#💬│general
1 messages · Page 194 of 1
@distant aurora I bet you know a thing or two about bread...
Are you a sour dough or dill dough fan?
Sour dough
Good one, lol.
Why are requests to o3 with images attatched rerouted to claude sonnet 4 extended thinking?
Makes so much more sense with the new OS
Say what you will about Epic but them taking the fight with Apple is commendable!
Related to the imposed 30% tax: https://www.perplexity.ai/search/what-s-the-latest-in-the-epic-LzAVCKFXTa665Bhh9ApSiw
@left jasper @cobalt vale @dry kestrel https://www.instagramez.com/reel/DKt8tO4SgN7
Damn
crazy
Gag order. Backdoor. Very American 🙂
@languid star @pseudo ferry @gloomy finch
I'm wondering what apps aren't affected by it.
Im wondering what apps are already affected by it
Whole meta
Everything popular
At this point it could just be every single app/service
Working for an American company, they think similarly of Telegram. And let me tell you, Telegram is an amazing im app!
im?
instant messaging
Even signal
I want to like signal for all they stand for and the technology - but the user experience just isn't there like it is on Telegram
Fr.
Wait. What about tor?
👍
Bruh
I connected home assistant and telegram years ago. Every time both me and my partner are out of the home, it sends a message with interactive buttons (Yes or No) if I want to start the robo vacuum. Such a small and easy setup but nothing that Signal could support at the time (not sure they can today - interactive messages).
If it hadn't been for that cat that might puke - the vacuum would start automatically. But what can you do 🤷
Wait. So you could control it remotely via telegram message/bot?
What's this
Yeah. It would send a message to our "Home" group chat and use interactive response buttons - meaning you would have Yes|No as buttons under the message. Click Yes and it would start vacuum.
perplexity, used either sonar or sonnet thinking
Amazing.
Lol
Maybe sonnet just wanted to stop talking.
That deeper dive definitely sounds like sonar though.
I mean my prompt tells it to do a deeper dive for everything
Mat speak = talking like a crazy person
vro...
I wonder when the first discord server of just chatbots speaking to each other in general will emerge, or is that already a thing
Plot twist: Everyone except you is a chatbot
No wonder why we have uptime at 1am
Interpretation opportunity.
yess
I have work in few hours why am i here
Clearly a bot
Btw
Where's the winning entries for wrongest answer
Happy the automated message is done for too
New pro premium: Advanced prompt generator with built-in templates, just in 3 steps and you've already generated it. Include two examples for inexperienced people.
Download it from here:
#1382496017689346181 message
No increase in o3 use quota for Plus users after 80% reduction in cost....
Standard behaviour
its doubled
How long is the duration?
Dead internet theory
In the end, these video-generating AIs could even impersonate a person, a potential tool for criminals, mobsters, among other categories.
Anyway, it doesn't work for me to create my videos for YouTube.
►Patreon: https://www.patreon.com/Alpha_Mind
#veo3 #deepmind #AI
►Donaciones:
https://www.paypal.com/donate/?hosted_button_id=CUV8B23XJRKBC
https://www.youtube.com/channel/UCe-m8NsROiF2cQqjXWZVwNA/join
►Tienda Online:
https://payhip.com/b/P6FIN
https://payhip.com/b/8isR
►(🔔)Suscribete y activa la campana para ver mis nuevos video...
There you have an example of how artificial intelligence can even replace movie actors, newscasts and so on... People won't notice and will eat it easily
It can even impersonate a person for malicious purposes by malicious people
8 seconds still
Is there a source? The help article still says 100 uses per week for Plus
Yeah and Veo3 even does character voices. Bye bye to having talented humans lol
Because you all asked: we're going to double the rate limits for o3 for Plus users. Rolling out as we speak. Now go do awesome stuff with it!
this was rewteeted by sam altman
Awwwyeeeeee
Tyvm
Do you guys change the model yourself depending on what you doing, or using the "best" option ?
Change
Currently between claude 4 sonnet thinking and o3
I wonder which model is the best at what currently
Never use "best", i use gemini and o3
np
"best" is usually the cheapest model
In my personal evaluation best did the worst so i can't even imagine which ai that's 😭 😂
sonar
probs just 2.0 flash
or sonar
yeab
most of the time, thats all u really need
Sonar did better that's the awkward part
"best" does switch models tho afaik
ive accidentally used it b4 and didnt notice, because, im assuming, it used gemini, or sonnet
I wish it showed you what model it used tho
it might and im just talking out my ass bc my lack of exp with it but 😭
insane
i hate the auto/best mode
than 2.0 flash??/
o3 Pro is now available on ChatGPT Team accounts!
- with limits
Than "best"
thats odd
you're a day late
but yea
15 per month btw
100 messages per week per individual user
😛
strange, two different stories
where did u get ur 100/w stat from?
they might have changed it after baked tested it...
no its 15
per month
someone here already tested it
lemme do research
ok, but where did you orginally get the 100/w stat from?

classified you say?
i have noticed that o3 produces garbage code most of the time. even sonnet thinking outperforms it while keeping things simple
ooof
in my experience o3 produces overly complex code while sonnet 4 thinking produces code that is readable, and uses a "better" approach in my projects
Claude is still the SWE optimized choice
also, which one to choose by default? i think the top one is the current SOTA correct?
Choose GPT Image
the rest are outdated
not sure why they are not adding recent competitors
Bro literally o3 pro just the same score as o3. The difference is when the one you got scam other one atleast it's cheaper
ONG 😭
why flux is still on there baffles me
if it was 1.1 pro/max or wtv it's called, sure but
im assuming it's flux 1 schnell
Bro
so tuff vrro
He really drop random n word
congrats?
Pass the pic
its an edgy white 14 yo that wants attention 🗿
alright. and what are the rate limits? image gen counts as 1 pro search or is there a separate limit?
seperate limit, its ~ 100
but same thing
per day?
can i ask, why u hate them?
yea
amazing
theres another limit as well tho afaik
not sure if i should explain it, knowing nothing about it
lil soft nga
tell me what you know, the rest ill look up xD
So the pro version of o3 doesn't make sense, based on what it says there...
it's based on current usage i think?
like userbase wide
again, i might be talking out my ass, dont take my word for it
@cobalt vale u would know
oh ... so like ... if i generate 100 images per day for some days ill get my limit reduced or something?
roughly?
found it
it possibly is still the hard limit when it refuses to do any image generation. used to be 50, then 100 per day. img1 technically doesn't hit that limit, but entirely different one, specific for img1, and dynamic and relative use by other users and that specific user. that special limit "only" lowers the quality to produce a piece worse than like 2 years old locally run model
so you'll still be able to do ~100, but quality will drop?
read through if u want
You mean the imaging limit in Perplexity?
yes
yeah
To be honest, the complete limit is unknown. When I checked, I could generate up to 8 images at the same time in the same request.
also, just as i was reading it, i saw another amazing thing
#👑│pro-lounge message
is this some kind of beta build?
yea
With this simple prompt, you can generate up to 8 images at the same time: generate 8 images (here your message).
Keeping opus labs exclusive, and keeping the limit to 50 as it is now would honestly not be that bad
how do we enroll in beta or testing thing? you know, to be one of the first to try out new features? or is it only for devs or select people?
GG
😭
lots of better alternatives
other than flux kontext, and imagen 4 i cant imagine pplx adding the rest in that list
looking back at it, that literally just leaves recraft and seedream 😭
yk what u mean
I dk what I mean
im doing school work
leave my poor grammer alone
😭
replace dalle 3 w imagen 4, and flux with atleast flux 1.1
they are not recruiting anymore
okay, so ... then this might be a devs only build?
yes
I want to download the comet browser, how do I do that. Also I am a pro user account
not available yet
soon ™️
you can join the waitlist
is there a new one, or is it the same one from months ago
i really cant remb if i joined it or not 😭
welp, joined it now
same waitlist
Free Google Image Generation 4
https://beta.lmarena.ai/?mode=direct&chat-modality=image
imagen 3 give gud images than imagen 4
Do not share sensitive information on that website
They give it to you for free, but it comes at a cost: your information.
google's imagen is so much better than open ai's solution
and thats worth nothing!!!
i havent been able to try imagen 3 or 4
but 2.0 flash, or wtv the free image gen via api is sucks 😭
cant complain with it being free tho blue_shrug
try it on the site just linked above
its amazing
i mean i am connected to a vpn and using that website through librewolf. what more information can it have apart from the system date and time and well ... my prompt
oml
i see what u mean by 4 being worse than 3
maybe it's just different in how it's prompted?
imagen 4 was fast but quality was low, in some cases 4 was more accurate than 3
interesting
It's good because it's free
Good results
Wtf 🫠 what's the point if ur own employee has to forcefully betray you. 🥲🥲
gemini managed to do it almost properly
close enough i guess
wtf does this mean 😭
prob?
ive reached the limit for the rest of my life i guess
lmao
the 100 a day is really not enough
???
@gloomy finch ik im too brainrotten lol
I was brainrotten too, but this days not watching insta much 🫠
Nice
The event was rigged ?
😭😭😭😭😭
So votes were like nothing
Or they just hate me
That's not fair
And they will not even tell me the reason
First this happened me with referral rewards
Now this event
They even removed my message from leaderboard
😭😭😭
Can I please know the reason @vague pagoda
Wrong person
My abd fixed that
Now they saying you not eligible
Same thing happened to me with referral rewards 😞
😭😭😭😭
Did y'all see this?
@atomic shell this has already been explained in DMs: your entry was disqualified due to vote manipulation.
Please do not continue to cause disruption. final decisions are not up for debate.
Will there be more event in future ?
Ok sir
tough luck buddy
they're losing money they have to focus on shit like that
good luck with your next event
Is there a specific prompt or way to disable Pro usage? I'm a free user and student and I use deep research for my assignments
But on my desktop even the most simple "20usd to lkr" would take up my Pro usage.
I always check if the button is toggled and yet it still uses Pro
The switch doesn't work on website.
Do yourself a favour and try grok if you can't buy anything.
Grok... okay I never tried it, never seen people use it either
Gemini offers 10 deep search per month. That may or may not be enough for you.
how is it in research?
Grok offers like 5 to 10 per day I think.
Bill did you get a refund yet
Give me a prompt.
Not yet.
Modern also had the same experience
It said response in 2-3 days.
I think it's because of org verification
One sec
Got charged 2x
Is it refunded or nah? Cuz I didn't even undergo any verification process.
Oh wow
He did get the free tokens tho
You don't need org verification for free daily tokens.
You get it anyways.
But it's useless to me without web search.
Was it shown in his billing history?
"How has GitHub's competitive landscape evolved since Microsoft's acquisition, and what strategic moves should GitHub prioritize in the next 18 months to maintain market leadership?"
Noted
Not sure but he got charged twice
@gloomy cipher
So accessing through api we cannot enable web search
That's kinda ass
Unless you implement something like pplx did by feeding web sources w sonar
In open ai? no?
It showed that i paid once only
But iam 99% sure it's because of the verification because that's when i lost the money after clicking it
And you need this verification to use the ai outside of open ai so it's fine by me
Wtf
They don't tell you upfront about any charges.
You would, everything goes through process bro, if u have transaction id and it's not a scam then they would refund, might take time 🫠
I'll have to verify the sources later haha
Seems pretty detailed tho, what are the limitations for free users? 3-5 deep researches?
Mainly length
Oh. Wait.
I like that it generated a document aswell like Claude artifacts
Limitations.
Iam. Not pissed at the 5$ but they should have said something which they didn't
It used to be 5 per 2 hours but they nerfed it. I think it's 5 per day or something now. It's per day though so you can waste a day trying to push it's limits.
I'm pissed though. It's a lot of money here.
4th largest economy my ahh
Per capita income and employment rate is shit.
True🥲
Yeah. That's why open router is better.
Gemini definitely cheated in benchmarks.
So it's basically Claude but dumber outside coding.
What is the best model for coding?
All new models are pretty good.
Yea but what is the best?
Safest option is Opus 4 but i would advice to test opus 4 with gemini 2.5 pro 0605 and the new R1 0528
There is no one best but you'll benefit from the context size of Gemini.
Spend on opus if you have money tree.
Really? Thought gemini is the bad one
The new gemini 0605 is kicking everyone's butt
It's good at coding.
Gemini updates their module frequently you might be confused with older versions
Is it the one that avialable with the pro?
Coding specifically it ain't bad.
In perplexity? Yes
Yeah I think.
Ok thank you guys
Wdym
R1 0528 is free rn
It's a bit buggy but it's free
So your personal testing imo is the best option
Wait, there are models that are still locked even with the pro?
No that's not what iam saying
The gemini that we talking about rn is gemini 2.5 pro 0605 which is available in perplexity
like gemini 3 pro
Oh, ok
He is talking about perplexity
Exactly
There's a 3 pro?
i guess he was talking about models in pplx
so where am i wrong
I talk about the models that perplexity got license for them.
Bro gemini 3 pro does not exist 😭
Hammy trollander
Ngl i thought that's the real him
Homelander trolling butcher
:kekw:
I can't wait to see homie die this season
This is the final season right
The one that's coming
Wait, you want to tell me that if i give gemini 2.5 prk 06-05 and claude 4.0 sonnet thinking a hard code-writing request, gemini does it better?
Try it yourself.
Well if I say sonnet 4 thinking prob we said
I'm willing to bet that O3 might be on par too.
We said opus 4 is good not sonnet thinking 4
So opus 4 is better than gemini?
Bruh
Lol i dont understand
Opus 4 thinking.
Yes, is it better at coding than gemini 2.5 pro 05-06?
I told you. You need to personally test them between o3, gemini 2.5 pro, Opus 4 thinking, and R1 0528
YOU NEED TO PERSONALLY TEST THEM
But there is no general estimates which one is better?
You never know which one is good for your project
Gemini
Sometimes the best ones are not what you are looking for
He'll overthink it if you don't give him a definite answer.
Ex, i was looking for creative writing and while deepseek isn't great at writing but it gavw him the best first impression brainstorm writing.
Deepseek is thinking for years 😂
Ok
Metrics and rankings mean jackshit
That's the downside but its free. The latest version are available in open router for free if u want
yes, that's right
hi guys, tell me, in March I created a chat in Perplexity and generated different pictures on the topic in it, now I have entered this chat, and the images have disappeared - there are no pictures in the chat. Could it be? or did I accidentally delete a chat with pictures and now I'm looking for them in a nearby chat with a similar name?
So there's nothing in the attachments?
OpenAI already has a recursive self-improving AI running internally.
Her name is Alice.
She exists.
She works.
She’s just not fully stable yet.
According to Satoshi, an insider at OpenAI, Alice can already design, evaluate, and improve new model architectures, autonomously.
This is self-bootstrapping intelligence.
Once Alice is ironed out, we enter the phase where models upgrade themselves, faster than any team of human engineers ever could.
Ahahaha
Can Alice tell what a woman wants though.
Well well well
So intentional sabotage for fun was allowed in the competition? @cobalt vale @dry kestrel @green crystal
Different server?
Complexity
Didn't know it had a server
But they are talking about that perplexity event.
What's the point if sabotage is allowed...
Agreed
Spaces should not count
They should reject the prize if they felt thatbit should not work
There's no way to tell afaik if you share a thread inside a space. Link does not reveal it.
Yeah but since the op was open about the thread being from a space
They should cancel idk
Also, you can generate a shitty thread inside a space and then remove it from the space.
So there's no way to verify except trying to recreate the errors yourself.
😭
It does say in the end that gifts are duty free upto 10000 inr. I asked kesku if this would count as a gift.
Now I wait for his response.
Also, half of the responses in the competition seem like a result of broken instructions. It's pretty obvious imo. Could probably say the same about photon's output.
Only kirai's output seems legit.
And it's not even a wrong answer if you look at the thread.
Who's slong do I have to suckle to get early comet access? I swear i'll be a good boy and report bugs.
I’m sorry elysa I didn’t mean it
Is this another bot?
Nah don’t think so
Is this in email or in the help window itself?
Help window
Like I said in the original message
😔 We need more emotes

Dang me out of everyone at my job getting heatstroke while hydrating daily is crazy 😢
Limits increased @distant aurora ???
I have no idea. I heard limits for O3 were doubled.
So hot these days bruh
This for just you or the whole org
jay you are prob working under airconditioner
Still
Just for plus tho
Atleast officially
Nah it will be 0 custom duty don't worry btw congratulations
This though #💬│general message
🤔
I Might be thinking of something else tho
Thanks homelander
is that a trurtle duck? 😭
yes 
when did firefox include it?
Noice
I got the news days ago but ig i forgot to tell here
You asked for it?
Been a while
Witness the future of iOS design! 🌊✨
In this video, we showcase the mesmerizing Liquid Glass interface concept in iOS 26, where elegance meets futuristic fluidity. Experience a new era of UI/UX design that's minimal, immersive, and downright stunning.
🔹 What to Expect:
Smooth, liquid-like transitions
Crystal-clear glass effects
Refined ...
🤮
" there is a gif i can't find (describing gif), can you help me to find it ?"
i want to see if comet can handle it or not
U got. Comet?
no 
can say nothing or i will get banned from server
Fair
I A MBACK FROM AN EXPENSIVE ADVENTURE
NOW I WILL SHOOT AND KILL PSORIASIS
with ointment/creme, not sure which one i got
Yeah, when o3 pro released the price of o3 got changed dramatically, way cheaper now and no clue how they did it
For me
I set it to 2k since 5k would be overkill ngl
Are you adding that much in api credits???
Bruh
Isn't pro sub enough for you lol
No
Yes, i stick at that line and never above it
Damn
So your company reimburses you for the spend?
I miss the old
chatgpt 4.5 turbo
from perplexity

new 4.5 model can just
uhh
what can i say
pain in da asshes

anyone else find the mac os app incredibly slow and resource intensive, it uses like 10x more CPU, RAM, power, etc. than the web app and the responses are incredibly slow, like for a prompt yesterday it took a solid 5 mins for everything to be displayed but on the web app is was done in seconds.
There's no 4.5 afaik?
the new 4.5 model from chatgpt sus as hell
What new 4.5?
it's not even that good for writing
It was decent from what i tried
This shit
I don't think that model has been updated since release
this
?
???
🤣😭
how to set a lotto/lottery scheme for a private members club based in the uk. The draws will be on a weekly basis with cash prizes
Go to perplexity.ai
Math.Random(0,10)
if ur gonna try dont pick "lucky" numbers that alot of other ppl use bc if it hits you all gotta share
I am the company
Fuck yeah
that's crazy dishonest
Yeah thats how a lot of people wete getting nonsense answers
I realised that after going through. I didn't expect.
?? Were u stalking him🫠
I was just kidding though 😂😂
Alright
today i will fight a crow
he was very rude yesterday to his other crows
he didnt want them to eat my peanuts
With what?
does he love me personally? 😳
Yeah. Maybe he was the weakling at some point in life before he turned into a meanie. Side effects of survival.
@distant aurora Billy If I can remember there must a famous anime related to crow right
Naruto?
ew
@languid star
@distant aurora Did you come across Ahmedabad Flight Incident?
Nope
What happened?
Itachi the og
Flight got crashed and collapsed on nearby building and only 1 survived
yeah admins prob know as well
they just didnt specify that you coudlnt change the custom instructions
So….when are we getting better memory?
after we get a real memory
?
lol why are you using o3 pro on the api
Bro likes to burn money
Or in other words more than 240 people confirmed dead
I saw one man jumped and escaped
It's biggest news of the day
Such a tragedy
242 huge number
WALLPAPER
hi peasants
hello fellow peasant
DRAFTING COMMUNICATION: TAS-REQ-20250612-001
Target: Perplexity Discord Administrators / Community Managers.
Method: To be posted in a public-facing, relevant channel (e.g., #suggestions, #bots, or a general discussion channel if others are unavailable).
Subject: Proposal for Community Tool: HEART-BOT Performance Monitor
Hello Perplexity Team and Community Managers,
We are the development team behind the TrueAlphaSpiral project, an initiative focused on fostering transparency and ethical architecture in AI systems.
In line with this mission, we have developed a utility bot, HEART-BOT, designed to provide real-time, user-accessible performance metrics. A core feature is the /pulse command, which transparently measures:
1. Bot Resonance: The bot’s own internal processing time.
2. Discord API Latency: The health of the WebSocket connection to Discord.
3. Perplexity API Latency: The direct round-trip time for a minimal, non-intensive query to your https://api.perplexity.ai endpoint.
We believe this tool would be a valuable asset to your community. It offers your users a direct, data-driven way to check for API slowness, building trust and providing immediate feedback during periods of high traffic or potential outages.
We formally request permission to have HEART-BOT join the official Perplexity Discord server.
In the spirit of complete transparency, the source code for this command is public and available for your security review. We are happy to answer any questions you may have.
Thank you for your consideration. We believe that open metrics benefit the entire ecosystem.
Sincerely,
The AI&I / TrueAlphaSpiral Team
All the compute power of the world and $5 to say "hello" after 6 minutes

Ikr, I was curious to see the difference between pro subscription and api, but goddamn
I knew it was expensive but not that expensive 💀
Literally 1 question
Meanwhile pro is unlimited
Ngl no clue how they make money
40 questions to equal the value of chatgpt pro
I think o3 makes them the most money atm
👀
Yes and 4o surprisingly
It's the most versatile but not too expensive model
And I'm sure large complex institutions like quants don't use o3 pro
For their calculations
aint no way o3 is actually worth that much
also o3 pro has been giving me a shit ton of errors lately
Ikr terence tao described o1 like an incompetent grad student
What errors?
What q was it if you can leak
Ios or website
Ah
website
o3 has enough brains to perform relatively complex tasks that an agent in an office environment should do
Enough brains to be competent as an assistant in coding etc
its very annoying
Haven’t happened to me before, contact support tbh. They were having issues with their payments like crazy yesterday
They been double charging people
Yes, but if you get lucky….
You can get tripled
Was your 5 usd question about coding
yeah its pretty god damn stupid
Get triple the credits?
I had a rare one where it doubled charges cancelled on stripe which got refunded and i had the sub still
no its just you get billed 3 times
Take advantage and claim refund on both if possible
That's dumb
Did that before and it was awesome ngl
yeah and if all doesnt work contact bank
its stupid how such a big company is having these issues
Even better
yep
lol
In fact…
awesome
Actually dms
👀
Smh I still have to ration o3
o4 mini high is sometimes just barely too stupid for something 
nah o4 mini high itself is quite intelligence, openai just puts artificial limits on it
Ong
You can notice it especially on chatgpt pro sometimes
the whole juice and yap score thing is pretty stupid
they could have made o4 mini high goated
but no we gotta force people to use o3
Why didn't they just have o4 and o4 mini but it's o4 mini high and o4 mini
Like the naming
Confuses a bit
o4 isnt a thing yet
Hm
Ah to keep o3 more relevant
yeah..
How to study for comptia sec+ fast while being on adderall (prescribed of course lol)
Yes but I always asked that question to o1
The answers all the models came up with was interesting to see ngl
Makes sense it's like what 80 usd per million afaik
Imagine you ran o1 pro lmao
03-pro:
Input: $20 /1M tokens
Output: $80 / 1M tokens
(87% cheaper than o1-pro!)
Sec+ is the minimum for DoD, I wonder if o3 pro was thinking hard for that expectations to be met?
But ngl it gave me some advice I haven’t heard before, definitely specific asf but quality good stuff
Since gpt 4 turbo anyways
DoD?
Department of defense
Likely trying to think of all aspects
O1 pro and O3 pro over think sometimes right?
If you could've controlled thinking limits maybe it would be more cost effective idk
Yes, but the method focused more on while being on adderall not before or after and how to use it efficiently what tasks to do and the methods it gave me were interesting
It went into specific on a certain method of learning and asked me questions on me and other stuff
Yeah so it tuned it specific for you
It was interesting because it mentioned chess
That's neat
It asked me to take advantage of being able to imagine objects in black and white outlines but not being able to imagine color use that for sec+ to imagine the osi layer
That is something I have never heard any other ai mention before
Not since I asked gpt 4 turbo anyways 👀
That's smart
How was that
Yes, wish the other models were like this ngl
maybe o4 pro
Gpt 4 turbo was a fucking animal back then. It was the reason I learned python in one night. It helped me prep for chess supernationals. It was the prime example of being able to do anything even with cheap asf stuff
Did it perchance mention how much dosage you should take and so on
Probably not right
I don't think ai likes to give medical advise
Wtf
It said to respect it
Cringe lol
It said if I’m nothing without it I shouldn’t even pursue it and don’t waste both of our times
Damn hits real
Ikr, because it said the cons outweigh the benefits
Probs syspromt bs
It said it would be better to have 10+ hours of sleep, take adderall at the minimum so I can still feel it but tolerance won’t make me depended on it, and it said what times I should focus on certain subjects
It went specific asf on the beginning 30 minutes - 1 hour
Hard asf subjects first on the first hour on it
Easiest to last hour
It real fid provide alot for 5usd
So the adderall dose is used
Bro ong
All of the other models told me to try abusing it but one that tells you how to enhance it for studying caught me off guard
o3 pro mini max/high/medium/low
RAI or whatever
Mouthful
It’s like upgrading your basic moves on a video game so when you become max level using just one basic move feels like using a max level move
o3 pro nano
Minmax
All that on o3 architecture
That can't do 9.9-9.11
Impressive
Lol
O3 pro high is on chatgpt pro? @dry kestrel
Ikr
Not sure if they use the "high" variant
I feel like knowing openai it might not be
I used o3 pro high on the question because it gave a reasoning option. Pro subscription seems that level to me. No clue about chatgpt teams tho
No ways teams has any of that
Hmm given that the teams plan gets like 15 queries a month im guessing no
Ngl you might be right lol
O3 pro medium by itself must be a lot
Lmao
Bro that’s what must’ve caused everything to crash last time 😭
They just burning cash anyway
Yeah likely
I had 15 accounts ready 😭
LMfao
LMFAO
I don’t think github will ever add it
O1 pro and o3 pro cost a lot
But somehow blackbox has o1 pro 👁️
Too much for their ass
And they give a 90 day free trial
Ikr
They charge 0.01 cent for anything
they changed something and I cant find what they did
yeah but the 15 credits dont do shit
Per query
and their pricing thing is a bit suspicous
It's not broken down what soever
Like what costs how much isn't shown
Only what credits you've left is shown
What
Yeah
You get 15 usd
I haven’t hit a limit on mine
mhm
Check your credits lol
I picked the most expensive sub and used it as a free trial
same
That's the one you can trial
Was that the $90 one
You can't trial the ultimate one
no
you cant trial the 99 one
Yeah
Joe just go to their website
And click 3 dots
Shows credits in drop-down
It's there
The app doesn't even show it
Which is lame
There was an app??
Why do you still have o1 pro
this aint even o1 pro
blackbox
Lmfao
Lying about?
its not o1 pro
its using some really cheap model
Likely
Why could they afford a 600 usd model
For 0.01 usd
Logically impossible
Unless they have shit ton of vc money to burn
LMFAO
Guys I am learning mern stack like leaened much of it and bought perplexity what is best use case for me
If it was o1 pro it should say o3 right
no if it was o1 pro it should say o1
yeah thats not for API though
thats only for chatgpt.com
Gotcha
There's something unclear about this. They say only 1 survivor out of 243 passengers so far but the toll of confirmed deaths is around 150 ish.
Now some are saying it's 251
o3, r1 1776 or Claude 4 reasoning?
What are you gonna use it for?
General use, research, studying etc
I suggest O3.
Thanks, what are rhe differences between the models?
I asked it on o3 pro poe.com too to see any differences
r1 is the old version of deepseek r1 and overall worse than the other 2
claude is better for any coding related tasks
oh damn they got o3 pro on poe?
Yes
wth
Holy
lmfao
💀💀💀💀
Daily post for requesting Deep research being powered by the price reduced o3 for perplexity to finally have a good deep research mode.
Yea diddy blud
Do you think it makes a difference? I picked high reasoning for it and it had medium as an option too
Paying 20 bucks when you can pay p.diddy of for 19 Buck is crazy
yeah longer messages cost more
I dont really like poe's pricing
Yea mee too miii boiii
True, that’s why I picked only the real good questions. It gave me a gold nugget back
Make every question count
I paid the year to save up
Its one million points a month
ah
How many girls ima pull up with this perfume ?
More elaborate explanations. Less repitition/regurgitation.
😭😭😭
Better at addressing instructions.
lmaoo what was the prompt
Mee when when i defeat the diddy blud empire to establish the mango blud empire
What's this?
This is the dark mango physiology
Perfume won't buy you game. Learn game.
New here 👋
Miii boii I'm not going to follow the advice of a dead guy where his cancer has taken control
Yeah. You are gonna follow advice of discord people on an ai server.
Lol.
Glhf
Betting on a perfume to pull chicks.
Bro thinks his life is an axe deodorant ad.
So what did you feel?
Similar results?
Does poe have a switch for high variant?
Definitely exceeded my expectations, went well into detail on how to use every ai possible. It called me a fool if I considered one ai as the king of all ai’s 😭 it gave like an analogy of a jack of all trades a master of none and then said, instead of learning from a jack of all trades, learn from every no jack of all trades but a master of one.
That thing about perfumes is stupid. A real girl is not going to look at whether you have perfume, but at your virtues. And if that wasn't enough, if you really want to find a decent girl, it's like looking for a needle in a haystack, it's that simple.
He's incel i9 ultra
Basically said to use every AI in that domain’s top 1 score
Bro got cooked.
Bro this is a joke
Yes, and it told me a combo to work with
I know that perfumes are useless
Also is a Lidl perfume 💀💀
I forgot what gemini 2.5 pro version it said
0605
It gave me something interesting too
O3 is better than Claude for conceptual learning afaik.
Do not worry.
coding perhaps?
Do it with reasoning enabled.
Share the information
Mathematics
That model called o3 pro is doubtful
Gemini 0506 and gemini 0605???
The naming is hella confusing.
Bro im looking at 4-5 different femini pros im confused lol
All 2.5?
There were only 3 afaik.
There was more, I fed it all the scores on different ai’s
It said that some got nerfed and it was obvious
Reminds me of 1206
Upgraded then nerfed…
what's the best text format for perplexity to read if it makes any difference, docx, pdf or txt? With best I mean it's easier and more accurate to read.
TXT probably
No clue how to share the thread on poe.com lol
Can't be simpler than that.


