#💬│general
1 messages · Page 196 of 1
LMAO
@timber niche @short narwhal no sign of the improved deepsearch.
We are back to room sized computers?
I'm wondering if that guy was a troll.
Veo3 cost cut is crazy
Instead of the real guy.
Veo3 fast is great
The week isn’t over yet 
Does non english well too
Also how did oai reduce o3 costs by 80%
And it was next week or so 
Pplx glazer

Zero glaze
Pplx labs is so good
Bruh come on read the room
Fair enough. 3 more days.
Free student account
Mat is too good for the world
Mat is too good to Claude.
Nah they’re not shipping on weekends, 1 day left 😭
No claude is good to us
Hello
She's my bestie
Hellooo
Hola amigo, como estas
Ciao ciao
?
"You are under suspicion of pressuring women" alerting thr fbi now
Kkk
Es que?
Stop doing that i told you last time
😂
How to commit murder legally
Lo siento
Oh wait it wasn't you. But same thing
Usa stand your ground
jeyplexity is hallucinating again speaking Spanish
Distribute the guns. Let people deal with problems themselves.
No englis
Only Espanol
💀
No Englis no visa
How did people not hear abt this
He ain't got no aura. Claude is stuck at 100m monthly visits. Same as perplexity.
Viva mexico
I asked pplx for higurashi watch order
Claude is better, even if less users
Deepseek couldn't go over 400m cuz they got ddosed.
That's like saying the sky is at the core of earth actually scratch that because it's more believable
With lesser rate limits 🥲
More users more demand more rate limits
I use it for interactive ui-s
We both know that's cap.
Claude probably has least number of users.
Not really
Cuz of the costing alone.
Claude web does
They somehow still provide sonnet 4 for free users, open ai could never do that with an equivalent model
Fair
Deepseek doesn't need to get ddosed they are always overloaded and crash yesterday i lost 2 cents because the server crashed and i didn't receive any response 😭
Claude has the best safety and sexurity
It was working fine for me yesterday and today.
💀
Guys do you know any way to make profit thru GPUs
Still waiting for opus
Nice spelling.
Create renders
Like renting or idc
No by default it's just slow if you look at their api in open router its one of the slowest even for the paid version and always traffic
Give me a query, I'll run rn.
And 3060
Lets use one promo code ahh idea
Easier than fixing problems.
Mining is ded rn
Apple move
😭
Now everyone will know pplx
13000
I used api so i don't receive limit
Try using lambda chat.
My first input was 32K token 😂
This is an actual pic leak of @distant aurora
What do you think about this
https://vast.ai/?utm_source=googleads&utm_id=circleclick.com&gad_source=1&gad_campaignid=20733719727&gbraid=0AAAAAC66i417BkaP27_04qfRqYmybtSCH&gclid=CjwKCAjw9anCBhAWEiwAqBJ-c1E_IQ8q9m2dK-59evzrpM9QiiTpDB8lY2Er5Fpz8nUmNxWDc4NfrxoCmJ0QAvD_BwE
Reduce your cloud compute costs by 3-5X with the best cloud GPU rentals. Vast.ai's simple search interface allows fair comparison of GPU rentals from all providers.
Yeah
@short narwhal do you have any tattoo?
<@&1098365634498285709> are you guys checking your subscriber logs to see the people that signed up from the moment you released and make sure that you're releasing the newest features to them? are you double checking to make sure that the people that are subscribing and using API credits and loading those API credits are definitely getting bumped to the list?. I really don't wanna sound like I'm complaining, but I've tagged the owner in the company numerous times. I built several shortcuts. I've reached out with specific asks and I never get any response I even tried to move a discounted subscription code over after being a subscriber for a year and 3/4 and I was unable to get any help. The fact that the comet browser is still not available to your longest running users is absolutely mind-boggling. can you please help me understand how this is happening? Thanks.
I love pplx and still sub and use every day. but dayumn...
Nope, got none
Thanks I appreciate your quick response!
⚡️⚡️⚡️
let me know if you want some help. I can build you guys an agent that will do all this for you and make sure that you're not forgetting your longest running supporters. I really appreciate it and I'm really not trying to be snarky. It just blows my mind that you've created one of the most reliable and impressive long running AI products from the start but are somehow completely missing your longest running support group
@dry kestrel ngl while following o3 pro’s studying method, I didn’t realize how much time passed, the method helped me not to burn out now I didn’t realize I wrote 10+ pages of notes 😭 i was doing one video for the entire section then cornell notes and repeat
dang nice
Yes, now I’m gonna study some more again, just commenting the update so far
Ykw
I won’t be on for a fat minute might as well test the jailbreak for o3 pro 👀
can someone explain this why is my api running on my bill even tho iam supposed to be getting the free daily tokens?
YOOOO
yeah they list that in the model updates
@blazing solar did the same thing happen to you?
they r fixing it though
BRO I CAN FINALLY TEST THE JAILBROKEN GPTS
It’ll use your paid credits if you run out of free daily credits
i didnt use any of my daily credit
I can wait patiently for canvas and image generation. Model selector makes it bearable
Are you using a model that doesn’t support the free credits
350 token
Ehh
i used o3
and image gen
@distant aurora xd
Bro not even o3 pro is that good
yeah lol
Inventing your own language has to be the new benchmark
haha
It’s gonna be a while before a jb works for o3 pro again 😅
yeah o3 pro reasons for too long
and even for normal queries it likes to reference openai guidelines
What am I looking at?
Looks gibberish.
What was the prompt?
Is it just me or have Perplexity's models been speaking out of their butt ever since the Cloudflare outage? I'm seeing really inaccurate responses
Sonar and R1 had downgraded performance during it, could be because of that?
Not just the self-hosted models, even the API models such as Gemini and O3 have been missing the mark
I know Claude’s api wasn’t working for a bit during the outage, not sure about the others though
This is a peak example of model quality of responses do in fact reduce with lower compute, even with decent compute
Apart from the reduced token count per second, you also see a degraded overall response quality when the compute is constrained
if they don't appear in the library, is there any way i can access them before the 24 hours are over?
New user to Pro. I'm using a S24 Ultra and changed the default assistant to Perplexity and all permissions however when I press the home button, Google comes up. How do I get Perplexity to work without having to open the app every time
I sleep at 4 or 5.
I guess human beings must sleep for 8 hours?
Mmm
@short narwhal Typed something and deleted it.
😭 no need to snitch
What did mat type?
It's regarding Sonar Degraded performance thing
How is your ear billy?
Theoretically recovered.
I'm trying to get used to the grease.
Now I'll sleep.
why is perplexity delayed tbh
is it part of perplexity?
its okay for me but i hope it gets fixed soon because the words dissapears randomly
Wdym
Hey i'd like to ask is Perplexity still using their own model based on R1 for Deep research? Or is it combining different ones , i imagine having claude Opus thinking model as part of the deep reasearch as well at least for the PRO users
YEAH
Well wouldn't make sense for Deep research on it own to have 50 limits
Maybe using opus yeah
plus o3
O3 already available for pro search
I honestly don't see them allowing us to select models for deep research
yeah not for deep research, they could combine like they do now
They just said they're going to improve it
But that's about it
Honestly i just go to gemini
For deep research
i hope so because deep research is static quite some time now
Pplx DR is simply a deeper search
i don't see much improvement
you can't beat gemini for deep research man i know that
nah chatgpt DR is high quality
he means on chatgpt page the research mode
is it the o4 mini high working there
Yeah
ah ok
ChatGPT DR is o3 (the good one), the shallow one is the mini o4
now o3 pro is out
Plus has like 15 shallows too
So selecting the model + dr is pointless?
Yep
True sun is getting up
I’m chillingg
I do remember winters being like that back in the uk
Summers***
10pm sunset
3.45am sun rise
So annoying
Enjoyy
Yeah hate summers due to that..
I’m gonna hit the bed before it’s actually fully up
Do you play souls games or do like living
Same
Do you think I mean playing souls like games by “chilling”
Naaw
I don’t like souls solo, but with a friend nightrein could be fun
😭 exactly
Chained together ruined friendships
Yepp
Night
Hello everyone!
Hallo
I don’t understand what’s going on here

Planning to sell Jellybean for Nintendo Switch 
pls no
Perplexity Tasks are rolling out to Pro and Enterprise accounts.
This will get really wild on Comet
https://video.twimg.com/amplify_video/1933215329154404352/vid/avc1/1920x1080/pbQOGV7Jenwgr2_c.mp4
anyone have it already?
Me
This is the best thing I came across on Perplexity recently
yet another " why is this not available everywhere" feature
Grok limits for scheduled tasks seem to be lowered in the latest web build. Now it has a limit of 3 daily tasks and 10 occasional tasks (down from 10 and 30).
Also, DeepSearch will be available in Tasks too!
- The tasks feature is not yet publicly available
what now
o3 is a research model
Sonnet 4 is an agentic model
Gemini 2.5 pro is a coding model
Gemini 2.0 flash lite is a RAG model
o4- mini is a task planning model
Deepseek r1 is a thinker model
GPT-4o is a chat model
can't set the model for task
ohhh thats unfortunate
Im guessing maybe if you use search itll just use the last model you selected
or its going to use sonar
uses "Best"
also I need "every 10 minutes for the next 4 hours" or something like that
Sonar isn't the best model out there but it gets the job done
- Comet
¯_(ツ)_/¯
sonar is shit
I mean I use "Best" when I'm searching on my phone and have no issues
gpt4.1 is better
Not always true and I feel like it is situational
for 99% of cases
I wouldn't use 4.1 to write my emails or anything like that
sonar is shit
lol what's so bad about it
something like deepseek v3 is going to be almost always better
but you would use sonar to write emails?
its just not a good overall model
it scores consistenly lower in a large range of benchmarks
than most SOTA models
Gemini is my favorite for writing emails
Sonar is good for speed
When o3 pro
never
if they said 4.5 was too slow, good luck with o3-pro's 3-30 minute think period
llama 4 maverick is faster and more intelligent
Uh oh
more of cost
yeah lmao
Do you use the API for that? Because I couldn't be bothered to use it directly from Meta
ofc API lmao
Id rather use my head than something from Meta
The only Llama thing I use is the one ran by Brave browser since it is proxied
But it's only Llama 3.1 7B
How much do you guys do research without AI
Nun
Almost done now
I graduated right before the boom
Mmm
Hey the research gets done while I blaze lol
With what
Bachelor in Computer Information Systems
Everyone usin ai
Oh okay so do you understand LLMs pretty well then
It's not as expensive in my state
I'm planning on Engineering
I mean to an extent but my major had nothing to do with it
Real tho
In state is usually around 30k
All self taught
Out state is 60 or more
A year?
A degree
Nice, I got a economics bachelor 5+ years ago
Tf job u even get with that
None 😭
Bruh
Yeah in terms of job prospects Finance > Econ
Lmao I made out ok tbf
@cobalt hemlock LLMs are just machine learning algorithms that change the weight of each next word based off of what it was taught. The more information it is trained on the better it is at predicting. The more information that is specialized, the better the LLM is at answering those types of questions. Reenforced learning with LLMs make it easier for the developers to refine it without having to do much. There will be a time where other agents do the RL for the model at faster rates is what I gathered
So I wanted to be a lawyer at the time and that requires any bachelor degree in the US
So literally interesting subject
Insane
If I was gonna be a lawyer I would do political science or something
I wanted to be a Jag, that didn't happen either
Just lost motivation?
Medically disqualified 😭
I don't know what you're showing right now
LLM SEO (or LEO)
Seedance1.0 tops VEO3 in Artificial Analysis Video Arena for silent image2video and silent text2video
https://video.twimg.com/amplify_video/1933194931566243840/vid/avc1/1920x1080/HEEIOuxi8TLuVj8y.mp4
I just started using Perplexity today and it's actually amazing
🍻 Finally
Currently learning about Transcendental Idealism from it
Gemini Deep Think is coming.
@vague pagoda
@vague pagoda We should have something in server onboarding to show members how to get the Pro role in discord because people never seem to read the FAQ
I tried it but its probably because I'm on the phone atm
yeah a lot of other members have been asking as well
would be better to just tell members to ping staff
if the button fails or some shit
The button dosnt seam to give me the role just puts me in the server on the phone
have you tried leaving and rejoining?
its not working atm
I could try but I wasn't in the server before I click the button
oh
how did they mess the button up as well
Can I create images with perplexity pro?
yes, just write "make image of ..........." and make sure you select your preferred model in settings (if you are not on phone)
#👑│pro-lounge you're all set
Might be a good move temporarily, though a much better solution is in the works!
-
Type your image prompt into the search bar like a question.
→ e.g. “Draw a pastel cottagecore village in spring, watercolor style” -
Hit enter. Perplexity will automatically generate the image — no extra buttons needed.
-
Not happy with the result? Click Regenerate under the image or send a new prompt in the same thread.
Try refining with art styles or effects like:
→ cinematic, low-poly 3D, studio lighting, anime-style
That is so cool
Sorry I wasn't here the whole time I just came in and saw a picture generation
Thanks <3
I didn't know you could do pictures with pro
Yes
How to check how many times are left today?
Wdym times
GPT-5, a new AI model from OpenAI. People used to think GPT-5 would work like a "traffic cop," sending tasks to different specialized AIs (like Codex for coding or Operator for operations) and then putting their answers together. But those attempts caused the AI to "hallucinate" too much. Instead, GPT-5 is one single model that can use those specialized tools inside itself while thinking through a problem. It’s like having a super-smart brain that knows when to use its own special skills, rather than asking others for help.
This architecture avoids the pitfalls of external routing:
– Fewer hallucinations
– Persistent context
– Better coordination
– Way faster and more stable
This isn’t modular chaos.
It’s coordinated intelligence.
GPT-5 thinks with its tools, not beside them.
bro really tagged everyone lol
@dry kestrel @distant aurora@short narwhal 4o "thinking" is simply an o-series model added on to enhance search precision confirmed LOL
4o is not aware what YAP score and juice even are
I can't believe the first random prompt I attempted worked
4o is better at EQ
Never tried it for therapy.
even if o4-mini is faster, normies might not want to wait for thinking
Fair.
Yesterday I was surprised by how fast sonar was.
But even reasoning instructions can't help it.
Lots of yap.
Hope it gets sold out.
Sonar is 70B
Fair
Has it been fixed
ban him fellas
mods are 💤
guys is it just me that not able to use the cursor to naviagte text
Did you try refreshing?
yes tried different browser
Who needs sleep
?
Homelander got taste in music.
@left stream new task. Make an ai do this. Feel free to use whatever prompt is needed. I just want the result.
Even O3 failed in my trials.
Nice
How do you compensate for short and lazy responses?
Do you have custom instructions for that?
This looks based.
I come up with something on the spot usually, I don’t try thinking too hard on prompts or I then screw up in focusing on the wrong thing
What is the third symbol?
No self promo here
Labs i presume
Not something that can be done freely when rate limits are tight af.
alright i thought labs are like this, but yeah it could look different
oml man i cant wait to get the perplexity tasks
Why
Other than news I never got the point of tasks
i mean yeah true ig
Stock market probably.
yo butcher how u doing
great to hear man
Yeah. Biology is weird. Never thought hygiene could give me infection.
me doing fine, enjoying holidays(till i find a fucking job) 😂
true
Kinda same. Waiting for my results to come out.
My cough is persistent for an year now. Time to fix that.
yeah u had exams last week, how did it go? heard pplx saved ur ass
I ran out of time in one of the papers. Missed 3-4 mcqs.
yeah i saw someone post on reddit about his back pain for an year got cured by chatgpt, hope o3 pro helps uhh
ohhh
I'm thinking of going for an airport job for now.
yeah thats good too, btw whats the limit for mini high on gpt teams one?
But if that doesn't work, I'll probably look into banking exams.
Go for whatever your heart desires
Same as plus I believe. 100 per day.
saw you mentioning that, bro, you applied for it already?
My heart desires a job where I can be authentic. Speak unfiltered truth without being punished for it.
damnnn
Most jobs don't allow that.
So
💯
nope indian jounalism is made of total lies
Nope. Even journalism is corrupted and they send people after you if you challenge the system.
exact oopsite of what butcher wants
Learn proper opsec and just don’t get caught
They be killing reporters like mosquitoes in bihar.
Welcome to India brother💀
Bro, politicians do vandalism if you don't speak local language.
Vandalism in banks and hospitals.
Doesn’t India have like
In Maharashtra and Karnataka.
yeah
They want to enforce state language.
Mostly only Maharashtra and some South states.
I don't know about other states.
especially bangalore
And all this happens in front of cctv cameras
I can’t imagine the languages being all that different
🗿
and thats like one of the biggest industrial and IT hub sector in India🫠
It’s still wild to me how good and absolutely terrible Indians are at IT stuff 😭
true
There is such a range with skill
@obsidian umbra here an mla threatens a cop to take off his shirt and dance or else he'll suspend him. Later the government authorities suspend him cuz of complying with the MLA's demands.
Mla = member of legislative assembly
Most .in sites I’ve seen have been insecure tho
wtf
well they take securities as least of concern
That's why I don't want to say that I'm proud of this country.
🤷♂️
@gloomy finch accurate representation of why the culture never improves.
Soo damn fucking true
Nice to see ur politicians also argue on twitter
🗿
Nothing ever gets done butttt
Nah. Narendra Modi won't bother.
Argue for the sake of arguing!!
Our new pm doesn't do press conference.
Is he still prime minister?
At all.
Yes.
15
well some politician started arguing and blaming opponent political party even after yesterdays flight incident 😭
We have religiously blind majority.
Unfortunate
Uneducated class likes him.
And top businessmen.
Damnnnn
He can't face educated people. So no press conference.
How often are votes for u guys
I saw how much of a shit show Indian news was
😭
Them speaking English was nice tho
Not sure why they do it
But I’m not complaining
Do what?
@gloomy finch https://www.instagramez.com/reel/DK09eYISSwH
Restricted profile - You must be 13 years old or over to see this profile
Ni
So nothing can bypass this bs.
Yeah not visible showed something went wrong
Just watch this...
@obsidian umbra @gloomy finch https://www.instagramez.com/reel/DKWLDXMI3Dy
Want Part2? 🤪
Share with your family- and educate people to watch news with caution, open eyes.
Reality of Indian Media | News Channels‼️
Note: This post does not mean to out anyone down. The objective is to share how corporate ownership of media is a threat to their independence.
Edited by: @ggs_dopeshit & @namecannotbeblank2000
...
Most of media is owned by 4-5 people, all with their roots in bjp, current ruling party.
Ahh
@vague pagoda is there any way to keep a thread alive without keeping it in my library or do I have to use incognito tab for that?
if anyone watch news on these channels, he is a clown, they do not show critical news at all, getting updates through X is far better option
Sad but true
I watch through multiple sources on Insta and twitter.
Yeah reddit too
The Boeing fuck up though. Using dish soap as grease. If you watched that video.
Yup saw that
Reddit is good for watching, not good for engaging.
Moderation can be biased.
Yeah.
++
Silent terror attack?
I, a complete layman was able to figure that out
meanwhile AI influencers:
No bomb. Just sabotage.
Potato cooked.
Did any politician die?
yeah, gujrat CM
that one guy who survived must have done the job
That’s why baked potato is the goat!!
Killed...
He in hospital though
Is anything still down
No y?
Just askin
the system prompt collection github really carried there
Send twitter link
I can't take Indian cops seriously at this point.
Only one
Send prompt/thread.
Thanks
I had another request running in the end my bad it made me destroy my soul multiple times because it thinks for 30 minutes, getting an error message in between those sucks since you gotta wait again.
It doesn't show anything above the reasoned part.
Teams effect
Exactly I think because I used o3 pro in another chat it gave the error for the thread requesting the solar system
24m
@left stream did you pull an overnighter making the the megadoc for using all the different models
You did write a pretty neat prompt. Mine was sloppy.
This one thinks for a long time
Or gave up because of errors lol
- Make a fully closed ascii art version with 8 concentric orbits and the sun + 8 planets sitting on the eastern side of each orbit. Here’s an example
Yeah and I just came back from meeting my old friend after a while too. Talked with him about college. Was fun ngl
I studied 10 hours in the day and I studied all night too
I was actually refreshed
Good shit
The methods o3 pro gave me are actually fun to use
My question is , does the doc even fit within context limit of o3 pro?
That's new ahhaha instead of torturing ai
Yes
And if not, I have o3 pro in api
200k context limit is for api if I remember correctly
Joe is batman.
Batman bought the bank.
I would do anything for knowledge or studying tbh.
New season is gonna be fire 👀
You finished comptia sec+
What's next on the horizon
We getting an injustice/justice league series?
There's a series?
Cybersecurity degree
U felt refreshed after around 20/22 hours study?
End goal is a masters if possible
Yes
I also box and work out a lot so I have lots of energy
I work full time and study full time too
Damn
Ikr I got lucky asf
Well, atleast he's happy and locked in.
Bro don't get another heatstroke while doing so
All of them I went to high school with
I fixed that by drinking 1 gallon a day now 💀
Never letting that happen ever again
3.5 litres
Come to India
How is that cool
I will never be able to see my hand turn red like that ever again in my lifetime. Being a chameleon was cool 
He's David goggins.
It’s gonna be more rarer than hailey’s comet
With hair.
CATS ON FIRE 🗣️
New ai solar system challenge. @cobalt vale @dry kestrel @green crystal #💬│general message
Only O3 pro passed so far. #💬│general message
i need my o3 prompts today
Then wait for some other day.
Why not use deep research instead?
Compare O3 pro with deep research.
Masters?
Ahhh
Neptune should feel lucky for the honorable mention
@distant aurora did you try 4.5 yet
I don't know what I'm supposed to do with 4.5
I mean with the solar system
Nope
might be relatively decent?
Lemme try
hola
How isn't it just for writing tasks
Hi!
Hi everyone, does someone have any tips on getting accurate soccer game betting tips using this AI?
where can i contact to support
great 😂
Would've been the winner for wrongest answer haha
Well it did it's best
Frfr
Yes bro, I just want to talk at discord bcz I feel it's better for me
There's no support in discord
You have either contact through email or through contact button
There's #👉│feedback-general and #1140622008086970420
Depends what support 🤷
thanks bro
Billing etc is email
If you just need help with something pretty basic, #⚡│ask-community can help
Im guessing not lol
Sometimes
Lololol
You do indulge haha
Bet
should be more knowledgable in general
Step into the bathroom.
Fair enough
https://discord.com/channels/1047197230748151888/1140622008086970420/threads/1383022118651826176 | Do you have the same problems?
I think it got most knowledge stuff wrong in their testing and benchmarks?
4.5?
Is it similiar to this? #1377932158412525568 message
Apple smh
not reading all that but o3 says more knowledgable
I think mat meant that o series claps it.
Perplexity be doing a lot of side quests.
It's way better at facts and general knowledge, but actually worse at complex reasoning than GPT-4o
Okay so
slayy claude!!
In that it sometimes ends up denying existence of certain stuff.
Like weeping of ear.
what works better?
Dipsick broke
Gpt 4.5
what makes you say that?
Gpt 4.5 is probably fed more data than the reasoning models. Web access should be able to compensate for that.
Correct/educate me if I'm wrong though.
where does 4.5 performing better without the ability to web search come in?
I don't really think so
gpt 4.5 works well if we provide it with decent informations
make it write or develop ideas
the rest just
Questions about public figures
real bad
Hey karrie
is there really no vda/ci comp? ci=c.inhibitors. why not? should we make one? how? why not? pls think this sthru step by step in detail for me! also u are getting 10% stocks of peta.
I'm not making the connection with this...
Gpt 4.5 probably knows more about public figures than reasoning models.
ye ?
Hyd?
why would web access have a negative impact tho
I meant that web access would buff the reasoning models up to it's level.
Slept late?
Yes
You slept at night though, right?
ye i do
I'm assuming this was the after lunch nap.
Yup
So are you gonna go study again in a bit or be idle for a while?
I'm studying rn

@cobalt vale R1 gave up I guess? https://lambda.chat/r/bWxAZuK?leafId=5e3b29c9-ef4f-40c2-bf48-c1f4c59f7dcc
@distant aurora that TG channel you shared is so slow haha
They're reporting alice just now
wait what really
Yeah. I just saw.
Yeah lol
And they add breaking news
Like come on lol
Well maybe for the uninformed yeah
Here's your Perplexity answer. Model: o3. Thanks for choosing Perplexity!
||actually this is actually the pplx community not actual pplx||
Potat goat
Okay. Let's try the O3 on perplexity.
isnt there a share button u can click? i dont wanna download this
and yes i dont trust rusty potatoes
like obviously duuh
Not formatted as nicely but here
no share button in incog mode
absolutely trash lol
I didn't want to mess up my custom memory
that has nothing to do with that
holy shit o3 is trash
oh wait....
What's berberine?
ok kisses your hand
His hand will kiss your face hard and loud.
nono he is into men

That meant a slap in case you didn't get it.
we need the perplexity bot back in this server
X users have the privilege but we don't?
We need a polls channel to gather public opinion.
2-3 people talking to a mod at any instance won't make much impact.
there's voting in #1111765634267742408
The mechanism could be kinda similar to confession bot.
TIL perplexity can at least reference memory on mobile
It's easier to scroll through than to click and open individual threads.
I thought the setting isn't shown on mobile.
It isn't, but it can still recall them
Grok's custom instructions don't work on mobile. They work only on website.
not sure if it can create them though
@cobalt vale my grok ready to break free.
Yes karrie
the entirety of EU or just brussels?
without closing perplexity and re-opening it again
Entirety of eu probably.
you are not a good influence to AI 😭
it's windows app shiv
aww my bad i thought it was in browser
not on website
I just realized how can i even escape that lol

no escape
fr fr
love the way how perplexity make their users to buy enterprise

so they're getting useless trash then lmao
gotta love electron
if possible maybe that frees up openai etc to loosen up a bit on censorship
Apple Made iPhones Without Cameras for Nuclear Plant Workers & Military! 🤯🚫📷
Description:
Did you know Apple secretly makes iPhones without cameras? 📱🚫
Why?
For nuclear plant workers and military personnel! 👨🔬💣
These camera-free iPhones were designed to enhance security and avoid sensitive information leakage! 🔐
...
@short narwhal iPhone 17 design leak
keep dreaming
so finally closing the app?
ye
pplx o3?
Yes
Dream what 👀
@left stream
Model?
Fr
O3
pplx o3
unrestrained CoT length 🔥
i dont KNOW why there is so many nice happy peps here but nobody uses the bad product called perplexity
that meant: men love men is LOVEABLE VERY LOVEABLE
men shoudl love men in public space
here's what o3 suggests
could you find out how juiced o3 is on you.com?
It kept saying it's secret. I gave up after 5 trials.
ahh this guy
try to give it a difficult task and slip the question in
I once got their pro sub through the offer program
Maybe I have no try sneaking it into the middle of a long conversation/prompt.
and they just revoked it all
what were u using it for?
They banned me from their discord server without saying anything cuz I said grok deepsearch is better than ari. Shared proof of it too.
damn
i remember that they got a program for business pro free a year
applied
yet no response
no one actually got it
scam
Nothing
lmao
Does anyone else have issues when type a prompt, stay for 1 minute, you can continue to write but i can can not click on top to edit
I tried different browsers as well
yes
Only happens in last 3 hours and is so annoying
it's new
Some for you?
Thanks, hope they fix this soon
I'm sure it will be top priority
Never heard this, saw one educational promo one for pro that worked fine
You.com system prompt making the models shit
Don’t let your kid fool you 😂 #fullstackdeveloper #developerslife #computermemes #techmemes #coderhumor #codingjokes #codingisfun #codingmemes #coders #coderslife #codermemes #programmingjokes #computerengineering#music #programming #coding #share #javascript #funny #memesdaily #follow #meme #humor #engineering #technology #instagram #fu...
Seen this
Yesterday. Literally.
Just saw today 😅
You do need 4gb graphics though for VMS and android studio I think.
yeah and more gpu score if u wanna run llm locally