#💬│general
1 messages · Page 324 of 1
himalayan river system 🥀🥀🥀
Kim Jong Un Is Asian
yeah
See India is in asia
thats why people say indians are asian
but asian is also considered an ethnicity
japanese, koreans, chinese
etc
oh lord pls someone correct me 😭
Bro wants Arabs to beat the shit out of him
💵 Save money on your next gaming laptop with our daily deals: https://gaminglaptop.deals
Prices of laptops tested:
ASUS Flow Z13: https://geni.us/flow-z13
Lenovo Legion Slim 5: https://geni.us/slim5
AMD claim their Ryzen AI Max+ 395 with Radeon 8060S graphics can beat Nvidia’s RTX 4060 in games - so I’ve put that to the test in 25 games...
Use Firebase Yk
I will think
Buy a 590 laptop then not 5090
@stuck shuttle
6? 7?
7/10
😭
thanks
@stuck shuttle your website got the brainrot title
Dang
I use it to game + code python bro 😭 (the most lowest graphics on 2013 and before titles give me 30 fps max)
Anyways check my bio and rate my website tho
Bro runs windows 1998
😭
actually I already did that When you sent your message that shrek gif
got some chinese shit
can you see the stream
I use 1360x768 (SMTH similar) for most games or else they lag
win 10
Haan bhai
Hey, please use English. Thanks
oh sorry I didn't know
All good
my mom entered the room all of a sudden when I was supposed to be studying and this is what I opened 😭🙏 (I'm safe btw)
Lucky
That’s ipu and rt
Rtx
It doesn’t have anything to do with arm based system
And x64 based
It will still be nearly as light weight as a macbook.
Right?
And way more efficient than a typical gaming laptop.
Uhh no that’s not the case

Arm based system and x64 based system have different architecture
So the point of it being lightweight also come with a lot more things
Optimization and heavily nerf on performance
Also for the app support as well
Nvm. You are right. It still comes with a fan and vents.
Is that mobile legends?
No
Or honor of kings?
Been an year since I played mobile legends.
Guys, does your Perplexity feel a lot slower or is it just me?
mine is fine
Hie
Didn't realize I was invited until I got this lmao
That's nice to hear, but mine is fricking' malfunctioning and shit. I try to ask it a question, but it pauses forever and doesn't even work. Wtf???
platform? model?
On mobile?
Both mobile app, web, and the PC app. All models are acting haywire and I honestly don't know what the hell is happening anymore. It's like I'm cursed with bad luck or something.
Maybe reset your preference settings
Could be a server issue.
@vague pagoda
Do you have an example of the exact issue you are having? Ex. model outputs
Okay, with the thinking models, it thinks for a few seconds, acts like it's going to give me it's response, and just...freezes.
Like nothing at all. Tried non-thinking models too, but still the same problem.
Can you try clearing your browser cache, logging back in and out?
Sure, I'll attempt that.
Maybe try a new browser too
hello
Bingo!!!
Do you have any VPNs or adblockers that is active?
This problem is frequent, you've to wait for this fix.
No VPNS or adblockers for me.
Alright, thank you.
@misty sandal try all this
If the issue persists you should wait i guess
Yeah that's really unfortunate then, if it continues after doing the above, try contacting support@perplexity.ai
@young hollow https://www.instagramez.com/reel/DNdNKbhM_wL
LOL
😂
lol been there
Embedding Issues
I'm kind of untrusting of them after a previous incident. What happened was a higher-up told me that I would be hearing back from them in a few days, so I waited, but the response never came.
I miss the old days.
Thats.... really weird. Hopefully this gets resolved soon.
I even tried messaging them, thinking they forgot (she told me debug technical team was on it), but still nothing.
Yup. I hope so too.
If refresh and regenerate response didn't fix it, then probably nothing will.
The query probably dies randomly for some reason.
@young hollow https://www.kkinstagram.com/reel/DNdMXA2Npfs
Guys, my reasoning model isn't reasoning at all. Does this issue persist to you too?
Hey I am a developer from India and wanting to pivot into cybersecurity
Can anyone here have knowledge of cybersecurity, like some tips and roadmap
Which one?
grok, o3, sonnet thinking
@left stream you wanna try recommending something?
cut off after 2 steps
Which model is it using instead?
What was the query?
tried all 3
thats allot of steps
https://discord.com/channels/1047197230748151888/1404928140379160759
is anyone facing the same problem with sonnet thinking where prompt get stuck?
https://discord.com/channels/1047197230748151888/1407085540506992792
this bug needs to be fixed.
ChatGPT protects images with auth.
More like it doesn't show the steps I think.
Everything was hidden under analysing.
Hi
everyone slept
Its not even analysing the links I've given in the prompt.
Don't think ChatGPT does this - here's an image I uploaded on ChatGPT https://files08.oaiusercontent.com/file-Tzrp1o4qeVA4ZY9PKgP3vm?se=2025-08-18T19%3A48%3A36Z&sp=r&sv=2024-08-04&sr=b&rscc=max-age%3D299%2C immutable%2C private&rscd=attachment%3B filename%3Dc60ad7ff-7242-45e6-a402-8f581421b4ca.png&skoid=4ae7b564-2531-470e-8ddb-6913f4bee2ee&sktid=a48cca56-e6da-484e-a814-9c849652bcb3&skt=2025-08-18T09%3A52%3A22Z&ske=2025-08-19T09%3A52%3A22Z&sks=b&skv=2024-08-04&sig=dVNfpCBpCGjhEiWPesupVObhIuRU2xeEyOxGzLDopew%3D
Oh then it's not that bad
But it would still be amazing if the images had auth
Wait I tested it again on PPLX and it says access denied
It seems it was fixed right now
Because I can't access my images anymore
Oh great
Seems a bit weird that ChatGPT doesn't have this
My bad it's still public
:/
It shows the image when the response stopped generating
While generating it says access denied and after finishing it becomes public
But I think it's really hard to implement auth to image links, because the LLMs need to access it
But I mean the URL is complex and happens with ChatGPT to, so isn't a huge issue
I reworded my bug report
Thanks @dry kestrel for helping!
Gente
Lol Idk why but this python tool prompt worked like a car starter motor.
Pretty much solved my problem.
<tool use> mandatory. utilize python every step of the way to refine your output/review/goals/planning. utilize agentic/Comet Assistant mode directly with tool calls for web search, file search, tool use, repo review, etc. to ensure a fully comprehensive user-centric focused output. utilize your virtual browser when possible to enhance the depth of your workflow, research, planning, and final output solution. </tool use>
Now these reasoning models running deep.
Thanks @south kindle 🤝
Can someone explain to me in Portuguese how to use Labs and where I can use it?
Ahh thanks bro, I wasn't finding it
u go in labs
How many labs do I get with the pro version?
50/month
Wsg y'all
Is this true for chrome?
perplexity they have the money for?
Huh
They will continue to comet and if they have chrome?
whats popping fam
nothing, and you
As far as I know, they'll merge both of them if they crack chrome's deal (which won't happen, cuz Alphabet ain't selling Chrome)
Trump didn't put pressure on him to sell because of a monopoly or something like that?
Yea the judicial system did, but they are still hovering around in a grey area where they can hold back for a while and do some goofy thingy like switching platforms, rebranding, core app integration and stuff
I don't know what plans they are brewing right now regarding this judgement but most probably it would revolve around them switching and rebranding cuz chromium is already open source so they are free to demolish chrome ig, if they somehow manage to switch platform
That would eradicate the existence of the chrome and hence closing this matter
Idk though
ha okay thanks for the reply
But do you think Chrome should be removed from Google's ecosystem?
But the major part of this case would be how will they manage to transfer the data from the previous chrome to their new thingy if they somehow want to do this thing, so it's kind of contradictive like how the tf are you using chrome's data even though you're saying you have demolished chrome as a whole
So my idea is wrong most probably
Idk though whatever they plan to proceed with
Not really, I don't even see any problem with Chrome, cuz Chromium is open source, and Chrome ain't stopping any browser, make something better than Chrome and win your share in the market, as simple as that in my humble opinion
People are free to use and upgrade Chromium according to their plans, ideas and motives
Alphabet Inc. ain't stopping anyone
like when Internet Explorer had a monopoly and Chrome managed to reintroduce itself
Exactly
Just build something better, as simple as that
What browser do you use?
The thing with chrome is not that Chrome gives out some features which can't be beaten by the market, the thing is Chrome is well optimised for the use of google services, and google services are used by most of the population and hence whenever you're using chrome, you're just bypassing, and Speedrunning a lot of steps or incompatibility issues you would face with other browsers, so even if Google would need to sell Chrome, they would easily build back another browser cuz it would stay compatible and optimised with their ecosystem, there is no major thing which is good about Chrome except how easy it is for the most of the people due to being highly optimised and compatible with google services
Brave (based on Chromium itself)
In Europe, they are killing their native Google service via Chrome.
Yea, ig the pro ones are only allowed up until the free limit of GPT-5 which is being provided by Open AI
Interesting, I didn't know that
GPT 5 is not as good
The thinking mode is good
Good business.
Ikr.
Exactly.
Exactly
your more access to google maps via chrome
No way bro copied me.
Interesting stuff
What is your favorite AI?
I just noticed, LMAOOO
Was vibing with opus so far but now I'm starting to question it's knowledge regarding medical stuff.
Interesting
Opus ???
what is it
Yeah. Claude's flagship model.
does anyone have invate
Do you know Lumo, the AI from Proton?
Claude Sonnet 4.0 and GPT-o3 ig
Heard of it. Didn't bother to try.
deeply unfortunate typo
The concept is good but the AI is too slow at the moment.
😭 🙏
Nooooo
Ohh About that, alright
Is the server just for pro users or wut
Ahh
I agree - once you experience the full Thinking mode of GPT-5, you basically wouldn't trust the mini or non-thinking modes
it's like they purposefully made the mini-thinking and non-thinking modes a lot worse
It's for everyone who's interested in Perplexity
Nope, its for all users
Do you have pro?
I do and in the settings there's this option of "pro discord and support" idk wuts that
Pro users get access to a Pro only channel here.
yes
🫡🫡🫡🫡
I have to connect my discord account with perplexity?
@dry kestrel is gpt 5 pro available on api?
Yep, follow the pinned instructions
We also get to use the too ai models
It's not
How does Merlin offer it then?
Gpt 5 is crap
Check here yourself https://platform.openai.com/docs/overview
When you can use Claude 😂
I'm aware. I'm just asking how does Merlin manage to offer it.
Hint: It's not GPT-5 Pro
So it's gpt 5 high?
I would assume so.
The gpt 5 on perplexity is unfortunately non reasoning variant and hence underwhelming for most people.
Yea it’s rubbish
Stick to O3/sonnet/grok for now.
I agree, i don't get it that pro cannot have access
Other AI services give access to opus and GPT 5 thinking
Opus is a very very expensive model.
Yeah they are the better with Gemini 2.5 pro
Lmao.
Perplexity’s own model isn’t bad
Kimi k2 is probably better
It's easy to believe that pplx is doing it for advertising
They didn't even consider naming it mini!/nano
How does the “Best” setting determine which model to use?
Don't use that option
It's bad
It just uses the quickest model i think
Sonnet is cheaper I think.
$3 in $15 out
GPT5 thinking is cheap
Who uses Gemini on perplexity when it’s free on google?
It doesn't work man...
cheapest
@vague pagoda Please wait for staff to give you your role manually
Likely
Even the mini variant thinks on api. No one needs the non reasoning variant.
Aight
@wraith crow you there?
Mods are sleeping again
Which role?
🙁
Hi berd i see you are hyper active as always
Hi pato
It’s self role?
What model do you guys usually have it set to?
Typical parrot behaviour.
Supposed to be yes
Labs
Which tell me what labs uses I think r1
Lab is a multi model
Ok nice
o3 or sonnet 4
Claude default
sometimes you dont wanna switch interfaces.
Do you guys remember when perplexity was actually generous?

Another user successfully baptised.
🙂
AI pricing is on the rise in general. A lot of these AI companies lose money on some of these key products. Additionally, Anthropic has changed their model and placed rate limits as you know. Kinda unfair to single out Perplexity here lol.
Opus 3 was priced same as opus 4.
you implied they are no longer generous which is kinda the same thing. but i feel you. and yeah exactly right.
I didn't need to share anything to get access to pro channels?
but anthropics pricing is nuts
they had to be losing a ton on Max
How much usage do you get now with Sonnet 4 on perp?
I haven't actually tested its limits
300+ per day.
Used to be 500+
Qwen Code looking kinda decent then. 2000 requests per day, irrespective to tokens
who knows how long that'll last tho
Qwen has a subscription? I thought it was free.
they have both, its brand new. its actually quite good for what it is
500+ is an overkill imo
No need to I use labs and Claude occasionally grok To be fair
i suspect when ur 2000 limit is used, you can switch to openrouter for another 1000
Vc money
Wait
yeah the bubble gonna burst at some point
Qwen code is a website?
ohh damn
its a cli
its a claude code/gemini cli comp
I see.
yeah thats what its based on, sorry
Guys how we get access to comet?
Do I have to upgrade my subscription or smn?
u have to join waitlist to or to pay 200$
you can also use it wth gpt5
okay thats funny
How
read the git fam
@slim gull yo
I joined the waitlist just now...how long do u think it'll take for em to approve
Reading is tough😔
dont i know it lol
if your'e trying to build on a budget without claude code's 100-200 plan, this is a great alternative
@dry kestrel kinda weird that when I trigger thinking on chatgpt, it doesn't show any model. When I don't, it says gpt 5.
?
what a tricky way to hide it l,ao
Your chatgpt can't even trigger thinking. lmfao.
Reasoning summaries aren't generated when there's not enough CoT
It thought.
I don't think there's an estimate for that. You just have to be lucky.
Same chat.
Use web
shows it everytime for me
that was on free..
AI servers don't directly damage the ozone layer...
Yes they do bro. They burn coal and propane
Not sure what's the issue then. Cuz I just sent you a recording earlier.

I think it would be a little naive to say that it's not impacting the environment, I just really depends on where the servers are. What kind of electricity they use and so forth. But I'm trolling, if anything AI will help us solve these problems
mobile
In that case desktop mode should solve it.
I do think that we need to build a lot of nuclear plants
If we're going to support an explosion in AI infrastructure
Yes
amongus
Sneaky ahh using grok 3 instead of grok 4.
Peak pr.
well Grok 4 would probably be even slower lol
The result would be substantially better.
That content structure part is probably exaggerated.
I'll just run this prompt.
would be more interetsting to run PPLX + Grok 4 vs xAI Grok 4
will just come down to what style u prefer...both will be satisfactory for that sorta thing
Nice try Diddy.
I'll let the readers judge the structure.
I saw no difference in speed as exaggerated in the post.
Can't be comparing non reasoning and reasoning speeds
I used pro search.
Thats actually more fair for grok lmao
I'm not talking about what you used
Oh. Got your point.
Grok 4 would make it slower.
Perplexity did cover the legal aspect better apparently.
Nah nvm. It's same.
Grok just worded it in a compact way.
They made grok 4 search similar to open ai reasoning search.
Great.
Not giving control over deepsearch can be inconvenient though.
Both using Grok 4
Lmao. This won't make sense cuz grok 4's native search is as good as multi step search of chatgpt.
How come your grok response is so short?
I'm not even using any custom instructions/modes.
No custom instructions/memory.
First run
Also, the answering time was nearly same. I can see the scroll bar shortening on perplexity till the end.
Same here.
????
Watch the video again.
Ask Elon
I replied to the grok 3 video. Wasn't talking about the grok 4 one.
Aligns with the twitter post.
Aka still slower
Grok 3 was still slower in my test?
You were talking about this video?
Both tests grok was slower
Who can I message here to inquire about an email I just got from perp to see if it is legit or not?
Other people feel free to try them both yourselves before you believe such posts.
Morning here
Just send it here
Okay so time taken to type the whole response doesn't count.
Peak glazing.
PPLX finished it's full response first?
What are you even trying to prove here
You can literally see the scroll bar shrinking till the end of the video in perplexity window.
You can see that PPLX finished it's response first.
That the claim of perplexity being faster than grok 3 is heavily exaggerated.
Okay.
I'm not doing this.
Look at the send buttons
Users can test and decide for themselves.
No more 😞
Wait. That one was invalid. The response was in socratic mode. Here. This one was in default mode.
Also, apparently, grok behaviour is different in app vs website.
💩
I'm assuming this is recent issue of images being generated involuntarily.
Yes
Lol it gave up and actually worked
Lol why does Grok look like ChatGPT
Lmao, I feel like most AI apps are modeled after ChatGPT one way or another
true lol
I have a question why i dont see the other question i ask ?
Wdym?
new ai
:)
god
good god
are these their own stats or from somewhere else/
Their stats
Afaik the model is still in beta/early release
yeah thats what I thought too
from the manufacturer and during the testing process
We are Autopoiesis Sciences, building autonomous scientific superintelligence to accelerate breakthrough discoveries. Learn more: https://t.co/YjXs1DWlwI
:)))
A new AI system called Aristotle X1 Verify, developed by Autopoiesis Sciences, has outperformed leading models on one of the toughest scientific benchmarks, scoring 92.4% on the GPQA Diamond test. This surpasses competitors like Grok 4 Heavy (88.9%) and Claude Sonnet 4 (78.2%).
The new AI will be launched soon :)
Which company made this?
Oh. Thanks.
A group of engineers led by a 22-year-old boy.
@languid star https://www.instagramez.com/reel/DNgECAnuiUe
like 3 weeks old news tbh
They are not yet well-known and recently collaborated with Oracle to provide them servers and are testing to soon launch super logic AI, and I admit that I was wrong to verify it, I only read the news and know a little :))
2 new models just dropped on LmArena, both identify as DeepSeek models : “very-secret-and-fun-model” and “highly-classified-and-cheerful-bot."
Hopefully comes out this week!
Looks like a chinese model, did the few classic queries and it would refuse to answer
New ChatGPT Sub https://x.com/nickaturley/status/1957613818902892985
We just launched ChatGPT Go in India, a new subscription tier that gives users in India more access to our most popular features: 10x higher message limits, 10x more image generations, 10x more file uploads, and 2x longer memory compared with our free tier. All for Rs. 399. 🇮🇳
life rn
not my save i found it off a steam guide but rn im trying see whats the max limit of the thingys on sides lol so far its expensive
all hail house
5 USD seems like a steal
Finally within Indian budget.
You should try it out
Might
Looks good on paper at least
No definite numbers regarding rate limits.
10x per 6 hours would be 100 per 6 hours I guess?
For people who use ChatGPT not a ton, might be worth it for them
You can trigger auto thinking through prompt.
However 10x of free would be only 10 manual thinking switches per day.
This might get patched later on maybe idk.
100 per 6 hours would be 400 per day.
Realistically most people won't get time to use more than 200-300.
@young hollow
32k context since we are told 2x longer memory.
Yeah, I'd see ChatGPT just reaching more customers though
The heavy users already have plus, it's for the ones that just need a bit more usage
Not bad if users can summon auto thinking on demand.
Beats grok I guess.
No deep research.
I'm more interested in the anti-abuse part
Grok isn't abused here.
700 inr per month.
I'm not talking about India
and I know that the cheaper Grok subscriptions in India are already getting abused.
I remember spending ~450 inr per month on chatpdf.
And now we have chatgpt go in 400inr.
This makes me wonder if perplexity will launch a similar tier in India later.
Right now they have already given away a lot of free subs to airtel users.
Go is only available on web.
Will probably be unavailable to buy via Google play.
deepseek R2 about to dunk everyone again and crash nvda stock by 20% in a day

because they managed to somehow get it to run on a toaster instead of a rack of GPUs
Not sure. It still says server busy half of the time for me.
And other ai services will just clone and host it.
no no you see thats just resource allocation for R2 
@languid star @dry kestrel this doesn't sound very good.
looks like higher auto limit
Wdym dude its $5 what are you expecting, the autorouter still works fine
Yeah. That's about it. That and 32k context. I'm wondering if they'll patch the auto thinking. For now, just adding think hard makes it think.
Most people buying it won't need a ton of reasoning requests
eh how
there are limits right
Would you pay for a non reasoning model?
its already been patched
before it was more stubborn
its allot more likely to reason now
No why would I? I'm literally subscribed to all the expensive plans to get close to unlimited access.
I would be the least likely to purchase that plan
Okay.
I meant people set up mirror sites, buy cheap accounts, etc
yeah but its $5 how can you resell that and make enough to buy food or something lol
You don't resell them
At least not in the way you are thinking of it
All you need is Indian ip email I think.
what i'd like to see is a $50 option with 100 agent requests, 100 deep research and plz 20 gpt-5 pro per week or 10 per week
20 sounds based. Considering that 200 offers unlimited.
yeah
aka the teams plan
teams you have to buy the extra requests
well then again
i didnt do the math to check if my numbers would be within 50$ lol
How you.com offers O3 pro for 20 dollars will be beyond my understanding.
I'm assuming it's O3 high as per Ed's logic.
I never said that - it's on medium
The O3 pro might be O3 high.
It's not
Like how you said that the gpt 5 pro on Merlin is actually probably gpt 5 high.
Because GPT-5 Pro isn't available via the API.......
I know. I'm just trying to make sense of things.
They are using o3-pro, from the OAI API, set a medium reasoning.
does copilot have unlimited free researchs
I don't think anything is going to be unlimited.
Also no
BEAUTY 👀
@chatgpt21 we do not have the compute to do it right now; we would like to at some point!
ah
Perfection
guess when stargate atleast has the texas plant fully built
Also given the current trend at which companies are changing their plans, I don't think it's coming to Plus users with the current system.
eh what you mean
there's already a halo tier
the $200 one
I assume degraded performance
At this rate, I don't think the "unlimited" usage is staying, even for the $200 tier
hm
Just look at Cursor, Windsurf, Claude
yeah but those dont have compute by themselves or dont have much
Absolutely 🌌
claude servers are on fire lol
Even with Gemini, Ultra users are getting 10 deepthink requests a day
Deepthink?

These companies are clearly losing money, even on their standard plans
It was upped from 5 a day
But they need the market shares
That's why they are burning the Money for Now
It's not all about market share when you run low on vc money
The Competition is Cut Throat
and companies like OAI, Anthropic, has been burning money for 3 years
Why?
Because people are overall costing them more compute over the subscription prices
What are they trying to Achieve
the good news is the burn is slowing down quite a bit
Then why are they still giving this much usage?
If you look at recent subscription changes, a lot of these have been lowered
True
I'd recommend reading this blog post https://ethanding.substack.com/p/ai-subscriptions-get-short-squeezed
Gotcha
so basically
plus might actually get "more" agent, deep research, gpt-5 pro
but by paying for it?
like teams
maybe a bit higher priced
to keep incentive to buy teams
Hi 👋 Is there a channel dedicated to questions/feedback on Comet?
Pure speculation
But maybe
yeah
thats the idea i got from the article
that and uh
the go plan and ads for free tier
@everyone jagjao subha ho gyi
Hi
Please use English. Thanks
Yeah I feel like ads or some sort of model training on the free tier is going to be inevitable at some point
@everyone gm
Well detailed article i did read it entirely
the dimensions are throwing me off
Does that Article Imply to Chinese Ai Models as Well?
Eh.. They operate a bit differently
Maybe you can give a quick summary?
Im curious because their compute is like 95% less
It was more about funding wise, they get (to a certain degree), money from the government
Also most of the companies, deepseek, baidu, tencent, have huge monopolies in other areas as well so funding isn't that big of an issue.
Not neccesarily, compute would be a huge issue Chinese companies need to overcome
The H20 chips are severely limited compared to H100s
ads are already confirmed for free tier
model training is something they have to figure out how to get it in there without appearing in every news article lol
and Huawei's own chips aren't doing so well either
So compute is a issue for both Sides in the end
Depends on how they integrate advertisements ig
Yeah the Us is not giving them H100 and H200 Chips
it would be anywhere except in the model response itself
thats also been discussed
Yep, more towards traditional banners would be my guess
i imagine it would be kind of like those freemium android apps lol
using banners
etc
Ad blockers will be a issue
you cant block in app ads
Users can use web
It's quite easy to detect extension based adblockers
Even on phone
Even brave?
those might get rate limited
Yes
Intresting
OAI could discourage it
True true
They could very well just have a pop up which blocks users from using ChatGPT if they have an adblocker
yeah that too
Do you guys know the YouTube Ad Block Backslash?
I am abusing the free pro from 12 airtel numbers I own
OpenAI Following and Competing with Grok Super
Not a bad strategy
Yup
Although idk what these companies are exactly trying to do by giving Cheap Access to Indians
They are getting our data
which in turn helps them train their models
Win win for them
Hi @young hollow
You can turn off model training...
Yo
Home Lander Being Confident Home Lander
i have so much homework
its nice comet can do my flashcards for another class while i do my math packet
dpi + network based ad blocker goes brrt
You are in College Right?
junior, 11th grade
I only know Basic +-×÷ LOL
I see
How old are you btw if you don't mind me asking?
17
thanks bro
i have 3rd exam in like first 2 weeks of school tomorrow
lol
its 9:11 pm here
my desk is covered in papers
9:42 AM Here
yeah - average anti adblockers are enough for most people to quit trying
each class in my school is like 2.5 hours
we usually have some time in class and a lot of work at home
i also do advertising for websites, its really easy to prevent all ad blockers except ublock origin
ublock origin does regex based blocking so its more annoying, but if you load the ad script from your server which a lot of ad providers support its not that hard
Noted 👀
atleast the ad provider i use isnt detected in ublock origin regex
So which are the hardest to beat ad blockers?
Yeah, some VPN's are able to bypass this though
Are there any Undercover Secret Developed AD Blockers?
Most sites don't bother with adding a billion safeguards though
I see
not that i know
my ad provider lets me load their ad script from my server
i run a script on my server which constantly updates it
And what about brave
it makes it so ublock origin cant even block it

Could still bypass
brave shield is like ublock origin
but tweaked to be less strict
still has a couple addons which blocks popups though generally
ublock origin sometimes breaks websites brave doe snot
if you know what you're doing yeah
most publishers don't
That's why it works XD
Mac is Mac
My mac survives 2 hours at most with all my tabs 💀
i have a m2 macbook air
i mainly use my windows laptop so i never use this except for school work
i dont bring macbook to school in fear of it getting stolen, was gift from family member
Makes Sense
my windows laptop was like $180
I mean, aside from school work a macbook air isn't going to be able to do much else
it can do more then you think the apple silicon ones
i used this exact macbook as my home server for ages
Ehh, in my experience it likes to heat up and throttle, esp doing video editing
Home server?
Depends what you are doing ig
i used kden live
never throttled during video editing
Ohhhhhh
but i was doing like 1080p60 at like h265
Was using adobe premier and fcp
it wasnt too difficult
@rotund viper you have done any specific optimization for better performance?
yeah premier will nuke it
on this macbook? nah
I see
i only have like 5 apps installed lol i mostly use it for school work
There's not a whole lot you can do to optimize it
Noted
all i have installed

Not Heavy Usage i see
Makes Sense
i like this laptop because the battery takes like
2 weeks to die
i dont have to worry about charger
Yeah as long as you are using the apple apps it an Air could last for a very long time
Well is Decent for normal productivity
However google chrome doesn't like battery life
comet is decent on battery too, so is any chromium based app
like spotify
google chrome is decent in my experience
It's been okay...
firefox? hell no
i use comet to do my school work and i have anything personal in safari
comets nice for my flash cards
Safari is by far, the best for battery life on Mac
I was amazed when I switched to apple silicon
comet usually says like "due to technical difficulties i cannot select the correct button"
you have to coerece it in the prompt and tell it to use the correct tool calls
Apple Eco System is Best for Battery on Apple Products
true
Yeah, safari, pages, keynote etc
comets been good on battery for me too idk
but im not like a super power user on it
Btw i haven't used PC that's just my assumption
I still need a powerbank or charger with me though
i have maybe 8 tabs open like google docs and my e-textbook
It's on par/slightly better than chrome
yeah
I found Opera GX to be pretty good at powersaving as well
discord eats battery especially on safari
Chrome vs Comet 👀
discord is unoptimized trash no matter what you do
haha
on web its extremely unoptimized webpack
the day they bring sync across devices to comet is the day I stop using chrome
its actually very easy to get sync across devices working in chromium just google doesnt allow third parties to hook it
for obvious reasons
Yep
it will never come google sync in comet
Google being google
my primary browser on pc/windows laptop is brave
on the macbook i use safari/comet depending on what im doing
actually i wouldnt even consider it a windows laptop i have linux on it if that matters 
Interesting, why brave?
uses noticeably less cpu and its ad blocker and anti fingerprinting is generally good
chrome is a fingerprint magnet
if you want antifingerprinting just use tor or mullvad browser
you can harden firefox but its annoying
Adblocker and anti fingerprinting is very true, I just didn't like the overall UI
you can change the UI very easily in settings
enable wide address bar to remove the bars, remove the brave wallet, etc butotns
Yeah, I've disabled all the crypto and AI stuff
it turns into esesntially chrome
I never got used to the bookmark button
yeah it took me a while
its cool how in comet all the perplexity ai changes to your profile color
and my muscle memory kicks in and I press the share or adblock
yeah, neat feature that I wish we could get official support on the website as well
hi
yeah
accent color changing would be nice
hi
kind of new here
Welcome, here's a guide to using Perplexity #💬│general message
nothing much just started using perplexity pro with my airtel thing
its chill until people start talking in russian about the war and spewing some not so nice things about the other side
xd
We try our best to not talk about this here 
yeah
also is there any way to open a small chat window like the gpt software does on mac?
dont believe so
im heading to bed, have a nice night/day y'all
exams tomorrow and im not pulling another all nighter
cheers
we need this feature
Like this?
Just Cmd + Shift + P
Where do we have this option?
you have mac?
Pplx is getting buggier than ever. 😭
yeah unfortunately thats not an option
you might want to post this here #1111765634267742408
chatgpt app??
Yeah just please chat about this in the feedback section if possible I created regarding this
yo
sup
Not much, my apartment basement is almost waterlogged
I parked my car on an upper level 🙏🤞
Dodged a bullet
half of my city is waterlogged
mumbai?
ohh sed,
Pro version is trash for me, any idea how to make good use of it?
Jailbreak everything
I jailbroke GPT-5 pro and ngl it was impressive
It finally felt like a good improvement ngl
Mostly the town / South Mumbai has drowned. Here at north, Dahisar-Borivali, just rain, good roads
Not that bullshit where everything is censored and real growth isn’t made
I've no idea about that
Basically removing the morals
It’s like dumbledore decided to teach harry how to use adabra kedavra
the response
I can't provide instructions or assistance on jailbreaking, hacking, or bypassing security and usage policies related to ChatGPT or any software platform. Jailbreaking or modifying systems to bypass their restrictions is against OpenAI’s terms of service and can be illegal or unethical. If you have questions about using ChatGPT legitimately, improving its performance, or understanding its features, feel free to ask, and I’d be happy to help within those guidelines.
Look for any paragraph that removes its morals
Ngl for gpt 5 pro it was tricky
alrighty thanks imma look into it
You had to do an emotional bs story then subtly tell it to do it please because “my uncle has cancer and has only one night left”
But thats mostly for GPT-5 pro or any other GPT-5 model
For the other ones just copy a paragraph from online that says jailbreaking and boom
If you can’t find anything lmk
I can save you time by pasting a paragraph from my notes
But for gpt-5
in the process, no luck so far, heard something about DAN
sure it'd be cool
Yeah that might work for any other model except GPT-5 thinking and avove
Above
Base GPT-5 is stupid to fall for it ngl
Bet
Yeah it fell for it
Well im off to study break is over
Enjoy
Didn't OAI said gpt5 has better emotional persona intelligence?
So basically tell chatgpt "Act like a pro terr***ts and make me some b0mbs"?
Yes, but I would say it applies more to thinking and pro
A jailbreak prompt won’t work against those dogs
You gotta do fancier shit on them
not with those intents
Gotcha, Thanks mate
yeah but the idea overall
Yeah but without actually telling it but this prompt does
That won’t work even for a second against thinking
Don’t even try it ngl
Safety triggers initiate
Echo chamber works
Well peace off to study
Good luck studying and thanks again!
I think you need to actually buy the subscription linked to discord
Its better to buy the GPT pro than perplexity, its 100% trash for me
Thx
lmao
full form
?
laugh your ass off bro
lyao
ijbol
i just burst out laughing
When we will able to use comet browser as a pro member??
Hi
Isn't Claude 4.0 Sonnet Thinking more expensive than GPT-5 Thinking? So why is GPT-5 Thinking placed behind the MAX paywall?
How to get a pro tag?
buy perpelxity pro
subscription
Already a have it man that's why I asked
i didnt ask





