#💬│general
1 messages · Page 520 of 1
And maybe that just means I need Grok Code or Claude Code Max
Try out Windsurf
I'll check it out
Overall I'm unhappy with Google in general. Gemini has gotten a lot slower. Perplexity handles 99% of my needs. When I do need a Pro model, I already pay for X and Grok. Grok 4.1 Thinking is amazing and really good at coding questions.
You can also use the Gemini CLI, you get 1000 requests a day as a free user.
Pretty sure it's higher if you have Ai Pro
hi gues
I may look at that but I just cannot justify renewing my AI Pro Membership
so unless Google finds a way to fix things before October bye!
As I said you can use it for free
Antigravity isn't free and not sure I want to keep providing my financial suppport and data to Google
hello
Hi! How are you on this most auspicious day?
"X queries left using advanced AI models this week" Now it's showing up after every 3 messages. I'm done with perplexity. Not going to renew the pro.
I'm not sure if it's your accessibility software thats acting up or something but I never once said Antigravity was free.
I am talking about the Gemini CLI
I'm not feeling well.
@vague pagoda Is there a plan to increase the memory for chats, because I don't do them very often and don't want to refer to them 250 times, but the bot sometimes loses its memory due to the size of the chat.
I communicate with problems to AI, how it reacts, and understand what it would do in my situations, suddenly I am not competent in my life.
@minor shard
You get 2,000 free requests a day on Qwen Code
I didn't have a good experrience with the mobile app for ChinaGPT is the Code product bettter?
I am also looking into smaller coding models that run locallly
My workstation has a GPU
I was responding to another user. I recommend you check out Gemini CLI since they provide 1000 free requests a day.
Ihn terms of paying I'm scaling down. Keeping Perplexity may make an enterprise max account in the future, keeping Grok+GrokCode
How is that calcujlated is it per message in the chat agent I use or work int he background counts too?
1 prompt sent = 1 request
thats a bit more reasonable
Sad that Antigravity is so limited
300-500 pro/sonar searches (which are useless, they suck). 10 AI advanced per day now. Perplexity Pro is useless.
You kidding just 10 advanced per day how the hell they pushed it trough?
This is untrue, I've used way more than 10 uses of Sonnet 4.5 today.
I used around 30 searches today I think. And after some time it gave me couple of more searches
Yeah, it seems like they haven’t updated their own model in years. Really behind, so behind it makes it useless and unreliable
Sonnet isn’t one of Advanced AI models, so you can use it for up to 600 a day like any other AI models.
It’s Opus that you get a free 10 use or so on Pro. That’s almost always been limited after they lied about unlimited and changed it to 50>10 last year.
Someone tested it before, it seems like resume one query after 20 mins when the weekly limit thing shows.
I need help! I can't generate images because it says, “Unfortunately, I can't generate images in your region at this time.” Why is this happening, and how can I fix it? I urgently need to generate images. I've already tried changing the models and leaving it at the default option. I've tried changing my prompt in thousands of ways. I can't find a solution, and I've tried the most common ones.
Guys today I lost my Gardfather
My deepest condolences. A big hug.
🙏 🙏
So sorry to hear that - may he rest in peace 🙏🙏
I don't understand the message
@agile musk is very racist let me know if u want a clip all slurs used n word and f word
also flexed hes super rich @agile musk
voice recorded and screenshot @agile musk
Hi
What is AI Advanced? 300 searches a day on Sonar does seem to be a bit stingy but I've never hit that limit
Whats the different between pro and free
Random (kinda loaded) question ... Are there any actual Perplexity employees in this Discord server? I'm NOT asking for any names or anything, it's literally just an honest question ...
Perplexity Pro unlocks more ways to learn and get work done. Upgrade your productivity and learning with unlimited Research, unlimited file uploads, exclusive Pro Perks and more.
Whats the best use case for perplexity ? Im currently pro user at gemini…i saw alot of models in perplexity…kinda confusing..
FYI - I've canceled all my other subscriptions ... because Perplexity Pro gives you access to all the top models from all providers, and ALSO the Deep Research is the most accurate and methodical I have seen
Research is the best "use case" imo
If you actually want REAL info, and not some LLM hallucinating half-facts ... Perplexity is king
I'm a full stack dev, and have used all the models ... Perplexity is the king for research and accuracy
(imo)
What different bt pro and free
So i just select “best” should be enough ?
pro gives model selection
the difference between "free" regular mode, and "deep research" mode is MASSIVE
Im in pro actually , just dint sure how to use ..
I use "best" for basic searches, and "deep research" for complex questions ...
If I want to know the weather today, I'm not using "deep research" mode... for example
Yeah - honestly, I typically use "Best" also ... the model router is actually pretty good at figuring out which model to pick
FYI - also, if you're a developer, for the $20/mo Pro subscription, they also give you $5/mo in API credits ...
pretty sweet 🙂
So basically just switch between best and deep research and no need to touch on other models ?
that's (basically) the way I use it 90% of the time...
Alright…let me just try using it…
I typically only "force" selecting a specific model if I asked a complex question, and it didn't think hard enough the first time. Then I'll copy my question to a new chat, pick a specific thinking model, and try again
no worries, sorry if I dumped a bunch of info on ya 😉 I've been nerding out on it
Thanks man thats not too much tho
okay good ... here's another 3 hours of info .... 😛 lol
I'll just offer one more thing I've noticed ... you can leave the model picker on "Best" and, instead of manually picking a model, just include something in your prompt like: "Do deep research" or "be methodical" or "think carefully" that will typically encourage the router to pick a better model
The advanced models have pretty limited use compared to the high limits Best/Sonar.
Perplexity isn't intended as a use it to avoid buying multiple pro subscriptions
Pro gives you more sources per search and higher platform limits not to mention access to advanced features of Comet Plus it's AI Browser
Perplexity is a search engine it's not a general AI tool so depending on your use case other tools may be a better option for you
Before this new thing with limits I used perplexity to double check and debug things across all models
Also the space mode is quite sweeet and deep search as well
Asking it to do a critical analysis usually does the trick for me
So I probably will keep it as a double checking tool instead of main tool unfortunately
I use Perplexity for most things. Grok for everything else. I've almost stopped using Google Gemini entirely
My work is very demanding and i always pushing models till its max limits thats where cross test and cross vakidation is the kind. It is a pity perplexity introduced this limitation. As a scientist it is a huge limitstiom for me xD
Gemini 3 is quite good tbh
And today started using opus 4
5 on claud and Claud still safer from weaknes in defending his positions he agrees too much wich is fine for coding but for humanity sciences it sucks
I'm a full-stack SWE with 30+ years in IT/SysAdmin/DevOps, etc ... I've used all flagship LLM models and providers, and I self-host using APIs keys from all providers, plus running models locally, and I can definately confirm, Perplexity is ABSOLUTELY a complete replacement for all 99% of users use-cases ...
what about chatgpt ?
You have access to it with Perplexity 🙂
I've canceled EVERY subscription I have now - except for two:
- Perplexity
- Github Copilot
You can literally do EVERYTHING with just those two.
I'll say this too ... don't fall in love with any model or provider right now ... things are changing so fast, and we are still in the "growing" phase ... what might be awesome today might suck tomorrow
explore and experiment 🙂 what works for me, might not work for you 😉
anyhow rant over - thanks for coming to my TED talk lol 😛
add the third
antigravity
free access to opus 4.5
wait wut
Isn’t antigravity free ?
yes
yeah you haven’t been using it?
I have
i’ll check rn
i did hear something about weekly limits
gemini 3 flash added
yeah claude opus 4.5 thinking still there
But, Copilot has better integration with Github, and I can't allow my IDE to run wild on a production codebase, and just assume everything is just totally fine
(as much as I want to)
yeah - I hear that about weekly limits too
Opus 4.5 is a beast
(for coding)
I won’t lie opus 4.5 is crazy for osint
but, there is NO model on the market that can be 100% unsupervised yet ... but, it's getting much closer ...
.
wtf? gemini 3 flash is non useable ???
Gemini 3 flash downgraded?
Used Pro because Gemini 3 Flash Thinking was inapplicable or unavailable.
@tired condor long term it’ll be a monster. Imagine opus 6 or 10
Good take.
don’t you ever get too lazy to the point you don’t wanna use your laptop?
However, perplexity has been having rate limit issues lately as per some reddit users.
it’s better to use it from your phone no cap
OMG FR. I don't even touch my laptop anymore coz it's too heavy.
I hate it.
That gaming laptop bulk with weak battery sucks.
I use laptop only for Microsoft office type shit.
Or software that I can't run on the phone.
ikr this saves so much time
and you can be comfortable from bed just using your laptop too
how do i set this up
you using remote desktop on your ipad?
I'm at my desk is my special chair 😄
I have never used this feature yet. Maybe I should.
it's designed for neurodivergent people to sit without a normal posture
What is YouTube OLED?
I've never seen that app before
I don't think it's about neurodivergence.
Tablet is simply more efficient.
Large screen, better battery.
Convenient input and output.
Wayyy less weight.
Sweet spot.
парни как подписку про из россии оплатить?)00
Idk why efficient and organised decision making is simply called neuro divergent by people.
@twin grail can you help him?
im tlaking about my chair
not your tbalet
lol
I can sit in weird positions and not break the chair
Would you mind asking in English?
No way. I guess I was more blind than you. 😅
Okay I noticed that now.
I need a good chair too.
But I don't have money for it now.
Maybe I'll get one after getting a job.
Or after renting a room in case my parents don't allow it in their house.
hello
guess ill have to run it in antigravity
hah, more like dyslexic
which turns out also affects braille
😂
Hello I already have a pro sub but in my phone it says get pro , what shall I do to resolve this
try Web login
Are you signed into the same account?
Guys,I am holding a Mafia (Murder Mystery 2 like game) among AIs
The AIs participating will be
ChatGPT-5
Claude Opus
Grok
Gemini Flash
Meta AI
Perplexity AI
Mistral Large
Cohere Command
Qwen
Yi
DeepSeek Chat
Reka Core
If you have any suggestions tell me
Also there Will be 1 doctor,1 Sheriff,3 Mafia and 7 villagers
BTW I have planned the game to be text based and hosted like a show on a Discord Server
i cant even access pro subscription after signing using my student email id
What is Sora?
Une application d’Ia vidéo créé par OpenAI
English please
In web login it works but in android it doesn't
It does bro just relogin in android
Now in web login too it's gone
Lol when did you claim that? Is your year complete?
Bro DM me
Done
Any discord developer ?
Hey am An ACCA student and i am looking for a Discord deveoper who have interest in finance and business and accounting.
DM ME I HAVE AN INTERESTING PICH FOR YOU. AM NOT HIRING I JUST WANT HELP?
Is Perplexity on vacation? His blog and changelog haven't been updated in almost a month, which has never happened before.
@green crystal is this real? https://x.com/i/status/2009734796897931417
i am only getting 10-15 advanced AI queries per day, then it say quota is reached. Is this also happening to anyone?
They're asking about how to pay for perplexity pro in russia.
Yep. Check out reddit for similar posts.
The team said they're working on a fix I think.
thanks, i see that people started discussing it two months ago, guess perplexity doesn't care
Okay wow. Did not expect this 😳
︀︀Grok 4.20 “Granite” casually showing up on Design Arena and apparently being SOTA for frontend? On Gemini 3.0 Pro and Opus 4.5 level? If this holds up, xAI just entered the frontend big leagues. My jaw is officially on the floor. 🤯
Quoting can (@marmaduke091)
︀
🚨 Grok 4.20 is on Design Arena!
︀︀
︀︀Nickname: Granite
︀︀
︀︀I tested it, and you guys are not going to believe this but it's SOTA for frontend! On the level or Gemini 3.0 Pro and Opus 4.5 for frontend. They got the same data as them, it does as good of a job as them, my jaw is on the floor.
Noice
It wouldn't surprise me but what's aardvark
I have no Idea.
Also
@green crystal @dry kestrel @gloomy cipher https://fxtwitter.com/i/status/2009707727405391905
7 trillion wtf
Lmao that's just faking it
ty
😭😭😭
Grok went with the pre training route?
Such a large model with long reasoning would have insane running costs
He looks popular.
But I wouldn't be surprised if health mode uses non reasoning models.
demo showed reasoning
accounts with lots of followers also do it lol
Okay. That's great then.
smh hopefully google stock gets a pullback soon so all this stops for a bit
whats that
interesting
for more creativity - web off
Hello all,
i recently got an email from perplexity saying that i need to add credit/debit card credentials to continue my AIRTEL TRIAL of Perplexity Pro -
I tried to add the details but was met with this error - any fixes?
this is for Perplexity INDIA
please contact support@perplexity.ai
alright
it was valid till?
october 8th 2026
idk mod said to contact support so will try
okay do let me know if you get any help
yep will do
Friend
I still dont understand what are you telling me
Why cant you answer in a full sentence
You need to have an active Airtel subscription for it to work afaik.
i have an active airtel subscription
You can add instructions to the "Introduce yourself" section in https://www.perplexity.ai/account/personalize, should help with its personality
turn off Web mode, if you wish to get more creative answers from the model, in your case Grok
Ahh okay thanks you 2
deep dive ➤ #💬│general message
Okay what does this mean
Anyone Instagram id banner?
Hello
inb4 they'll now start posting how grok 5 made them a crapload of money by trading on TSLA
Is there any way to get free perplexity pro for students
What are its offers
And what about referal programme
Why my free plan ended i think they promised everyone 1 year free pro plan
Hey everyone 👋
I’m currently using Perplexity AI and I really want to unlock its full potential for studio work (research, ideas, workflows, planning, etc.).
If anyone here has useful tips, best practices, or powerful prompts that you personally use, I’d really appreciate it 🙌
I lost pro account from a 1 year reedem code. I still get emails from tasks such I created. I don't see manage subscriptions where I could add credit card information. Is there a way to resolve it? I contacted perplexity Support. Now I'm waiting for answer from billing team.
I'm no longer using Atlas. Comet is my main driver
why did you make that decision?
Atlas does a very bad job of protecting your privacy on the web. Plus Comet is better at agentic tasks
eh how? do you tell it to post on social media for you? lol
This is what I mean
where comet
Have the same problem
Use Gemini I guess.
paste the website and add the title of the video, see if it can solve the problem.
yeah an ai browser that sees everything and uses everything for its functions will look very non private compared to non ai browsers lmao
if privacy is a high priority for you then you shouldnt use any ai browser
Gemini 3 Pro went crazy again
This appears to be the "Command Execution Loop" bug (sometimes jokingly called the "Agentic Seizure") that has been plaguing the Gemini 3 Pro model inside the Google Antigravity IDE since the January 2026 updates.
It is a specific type of hallucination where the model's "agentic" layer—the part responsible for deciding to act rather than speak—gets stuck in a self-reinforcing probability trap.
Why exactly does this happen?
Technically, this is a token probability breakdown between the model's internal reasoning and the IDE's tool execution layer.```
generation still not working?
Yo
Has the 12 months free student offer ended?
huh
Why does generation still not work even on Pro account.
I just complete the first step of AI project.
Now I am deploying now, thanks to all
Hello
Hi everyone, I had a question regarding the Airtel Perplexity Pro offer.
I claimed the 1-year Pro subscription through Airtel, but my access ended after around 4 months. I later found out that adding a payment method might have been required, which I wasn’t aware of at the time since it was an Airtel-sponsored plan.
Has anyone faced a similar issue? If I add a payment method now, does the remaining promotional Pro period get restored, or is it permanently lost?
Any insights or experiences would really help. Thanks in advance.
check if the subscription is 'paused' or not.
also check the messages after that one
How is my lander?
Has the offer for students to study for a free year ended?
Hello
Do you mean scholarships ? How do they work ?? Where can I get more information in this article
I mean a Preplexity offer that gave students a free year, and I think the offer has stopped.
If you didn't understand me, it's because of Google Translate.
Ooh nice
But do you have idea if they run scholarships programs?
They do not
No NSFW content, inappropriate usernames, or profile pics. These will result in removal or a ban.
Hello, is there an option to disable memory so that Perplexity does not remember what I wrote in previous chats? I am a pro user. Thank you for your reply.
bro
reply me
Why don't I have this option?
Oh thanks you
YouTube Premium $$$/month
Comet Broser for PC and Android Free or $5/month for Comet Plus (free with Perplexity Pro and MAx) full access to the Perplexity and also happens to block YouTube Ads
Wait YouTube premium with PPXL/Comet
I'm saying you don't need Premium
Unless you want background play on mobile like screen off youtube
Oh ok it blocks ad
@coarse dome any new projects you're working on?
For my most up to date works visit my website https://www.elaine.is last month I shipped:
- NeuroScan: An text classifier to mitigate AI safety risks
- FB Post Scaner: Wipe your Facebook post history and keep your account
End to End Encrypted Pastebin: The server cannot see the contents of your posts as the encryption key is stored aft er the # in the url meaning it's never sent to the server and they expire after 7 days. Very secure!
Password Generator - V2 of Polish My Writing. Less AI-like traits in the writing and when used with Grok and fine-tuned to a Twitter account for randomness it defeats GPTZero with a fully ChatGPT generated essay
Elaine Jackson-Pimentel - AI Consultant specializing in web development, DevSecOps, AI integration, custom chatbots, and AI training.
I prioritize on-device processing including end to end encryption AES-256 Bit GCM and only use server-side processing when required. APIs are built on top of Cloudflare Workers and their Data Platform meaning no insecure docker containers just secure JavaScript functions with minimal (usuallly no) external NPM packages
For data that needs server-side processing data is retained for as little time as possible. Usually 72 hours for security loigs which do contain the POST data sent. I always strive to retain as little data as possible to protect user privacy.
Well looks like GPTZero took my concerns seriously
It's flagged as AI now
Sometimes defeats Quilllbot and other detectors
IF you edit actuallly human input it's a lot better
One message removed from a suspended account.
One message removed from a suspended account.
One message removed from a suspended account.
One message removed from a suspended account.
yo
Eat a chicken
Are there any Perplexity Max users here?
Why?
Perplexity Pro no longer allows image generation due to regional restrictions? Previously it is allowed with cap, but now no longer.
Perplexity is aware and in the process of fixing the issue.
what is the daily limit for Comet browser help on pro? Says I reached the daily limit
Wdym?
I shared an interesting bug I found in Perplexity Comet with X. Has anyone here also had this issue?
https://discord.com/channels/1047197230748151888/1459771223276912722
Please post this in #1393681130842226698
The previous post was deleted.
Hii
Ty
So uhh, I have a problem
I was enjoying Perplexity pro that came bundled with my Airtel carrier subscription, but it ended earlier than it should've???
Why??
How to solve that issue?
Guys what is happening with perplexity?
I have pro subscription but they are still asking me to upgrade to pro wth is this?
Weird, are you on a free trial? They might need to put a credit card on file.
It was provided with an airtel plan
yeah they're doing a lot to try and cu toff the freebies early and well as setting stricter
Lots of anti patterns
Well if that's true that will be betrayal, will be shifting to claude pro subscription because that's the one I only use and perplexity used to automatically switch to best even when I haven't used the claude for 2-3 days .
Hi bro
wats up
How's everything tho
Do you also have any products ideas to sell this month season
no?
Ooh you didn't sell online right
Do I look like them
Or do you saw it on my profile
i just smell it
and i dont sell anything online lol
Okay could we connect to each other
no
But I already sent you a request
Lol 🤣
Hlo , smt happend with my pro ...... M an Airtel user so i got pro in free of cost ...... I think it's for life time ..... But now it's gone .... Someone tell me why
SAME
I have been using for months btw
It is for one year
Exactly
Ok so many people facing problem so it will solve itself 👍👍👍
Cool 😎😎😎
No worries
the comet browser is soo cool
Yes 👍👍👍
Oh 😮😮😮
I asked this and still no answer? what is the daily limit for Comet browser help on pro? Says I reached the daily limit
same problem
sup
It was not working. on mines
hey
@green crystal https://x.com/i/status/2009963629500870977
At this point I don't even understand what people want.
That's freaking awesome!
I've been wanting to use other models from within CC.
Any one here
Me
Hello
hi
What do you mean, this has been rolled out weeks ago
Claude code has their own instructions and tool calls for the ai which makes claude ``code More advanced than other cli
I have used gemini flash and deepseek inside claude close although deepseek and some other models didn't work properly they kept breaking down and outouting tool callds to me
And?
It's not new
So?
And don't be too hyped
Hyped?
Its not optimized for a lot of models
Even for gemini flash i had to rerun some queries
Because other models keep messing things up for some reason
Poor context management by the delegator
No it wasn't context
You know when ai outputs tool calls instead of responses?
And poor system prompts too
Speaking about context, cc runs like 18k context minimum start up and it runs multiple time
That's how large their system instruction is
So maybe the models went halulu
How do you know it's uses that much
Via open router
I wish you would use full sentences
I don't what that is
Open router is an app that connects many ai models
I used the api key from open router
And open router has a dashboard that let's me see what happens with my api such as which model was used and how much context
I also noticed that every Time i made a query to claude code it would call the api twice so i would get charged double :c
This is probably done with the good intention to make sure the ai tool calls hut this methods costs a lot
GM chad
Hi
Lmfao. Had me in the first half.
Wdym 
how do i put a icon or symbol on spaces like these
Open the space its next to the name of the space
ty
I read "good intention of making this cost a lot" .
hello
??
Add ur credit card and it should be all good
i have not any credit card.
welp that's sad
coz I am student
because it's compulsory to add ur credit card to keeo using the pro plan
..
just a new way to prevent fraud accounts
Simply refuse it. It's unnecessary unless you have a location based query.
what happened here lol
I was asking/researching about how limited options in prison force people to go homo and how lifelong rejection might be a prison in itself leading to people being homo/bi when it said prison behaviour doesn't define identity and people engaging in homo activities still identified as hetero. Also how bisexuals have a preference for hetero over homo. Where transmaxxing came from is beyond my brains.
hmm
i hope the rumored shift to more pre training for gpt models doesnt make them nuts
yeah thats sketchy lol
i gave the dr report to 5.2 pro
curious what it says
the only thing holding back 5.2 pro from outright beating dr is its limited web calls 
Is it limited on api too? Or unrestricted there?
most likely not
you're the one paying for the web calls lmao
the limit is context and your wallet
So even 5.2 thinking with potentially unlimited web calls would be crazy.
yeah
but context would probably hit a snag somewhere if its a really indepth research
True
transmaxing is better than looksmaxing?
have to choose way
The thai secret.
good txt about case
hé
@distant aurora its still going for 21 min lol
thats on standard thinking
extended would be like double
Give the prompt at the start of my conversation to gpt.
Is it possible to use cloudflare development as an MCP for perlexity... unfortunately cant see it bc im on my phone rn and travelling..
would appreciate a response
Yes you can
Yk the white screen problem after deploying.. any idea whats likely the problem, i did even used files where it doesnt make sense that they dont function.. i got one worker as a password checker for my main website, now i wanted to deploy another one though no matter what i deploy now it goes white screen
You can send me what it shows
Thx
Do we have any hints as to when Sync will be added to Comet?
@admin ban obviously the oldest scam in crypto
Please dont fall for something like this
Perplexity did us Airtel pro users dirty 🥀
Pulled the rug beneath us 6 months prior to when they should've 🥀 🥀
Was gonna convert to a paying user (only during the months of exams) after my free trial was over
But now the trust has been broken, gonna claim Gemini pro trial using a Jio simcard and never look back
What
I need to edit one photo
As for me, I'm leaving this place once I get some more answers
Oo
I also thought the same but apparently perplexity changed their transaction method. You would have got an email. If you don't have a default payment method, try adding your card details then you will get the airtel trial benefits back
I see
Well, good thing I waited for some answers
This server for things? What is perplexity?
Lol
@vague pagoda do you write all your code yourself or do you vibecode sometimes
He ponders all his code 
Did you see ChatGPT health
No... what happened ?
There is a waitlist
ayy nice
Hell naw, i'll pass
I applied when it first released and still nothing...
Ig they will drop the beta soon
Not sure
first time hearing about Chatgpt health
It's not in beta
I've seen other users get access
just slow release
Send that n*de to willy
Guys what the hell
Des français j’ai besoin d’aide ?
pour quel sujet?
Avec la version pro on a pas accès à Veo3?
je crois que oui
Que la version max du coup?
How do i upload zip files? I have pro version.
oui je pense
sauf si il vont rajouter pour la version pro
tu t’y connais niveau ai? car j’aimerai faire la trend tiktok viral sur les objets du quotidien qui parle
tu peux utiliser sora
je ne sais pas s'il est si bon que ça
lmao?
why does every time i type something in the app it says something went wrong
Please use English, Pro users get access to video generation with Veo3.1
Hi i need an AI gf
@formal tangle
@copper tapir
I don’t speak english excuse me, but The app is telling me that I need to switch to the Max version to be able to do it.
have you generated any videos this month?
no
it anybody’s app not working say something went wrong when you search something ??
No, please make a #1140622008086970420
Weird, please contact support@perplexity.ai and make a #1140622008086970420
Unfortunately, the video generation function is not available on my platform. However, I have created a high-quality reference image of your anthropomorphic washing machine character in Pixar style that could serve as a visual foundation for your project.
@dry kestrel
oui c'est une ia a part, sinon tu as nonobanana aussi
moi ça me dis que ce n'est pas disponible dans ma région alors que depuis que j'ai perplexity ça fonctionnais
Sa me saoul je veux faire cette trend mdr
English
Hey! Just a heads up, this is an English-only channel so everyone feels included.
Appreciate you switching back to English @formal tangle
@halcyon coral
hi i was trying something on my laptop and accidentally changed my laptop ID. Now I've lost my Pro subscription; can you fix it?
Hello everyone
@vague pagoda
any mode's can i get it back ??
You need to add billing details to continue, as is stated on the page
This is an anti-fraud measure and you won't be charged for doing this
Click Update billing Verify your login
I've been coding the normal way for years but I'm trying to get comfortable with AI tools for coding, some are pretty good
self-promo but https://github.com/perplexityai/modelcontextprotocol has been very helpful for getting snippets of info into the agents context while working with libraries I'm not familiar with
Wasn't expecting you to ACTUALLY reply so thanks for that.
Back on topic, I used to vibecode, but then I just found it wasn't what I needed. Something about it was off, maybe it was the colours, the gradients or something I just don't understand. Good luck asking it for an exploit for iOS (I do cybersecurity and jailbreaking) and it just cannot do it. I learnt Swift some time ago but forgot about it (more on that later) and started to do HTML seriously a few months ago. Since then I have learnt Swift/SwiftUI again and made my first project, NanoTap that allows you to scan a QR Code, NFC Tag or NanoCode (a ten digit code). I am learning React and made a site yesterday with it. I now make 10x better sites and actually know what I am doing (hopefully 🙃).
I am all for AI browsing and letting AI teach you programming. I do it all the time, but vibecoding isn't for me.
looks interesting, i will look into this
What do you mean by laptop ID? Did you forget your username and Password ?
I think it's more of a clean up inactive SheerID/AirTel deals lol
Any news on Comet for IOS
Hi I'm Blind Goose 🪿 your digital girlfriend
The fastest way is to get rid of your Apple devices and replace them with Android ones. Apple forces third party web browser to use Apple WebKit meaning the Firefox and Chrome web engines aren't available. This has the added effect of limiting the advances of third party web browsers. In fact you get a separate JavaScript engine which is much slower.
what is the highest level of thinking for Grok platform?
Expert or Grok 4.1 Thinking?
I believe 4.1 Thinking when you ask it to think harderr
Technically
heavy
Probably the most for free and low cost users
Between Expert and 4.1 Thinking you're better off with 4.1 Thinking
Hey i just had a question, what would be the best model for Advanced math/accounting
Expert is Grok 4
ah right, sorta confusing eh
Pretty much all the models that support reasoning are great at math, I'd personally use Sonnet 4.5 or GPT-5.2
cos "Fast" mentions Quick responses by 4.1
Thanks dude!
4.1 has a non reasoning variant
so Grok 4.1 Thinking would be the best of that list in screenshot?
in terms of combination of running latest Grok and having some degree of thinking
Grok 4 has the highest benchmarks
so when would you use Grok 4.1
Hi can i have a digital egg?
I can't create images. First, it told me, “Unfortunately, it is not possible to generate images due to restrictions in your region.” That's impossible; I've always been able to generate images. Now it tells me that I have to subscribe because it says I need to be a pro, and it even gives me the steps to do so, even though I've had a PRO subscription for a long time. Are they making fun of me? I need a solution right now! Can anyone help me or has any ideas?
Thank you for clrifying
What region are you in
Well Grok heavy is but ou’re excludng it so…
yea but that's like saying PPLX Max is the best, but ur excluding it so...
not that helpful imo
You gotta learn to think critically, try multiple models and know when to use which odel. There is no best model or best sub. Different problem require different tools
Your dub.co account has been banned from the Perplexity Partner Program for fraudulent activity.
If you believe this was a mistake, you may appeal by emailing support@perplexity.ai. Please include all relevant details in your message.
Important:
- Do not open bug reports or use any Discord channel to request a ban review or appeal related to this issue.
- Do not contact Perplexity staff via direct message or @mention them, including <@&1098365634498285709>.
- Misuse of Discord channels or attempts to bypass the official support process may lead to further consequences.
All appeals will only be reviewed via the official support email.
Please be patient: Support may take 3-5 days or longer to respond to your appeal. Avoid sending multiple follow-up emails, as this may delay the review process.
Thank you for your understanding.
hey I am 40 years-old man from Germany who is trying to integrate AI into his life
I gave the context already though
basically my question was which one used the most thinking by default
but it seems it's quite a complex answer, since they are not even using the same model
one uses 4.1, another uses 4 etc
imo it's quite confusing - it'd be akin to something like ChatGPT giving the model selector options as "GPT-5.1", "GPT-5 Thinking", and "GPT-5.2 Fast" or something
looks like PPLX is trying a new COT thinking ?
uses the term "Researching" now
rather than "Thinking"
but only for certain types of prompts interesting
Well, contrary to popular belief, the prompt affects how long the thinking happens. So we really can't tell you about looking at a specific prompt how long it's going to thing for her. In general. Heavy is going to think longer than 4.1 thinking or expert mode but it depends
okay let me ask another question
which one thinks more in ChatGPT platform
GPT-5.2 Fast or GPT-5.2 Thinking?
In general thinking but it depends
what would be an example of a situation where GPT-5.2 Fast thinks more
Hello Guys
GPT-5.2 Fast is not a valid model.
Unless you are referring to GPT-5.2 Minimal Fast
Anyone else got Sorry, something went wrong error message right now?
No, please make a #1140622008086970420
yes sure, that one
Minimal is the reasoning setting. Fast is just the output speed.
actually wait, what about GPT-5.2 Instant?
which one thinks more - GPT-5.2 Instant or GPT-5.2 Thinking
It's in the name
doesn't it "depend"?
They have a hard limit
when “espresso”
hard limit for what
When GPT-5.2 Codex Max x-high Fast?
still trying to clarify an example where GPT-5.2 Fast thinks more than GPT-5.2 Thinking
GPT-5.2 Fast and GPT-5.2 Thinking are not valid models to compare.
You can review the reasoning effort in the API documentation
Shouldn't be comparing fast to regular models in the first place since the only difference is output speed
in the ChatGPT platform, what are the corresponding thinking levels now?
and descriptors with x-high, high, medium etc correspond?
there used to be "juice" level descriptors - are they still valid?
Heavy lies the crown
there's always been a lack of clarity imo
most of the info iirc was from reddit posters "discovering" things about juice level etc lol
Hello everyone I’m a developer who’s working with AI and automation lately.
I mostly work with web stuff (Next.js, TypeScript) and I like building practical things like bots, workflows, and small AI tools that actually solve problems.
I joined to learn from others here, exchange ideas, and maybe collaborate if it makes sense.
Nice to meet you all 🙂
x-high is not available on web
all the others correspond
okay that's what I assumed
and PPLX still uses Standard right?
(previously known as Medium iirc)
They introduce new limits for pro?
a lot of people have noticed this, but it's hard to figure out what exactly the limits are
Hello all, I was able to generate images with perplexity until fairly recently, but now I get hit by this?
thanks
Yea, willy loves medic related ai stuff
I think so, not sure what
Well it's a bit more complex than that. The LLM paused for some time to produce a chain of thought that's used to prepare it's response. It is a different approach to problem solving in. The models are designed for different types of tasks. You probably do not need a model with reasoning for quick text generation, whereas a model for coding might include reasoning because it's necessary for the task, especially if the developer cannot write you detailed enough prompt about what they're trying to do. The extra thinking time helps give a better response, even if it's not exactly what they want
I had never expected guavas to be so difficult to cut with a knife.
This has nothing to do with that conversation and my reply.
It does because it adds the why the models aren't really comparable.
No lol, we were talking about OpenAI models
The "Fast" just means you pay priority pricing for higher token output speed
That's not true when you're comparing reasoning vs non reasoning models n they are separate. The blog post about ChatGPT Enterprise looks interesting but normal non-Enterprise users don't have access to priority pricing.
- They asked the difference between GPT-5.2 Fast and GPT-5.2 Thinking
- I explained that these are not valid model parameters.
I have no idea what you are talking about
No idea what this is, please read the entire conversation and the link I sent.
I believe they were talking about Grok not ChatGPT
Please read the entire conversation
Also should look at the message that I was replying to
It's in reference to this user. They're mixing up terms and derailed from asking about Grok to GPT they were also DMing me questions
They are talking about the buttons in the Mobile or web apps not talking about API access
Mares is a certified Sabrina carpenter fan 🤨
yes was not talking about API access
web app only
They are certainly not asking about priority API pricing 😕
the reason for asking about that from the initial Grok conversation was to illustrate the confusion
that many platforms have imo
Gemini might be the cleanest i guess lol
altho there's another layer of variable with that since new Fast came
Gemini has Fast Thinking and Pro. They are by far the most confusing to the end user
the GUI for Grok is more confusing imo
but it might be because I've used Gemini platform more in the past
for Gemini, it'd be clearer if they labelled it "Fast Thinking"
but they just label it "Thinking"
or even just be consistent with previous naming convention - Flash, Flash Thinking, Pro
😂
- They asked the difference between GPT-5.2 Fast and GPT-5.2 Thinking
- I explained that these are not valid model parameters. And the fast variant only exists over the priority API
Fast has commonly been used if a colloquial name for non-thinking AI models
Is anyone having issue adding a card? I emailed them but no response, really great customer service
No?
So if the average user in the interface of Grok sees Fast and Thinking and then they try another AI app. They're going to ask about the same modes for that other AI app, even if they don't use the exact same technical labels.
Average users don't use Grok
Has anyone seen this before?
They do. Join X people tweet at it every day and use the app to have fun.
Lol
Thankfully, all of the AI apps copy each other's user interface so it's not hard for users to jump ship from one project to another
Grok had 271 million visits in Dec 2025 while ChatGPT had a staggering 5.5 BILLION
You are severely overestimating the usage of Grok
Regarding the use of "fast" - ChatGPT, Z.Ai, Qwen, Claude, MiniMax, Kimi, Ernie, Cohere, all do not have this.
I did not claim they had high utilization. I just said that there are normal people who are not technical who use grok
Grok isn't that popular, from what i see is people using grok for it's new uncen feature
And it may be a very small minority of Internet users, but it's important to point out. They're also a much more platform much newer and have a much more Keen than the other big AI Giants and are doing way more
The average user is far more likely to use ChatGPT than Grok
The average user does not even know what Grok is
Grok Imagine getting more popular 😛
They are, but that is slowly changing mostly because of Google, not because of smaller platforms like perplexity or Gemini
They have disabled generations for free users.
just get Max?
Perplexity is not a "small" platform, they had 179.6 million visits last December.
I'm one of the 0,0000001 of users who use 5 or more AI apps a day
Compared to ChatHPT it is very dmall
I'm curious about what you're trying to prove here
as i'm lost atm
Yet you say the average user uses Grok
I'm more curious about what both are trying to prove imo
Relatively speaking, most AI platforms are quite small compared to chatGPT so yes Grok is small but so is Perplexity and Copilot and to a lesser extent Google
However, most of the data we are seeing is what visit data. It doesn't tell us what percent of those users are subscribers versus free users
What percent of ChayGPT huge user base is just teens without a subscription
you didn't say compared to ChatGPT
Fast is not a commonly used name for non-thinking AI models
pretty sure he threw in terms like "colloquial", and also there was a lot of context behind it
Google has the most users by far out of these companies with over 2 billion monthly users accessing AI Overviews.
guys, I didn't understand the real difference or which one is better
The reasoning models often perform better at complex queries, Sonar and Gemini 3 Flash if you want a quick answer
also please don't spam channels with the same query
You used data about ChatGPT as basis to call Grok small. That information is now in the context of this conversation. We are comparing the sizes of AI chats and chat gpt is our measure of what a large one looks like. I don't need to repeat myself every single time in the conversation because everyone already knows that we're using this as the basis now
If you put that into context, they added AI overviews to Gmail and to search. So users are being displayed AI content but it's not because they tried to use Google AI. It's because it got added to the applications they're already using without them having to ask or click anything
not just Sonar and Gemini 3 Flash for quick answer, but GPT-5.2 (non thinking), Sonnet 4.5 (non thinking), Grok 4.1 (non thinking) are also fast
You said the average user uses Grok. I used data to show you that this is not the case.
It would be no different than if you redirected Google to chatGPT and then tried to say oh wow chatGPT has more traffic. Now they're popular. No traffic got redirected
I said average users grok I never made such a plane that it is their go-to tool chatGPT remains more popular
Their AI mode already has over 175 million daily active users as of Q3.
Ai mode is seperate from Ai overviews
I don't think we're arguing that Google has a lot of traffic too, but it's traffic they've had for years. That's being redirected into other products such as prompting you when you are typing a query into Google Chrome to try with AI mode is not that users had any intent on using AI until they got prompted and that traffic was redirected. The Gemini overviews in Google search are another thing. It's very different than a user making the conscious decision to open ChatGPT or perplexity and then search.
Please read what I just said.
Pair this with the 650 million monthly active users on Gemini APP should show that they are not a "Smaller platform"
Bear in mind this is monthly active users, not monthly visits.
An app that is pre-installed on billions of Android devices and gets used every time you try to set a timer
What are you trying to prove here?
Google was grouped in with Grok and Perplexity as smaller AI platforms. Compared to ChatGPT. The current data being released by Google about their usage of their AI platform is not credible because oh, you used an AI overview that's more Google AI traffic for us. Let's show it to our investors. When your Android phone is silently updated from Google Assistant to Google Gemini and you try to set a timer and it does so. But now that count is using the Google Gemini mobile app, that's not in good faith and it's very misleading when you're trying to look at how much interest there is in Google AI versus people whose traffic just got redirected into using it
of all the people i've spoken they all know chatgpt
1 person i've ever heard gemini
but its not their main llm lol
heck the local news papers use gpt image v1.5 for stuff where nano banana pro currently is better lol
thats because ai overviews are automatic
I've addressed this here #💬│general message
Original point that I was trying to prove is that the average user does not use Grok
its because ai mode is put right there where people see the "search" button
yeah true
even for people like us
grok just isnt reliable
sorta depends what is meant by "average"

and also, is the claim, zero "average users" use Grok
since the day i cancelled the grok heavy plan
rando person you bump into
if i ask a rando now what ai they use you'll hear chatgpt lol
It's only useful for finding X posts
it'd be hard to prove that zero "average users" don't use Grok
you paid $300 for that? 
Even with that, Gemini def is not "small"
3000$ to be exact
it was a 1Y plan
yeah its not small thats for sure
That was my point.
that's curel
its also not as insanely large as people on X and reddit claim it to be
hopefully this changes when Grok 5 releases
according to those 2 platforms even a baby knows what gemini is
My point was that "Gemini is not small"
It's quite useful if you want it to write in your writing style it can use your posts on X as inspiration
BRUH
well how was grok heavy
was it atleast useful?

I hope so, grok 4 was a nightmare
Do we have a source of data that excluded traffic redirects? E.g. No hey Google, no search or Gmail overviews, etc?
Perplexity works for this although I try to avoid using AI when posting.
Yep
Really? Most of messages longer than a single sentence at the very least are edited with MS Copilot
Also they should really stop redacting the reasoning over the API
As a developer that information would be very useful, but in production I imagine it would just constrain networks with more data that's never going to get looked at
which model?
Grok 4
ahh
and 4.1
grok 4 wasn't worth it
although i would love to use it and find random posts that kesku liked
and start tagging him
@dry kestrel we should do this fr
I get increased access to Grok with my X Premium Membership but do not pay for Super Grok. Perplexity handles 99.99% of my needs. Gemini and Copilot are a really good glorified autocorrect
same but it gets sent to chatgpt
copilot is if i wonder if my grammar and spelling is correct


Copilot is basically ChatGPT Plus for free
hmm
you mean go
not plus
reasoning effort is set to low
with copilot
I mean plus. You get unlimited file uploads and a access to GPT-5.1 which is generous
yeah but low reasoning effort
something that go tier has
plus gives you high reasoning
there's a large difference in ability
i think that's more like Go plan
but it doesn't make it any better
With copilot you can reason for up to 30 seconds just like plus
What's the reasoning level for copilot on free plan
i think your prompts force it
is it instant or low reasoning
Limited Deep Research, less image generation, no Sora, no reasoning control, no access to older models, no Codex, way less connectors
both
This is just wrong
I can very much likely do the same with chatgpt free plan
so basically just instant model
nah you get auto and thinking mode
the smart mode is auto
thinking is "think deeper" mode
The UI is quite confusing
indeed
This is without logging in.
oh i got logged out
Hello everyone
I'm currently a full time student but when I return to full time work early next year I'll be advocating for Copilot in the workspace and buying Copilot Pro as my 4th AI sub lol
I can't see this
Mine is on a enterprise account
millions?
Hello
Enterprise and school accounts have different version
oh ye mine is school one
ohh
think its due to the benchmarking news, betting markets aswell as wallstreet
there's allot of gemini talk there
Hello @vague pagoda I have a question
Yeah, just check if you are on https://copilot.microsoft.com/ or msft copilot
The latter is the enterprise version
Please just ask the question
Okay
I think it's the school version
What do you guy do here trading or what
interesting
uhhh no ?
This is the Perplexity AI server, we do not do trading here.
Oh really
I'm guessing it looks like this? https://www.youtube.com/watch?v=_BSGrykqFAM
GPT-5 is now available in Microsoft 365 Copilot. Users can tap into advanced reasoning and fluent responses tailored to their work. Copilot dynamically selects between GPT-5 Chat and GPT-5 Reasoning models to deliver fast answers or deep analysis — helping you make smarter decisions, faster.
#Microsoft #Microsoft365
it just redirect me to the office 365 homepage with copilot in it
i can't watch the video rn
i think imma leave that when i'm home
guys, how to make perplexity forget all of my personal information?
delete all the chats
bruh
😭😭😭
In Comet Browser, go to Settings → Privacy and Security → Comet Assistant.
Disable personal data usage for the assistant. This prevents Comet from using your browsing history and personal info for future queries.
i used perplexity only, do i have to download comet
hell nha!!
just Go to your account settings (usually found by clicking your profile icon in the top right corner)
Look for the option to delete your account or request data deletion, typically found in the privacy or account management section
Follow the prompts to confirm deletion. All your personal data, including search history and saved queries, will be permanently deleted after 30 days
You can delete your account and data under https://www.perplexity.ai/account/details
but i dont want to, i just want to delete my personal info, not the whole account or chat history
What exactly do you mean by "personal info"
but if i delete my account, i have to create a new one right?
If you want to use the service again. What exactly do you mean by "personal info"
I made quite a bit of progress on my AI Writing Assistant. It now works perfectly on Sonar, it previously needed a model with reasoning to work fcorrectly.
Hello!
Hello I'm Goose - Your Perplexity AI Expert - Ask Me Anything!
What is the meaning of life
To serve the lord almighty, Goose!
Goose creatred the universe in seven honks
And ?
so many mutual servers
Image and Video generation is no longer available in India?
You know what I didn't think about. Is there a Microsoft Copiolot Discord?
Yes there is a server for it
idk why... but most of the tech discord servers is boring 😄
I joined the Microsoft Copilot one caseu why not
image generation have some problem
why is image generation blocked? i'm based in india
okay, thanks
i should be asleep, I have a 3 hour lecture soon 😢
Yes, please, or else you'll sleep during the lecture 😂
I’m just too excited 😆
Hi
hello
Does anyone else having problems with image gen?
I am thank you I just messaged him 🙏
Kya koi meri help kar sakta hai bolo
the two best image gen models right now might be gemini and grok
fixed thank you @vague pagoda @wraith crow
Nano Banana Pro
GPT Image 1.5
Seedream 4.5
Flux 2 Series

chinese z-image its good too
10% of the price

am i the only who is facing that pro subscription issue..???
could you describe your issue with your pro subscription?
I had the 1 yr pro plan of perplexity ..
which was offered by Airtel with their recharge plan (for free)...
but now my pro subscription is gone...
when did you apply for it?
June - July..
there are still 3 months left for it to end naturally...
