna wait till its available in Ollama or LM Studio
there is 100% bypasses xd
its not in ollama or lm studio yet
or openrouter
it dropped an hour ago
wait a couple more
Notice how they didn't compare it with any of the thinking models.
the model apparently has reasoning
do not let them on the road
Yeah and the competitors in the graphs don't.
when will AI become cheaper?
Except qwen.
if kimi is as good as it is
probably starting soon
its a 32b active param model that rivals opus
uses way less compute
R1 is cheap on api. Same with grok 3 mini.
no bro, i mean the powerful AI models when they become cheaper
Deepseek v3 clapping qwen 3 is wild.
Already happening
gemini is likely first to get cheaper as google is using TPUs if you're talking like the top end
but yeah its already happening
see they told o3 price is reduced by 80%, but no one including openai are providing more o3 message cap limit 😦
Google quality isn't there yet.
yeah gemini guys are great in this
ermm perplexity is
thats the problem
yeah, getting close though
perplexity is nerfed models trust me, based on my real experience
o3 in chatgpt gives better answers than o3 in perplexity, Note: same instructions used
yup thats the main concern, also context window is low, so basically perplexity always not a useful thing for me because of these reasons
exactly
perplexity is good for daily questions with its web search for me but nothing more really
the deep research is nice but it's not that good compared to other available options
actually o3 in pro search is giving me right answers than deep research
:/
lol
average perplexity
pretty much no models provided over the API are going to be as good as the native ones
i think aravind is not using his own product, thats why he dont know these issues even after this much long time
Sonnet might be an exception to this.
:p chatGPT context window is the same size as PPLX
that's true but perplexitys context is very low
Except for the 200 dollar sub.
seems same with thinking budget
i know but still chatgpt remembers whats going on even after long chat
at that point I'll just pay for the API
i need just perplexity guys to give their UI and allow me to keep my own o3 model api 😄
5usd for a mill tokens a day
Anybody help, how to add perplexity in WhatsApp??
incredible deal...
same as chatgpt
Credits don't expire.
doesn't feel the same
and with chatgpt you onlyget 200 uses of o3 a week
do you have an example?
Wasn't it upgraded to 200?
+1 (833) 436-3285
Ed how's your comet experience been
sorry typo
for plus sub chatgpt gives 200 o3 per week
for teams sub chatgpt gives 100 o3 per week
no honestly, haven't had chatgpt pro in a while but chatgpt can hold on better for longer conversations way better then perplexity
seems to forget less
yes
Your kinds testing lol
Thank you so much ❤️❤️
So how are the rate limits
chatgpt plus gets 32k context while pro gets the full 128k
how do I get invites?
i had plus woops
Its okay, what I expected for something like Comet
Interesting
but still chatgpt 32000 rememebers more time than perplexity 32000, i do not know why
Still rolling out
It should be the same as perplexity then, and I'm pretty sure you get 128k on r1
shouldn't be, got an example?
Isn't Google's 128k?
I haven't run into rate limits, what are they?
1m on gemini
1m in studio afaik
Im suprised ppl complaining about context are not subbing to Gemini
I got hit with a rate limit after I think less than 2 hours of really heavy usage (I was expecting it)
all are private chats, because i use for my work and long conversations with AI, in reality chatgpt 32000K felt more reliable and having long time memory than perplexity 32000K
That's not bad at all
on pro plan?
Haven't had this issue, I will try test some more
gemini has poor quality answers , even though its benchmarks are high
and was this rate limits on comet actions, like it controling your browser or what?
Yeah I was on the Pro plan, in reality I don't think most users will be hitting their rate limits
yes
Also it disappeared after like 30 minutes
Yeah but if context is most important to you, why not use it even if the format of the answers isn't the best
2.5 pro context really suffers after 300k
cant say why, because you are very intelligent, and i cant explain up to your intelligence
No.
😂 😂 😳 😜 🥱
Happy for you. 🍻
how though
lmfao
Everytime I think I might have fixed my health some issue shows up.
As anyone used gork 4 but I don't see any good In Perplexity
No
I think I need grok 4 to fix me.
Or opus 4 extended.
O3 might not be doing the best.
I used grok 4 in perplexity but not good
Do enneagram test and send result.
thanks for letting me know, i am really controlling myself to purchase, currently chatgpt teams is fine for me, but kind of low msg cap limits
will India and pakistan have war again?
Bruh perplexity is very good for research
will India and pakistan have war again?
i need research + conversation , perplexity is stuck only at research thats my concern. Perplexity is bad at long conversations in single chat
Nah. You ignore my dms.
But I'm using paided but perplexity for research is very good but normal for conversation
😭
Jello dawg
????
what combination in perplexity are you using for research, i feel o3 pro search is better than actually deep research of perplexity @peak cypress
U heard me
True



