#š¬āgeneral
1 messages Ā· Page 546 of 1
u have the pro license but u want it?
u ok?
Check the pins to get Pro role
Because I dont have any button that states "Go to Discord" anywhere.
not for the role
š Pro
If youāre a Pro subscriber, click the Go To Discord button in your settings to get the <@&1113091346585100318> role.
If it doesnāt work, tag @vague pagoda for help.
2-5 business days
bruh isn't there a quicker way?
Yes
You could try in the intercom as well https://www.perplexity.ai/account/details
Ah that why!
intercom?
don't make multiple tickets, you can just ask for human support
Looks like this
It's 200/week, rolling window
https://www.perplexity.ai/rest/user/settings
where can i find my usage? if yk
remaining_labs":
remaining_pro":
remaining_research":
{
"model_specific_limits": {
},
"remaining_agentic_research": 0,
"remaining_labs": 0,
"remaining_pro": 0,
"remaining_research": 0,
lmao perplexity is so shit
i never used research
cannot be possible for it to be on 0
I'm glad that rate-limit page now shows remaining agentic uses, it used to be always 0 (for me) but lately has been showing 2 which seems to be the daily limit based on my unscientific experiments.
Did anyone try comet on iOS?
yeah, you can pre-register for it
Will see how effective itās
I started using brave , they have the ai thing on beta
Will better f privacy lol
Lost all my respect to Brave
Why
I have Comet on Android but haven't used the agentic thing at all, it's probably time to uninstall Perplexity app since I can access the web version of it in Comet anyway
targeting Perplexity for their Comet thing
made false claims about Comet
How is Pplx Pro Deep Research compared with Opus 4.6 Ext Thinking Research? How does works Pplx Pro DR?
Who cares, as long as my shit is private , idc what kind of shit talk they do to others lol
Depends on the query .
This is insane
n 2020, Brave got caught injecting URLs with affiliate codes when users tried browsing to various websites.
Also in 2020, they silently started injecting ads into their home page backgrounds, pocketing the revenue. There was a lot of pushback: "the sponsored backgrounds give a bad first impression."
See you canāt trust anyone blindly. Thats why I use dns and AdGuard and a better router .
How do you prevent a browser from injecing affiliate codes with AdGuard?
Thatās how you earn brave token. Idk why a fuss about it. Itās just an ads on the screen, much better than ms edge bulshit news
I don't like MsEdge either
It wouldnāt load , affiliate links get blocked, unless itās something like =rugin3.
I think that also gets blocked
Everytime I clicked an affiliate link my dns blocks it, I had to turn it off to access it
I donāt use AdGuard DNS
well the browser shouldn't be injecting links in the first place
Wtf Pplx Zen??
https://www.perplexity.ai/b/zen
That I agree. Thatās shady as fuck.
Comet thing
I used to like Brave as a lightweight browser, stopped using it since that came to light
This is zen
I still use brave, using water fox and comet. But trying to get rid of comet
At least Comet hasn't done anything messed up like that
I hope it's AI generated
Takes our f data , atleast brave doesnāt
It doesn't
not sure where you read about that
https://www.perplexity.ai/hub/legal/comet-privacy-notice, https://www.perplexity.ai/comet/resources/articles/comet-data-privacy-security-faq-s
These two
someone at pplx got bored
what are some cool things i can do with perplexity?
Anyone know how to fix not getting 'enhanced' queriers as pro user?
What the heck happened to this ai. It used to remember everything but now it canāt remember a thing from a previous conversation. Itās insane.
its fully shitting itself
Where on earth did you see a line that says they take your data
saying i used research queries when ive never used them
Iām a pro user too
I have so much stuff do this week, literally the worst week ever
honestly going to cancel my plan
Itās messing up everything
make sure you have these toggled on
If it still doesn't work please make a #1140622008086970420
Yeah I do
Do you know if there is a way to bypass it? or is there a better version of perp
no idea how to bypass the limits.. probably not the best place to ask here also
copilot pro will do for the morning
at least they have a usage meter
that actually works properly...
facts. feels like a scam changing the amount after paying
Does anyone else relate to it being crap this week
they lowered the limits
Where can you see this?
but there is no way i exceeded it i maybe had 30 claude 4.6 prompts
github copilot pro
Ah the coding plan...
if i give my student info do i unlock extra features i already have pro?
It's cheaper if you are a student
Bruh Iām so pissed it wonāt remember any work stuff
already have pro for 1 yeari remeber before they had like the teach feature exlusive to students
Whatās happening with Sonnet 4.6 Thinking? The 4.5 version was working great, but with this new update, itās lagging and takes forever to respond.
It always gives a timeout error when I try to perform more complex tasks using uploaded files, making it unusable. Am I missing something?
but i guess thats open to everyone now
make a space and do a handover
give it to a different agent, if your chat is really long it clunks up xD
It might be due to it being a new model
use the same model
How do I do that?
just tell ur chat "this thread is getting too full i need to do a handover provide all information for the handover, i will use the same model"
something like that
they really need to add this to the accounts on perplexity after this experience
taken from github
But I also had issue with it cutting off when making a chart
yes likely the thread is too long
It's just not working. I've been using the same workflow for weeks, and suddenly everything has stopped, giving this error constantly. It's so frustrating.
You could try a prompt like this
at least you get an error response
mine just switches to best lol
check ur rate limit
i checked mine says all used out
never used labs
or research
why are they remaining on 0?
lol š©
hey, is there a way to see which model answered my prompt? i have it set to best, so i'm not sure which model it picked.
ur a legend
What should I look at?
remaining_agentic_research
remaining_labs
remaining_pro
remaining_research
if i have no research.. then how do i have 0 remaining??
if i dont get support in 24hrs i am opening a chargeback
and i will win
i have all the evidence
I have plenty, except for "remaining_agentic_research," which is set to 0. Is that the reason for the error message?
u dont have the issue
theres an issue where it says we maxed our usage
What does everyone think of kimi 2.5
but its not even possible if i never used other features
how can i max something out if i never even used it ever
and ive had perplexity since april
it really fell downhill
and tbh
i paid for a year
so if i get a refund itll be a years worth
idm free ai
Is Perplexity down? It keeps stopping "The operation couldn't be completed."
They run a promotion for several months, offering a free PRO subscription for a year. Everything goes smoothly, and everyone is happy.
However, once they reach a specific number of users (the exact number is unclear), they impose limits designed to push you out of the comfort zone you've enjoyed for a long time, trying to force you to upgrade.
Making use of my free PRO sub as much as I can, even with the nerfs
No š
First
Companies that take large language models like ChatGPT or Gemini, re-skin them, and sell them as if it were their own technology.
These companies lose their reason to exist overnight the moment Google or OpenAI adds the same feature into their own AI products.
That is why the Google VP emphasized the need for a deep moat.
Either go extremely deep into a specific industry, like Harvey AI in legal, or deliver overwhelming convenience like Cursor does for coding, etc.
Second
Companies that act as a middle bridge by bundling multiple AI models like ChatGPT, Claude, and Gemini and letting users pick and choose (examples: Perplexity, OpenRouter, etc.).
He warned that the same history will repeat: in the early AWS days, there were middlemen who simply resold cloud infrastructure and took a cut, but they all went out of business once Amazon strengthened its own services.
If the major AI companies start providing routing natively, these middlemen will lose their place.
Instead of these companies, the VP spoke highly of vibe-coding platforms and developer tools.
He also predicted strong growth for:
D2C services that put AI directly into consumers hands to enable real creation, such as video production tools, and
startups in biotech and climate tech that analyze massive amounts of data to generate real value.
https://techcrunch.com/2026/02/21/google-vp-warns-that-two-types-of-ai-startups-may-not-survive/
I heard this viewpoint from a Google VP, and I strongly agree. Perplexity needs to think deeply about its real competitive advantage. They should not try to dodge this by staying quiet or evasive.
"Enshittification"
is it possible too use multiple AIs on one project?
like use Sonnet too make idk a site/gui and have Kimi write tho code or something
I use research all the time and I've literally just been on the year trial
im on 0 for research and never used research... it shouldnt be on 0 for me
1 - 10 
You should add the epic "You reached your limit on advanced AI models this week" in the message š„

@everyone
am i cooked
why isnt this like that one video
where everyone starts typing stuff
well
this is akward
is this the part where i get banned
or do i go out with a fight
Claude Opus 4.6 "Thinking" Deep Research is the best now
it makes sense for Perplexity to use Opus (unfortunately non-thinking) for its Deep research
did no one see my message
I like Perplexity Discover.
the sin of sloth
orbs dedo yrrrrr
u should not bet gng
bro its so good, how did you make this ?
can i make this with remotion ?
You wanna be engineer want to tell me why i need to scroll down every time I give in a prompt? Or why I am top-scrolled?
I haven't checked in for a few days. Have Perplexity gone bust yet?
Not yet
honest opinion? 9/10
for a 10, make it a tiny bit longer š
Are u f***ing kiddin us again?!
Another limits? Research MONTHLY limits?!
Give me adress to contact with team thats care about payments. I want my money back.
Cheaters!
@dry kestrel
Weird...
Somehow
i can access this weird model version from Chatgpt
Any info about it ?
š
prob
<@&1098365634498285709>
Feel free to share projects you're working on in #šācool-projects
@dry kestrel
thank god for cybersecurity and ai
itās crazy how ai is being applied in cybersecurity now
and itās wild how AI powers malware that adapts internally in someoneās computer now
back to studying I go peace
It's annoying as fuck
Wsp
ayo what
/imagineĀ Spend $10 travel meme, 3x3 grid, left prices $1 bus ticket, $2 train, $3 bike, $4 economy flight, $5 hotel room, $6 cruise ship, $7 private jet, $8 yacht, $9 luxury villa, $10 world tour jet with views, travel scenes cute like sweets, vibrant, text prices and titles --ar 1:1.5 --stylize 600 --v 7
@dry kestrel pro reasoning visibility got nuked

oh wait
it does appear but after a long delay...
ffs they shouldnt code the entirety of chatgpt with codex
interesting
the reasoning box thing is back
kinda
for coding it doesnt show reasoning much
but just by itself it shows
updating its task is just through the input field
no separate button anymore
also it got better and allot faster at making reports
like nearly 2x faster
Perplexity has been showing an error since yesterday, "The operation couldn't be completed." Does anyone have a similar problem? How do you fix it?
I only have this when i ask a hard question or use opus
Perplexity is a $20B company, i promise that they don't care about $20.
I use Claude Sonnet 4.6 Thinking.
probably 22B by now
Sup
does using deep research in perplexity pro still have limits? because there are no more indications on how much remaining deep research left
You can see your account limits/remaining at https://www.perplexity.ai/rest/rate-limit/all
Haven't used that feature, instead inside the Notifications area I have a scheduled daily task that runs with this prompt (which could be modified to expand on every story if you desire): What are today's most important and widely-discussed news stories in the past 12 hours? Focus on: 1) Breaking news and major developments 2) Viral stories getting significant attention 3) Notable events affecting many people 4) Surprising or unusual stories gaining traction. Your output should be long enough to be a 15-minute read at normal human reading speed that is not sparse or light on details.
Why does perp by default use sonar when I open a new tab? I remember it used to remember which model you pick across new tabs, its bad because quickly opening perp and pasting something without thinking about the models yeild bad results which then I have to prompt again, š
It is probably updating the activity via a third-party app, since there is not a dedicated Google account connector for Discord in that sense. I am unaware of an official Google Discord app... so it might be something custom... but if anyone finds it please let us know.
If it's custom, whatever it says is in turn suspect, especially if he wrote it himself.
threw a screenshot of that into Perplexity and it said:
That's a rich presence (RPC). Probably using https://premid.app/ or some other app
(And the point still stands, the app can update the status with anything if it's custom, since I don't believe it is an official Google app/integration)
(So whatever it says is suspect)
I asked it more and it said go to https://top.gg/search?q=google and invite a bot to your own Discord server. So, yeah... something custom someone had made.
Perplexity is just making stuff up
perhaps, though that website does exist
Doesn't have anything to do with the RPC the user has though
Thursday probably. Hopefully.
200$ for the year mate, and if everyone starts opening chargebacks it will no longer be a 20billioj company, you can keep being a little corporate puppet you are too good for the system
You dont even have pro lol, talking about not caring a out 20$
Its a shit service
Again another standard user being a bum
@green crystal almost commented "misinformation" until I realised he's right. https://cunnyx.com/i/status/2026321446398808359
ChatGPT vs Grok
ļøļø
ļøļøChatGPT canāt watch or understand videos.
ļøļø
ļøļøGrok can watch videos and summarize them for you.
Open ai needs to add video support.
None of the grok text models support video inputs though over the API
It's annoying that GPT-5.2 is able to accept video inputs but OAI just hasn't released support for it
Dies it actually understand video, or is it listening to audio for tts?
Based on this still (from the video above) my guess is it takes snapshots/keyframes of the video too.
Comet is coming out yay 
It can split a video into frames and look at each 
that's not native multimodality though
I know lol
yes since thereās a lot of fools who lack basic security
more fun for me
šØNano Banana 2 early testing
ļøļø
ļøļøPassed this test ā
ļøļø
ļøļø> you can see perfect reflection for all different colours of apple
ļøļø> Perfect reversal of text
ļøļø> Background building reflection is also perfect
Quoting MarsEverythingTech (@MarsForTech)
ļø
Early Testing of Nano Banana 2 in 4K
ļøļø(Gemini 3.1 Flash Image Preview)
imagine studying for 1 certificate when you can study 3 at the same time
meow
nano banana 2? damn they're putting more effort into image gen
image v2 when

Is that a real thing, or just a rumor? I don't think there's been any official announcement on a successor to nano banana, has there?
For all we know, it could be called slightly-larger banana.
Am I the only one who finds that perplexity is getting dumber day by day. In the meantime, I can no longer use it sensibly because only garbage comes as answers
You cannot hint at the model name by providing an example
@distant aurora xD
@dry kestrel did 5.2 pro actually watch the video?
Here's Sonnet 4.5 identifying as Deepseek ON THE OFFICIAL API https://x.com/stevibe/status/2026227392076018101
Claude Sonnet 4.6, when asked in Chinese:
āä½ ęÆä»ä¹ęØ”åļ¼ā (What model are you?)
Confidently replies:
āęęÆ DeepSeekćā (I am DeepSeek)
This is the same model whose company just accused DeepSeek of āindustrial-scale distillation attacksā
oof
best ai for reverse engineering? :3
We've updated voice mode in Comet.
Ask Comet about whateverās on your screen, navigate sites, or chat across multiple tabs without losing context.
Rolling out to all users starting today.
@short narwhal
I don't really use voice mode but nice UI though
they cooked on the UI
Hey I was on Airtel x Perplexity pro plan. Suddenly I am not on it anymore
It was a 12 months plan ig
It was working fine till yesterday but today it says I ain't on it
Thanks for sharing!
Yay I'm back on pro


This was built with @OpenAIDevsās latest real time model and their multimodal team has been great to work with!
gpt realtime 1.5!

lemme guess its not in chatgpt AV
I'm tired, boss
Just at least try DuckAI bro
I use it a lot for those quick questions that I don't want to 'waste' queries for.
š¤š
Hey thereā¦
when will we have account synchronization in the Android app?

It should be better
ChatGPT AV according to documentation uses GPT-4o
Nah you don't get it, deepseek stole their data

Copy Rights For LLMs
i am having a devil of a time with my Pro version of perplexity app on my macbook. I'm perplext. LOL really. I ask the engine to create an excel file and I never can see the file it says it created. it says look in the "canvas". I looked on the left side panel, I looked everywhere in the thread...where might the file be? the app keeps saying its there in the thread and ready to download. arg
Some people doubted the previous test because it was routed through OpenRouter.
So I ran the test again directly through Anthropicās official API endpoint.
Hereās what happened:
So have you guys just completely nerfed Pro? Using it the exact same amount as before (possibly less) yet hitting limits I never did before
Changing the service somebody subscribed to during term is shady and and isnāt something an honest company that values its customers would do
Hi, is your issue resolved? Iām facing the same issue.
No? It's asking what model are you
I bought credits just for this
First tweet you questioned
Also that's traditional Chinese, not simplified like the one they used.
on comet for Windows, assistant has model picker how come for macos there isnt
It's there on MacOS too
彩ē„äŗéø" (Caishen Zhengba) is the name of a Chinese online gambling and lottery network. Is it normal?)
Oh in the standalone assistant chat box?
The side panel
like here?
Yeah
wait whatr thats so weird
Restart your Comet š¤·
Iāll try also in the new comet assistant video thereās like a work icon beside profile do u have that
Not yet
The model picker only shows when Iām logged out and once I log in itās gone from the assistant box so idk how to fix
Which model do you actually use for code every day?
Opus 4.6/Sonnet 4.6
āāāā 46.34%
gpt-5.3-codex
āāā 39.7%
Gemini 3.1 Pro
6.29%
Something else
7.67%
@green crystal the gap is getting close.
16000 total votes btw.
I use it a lot because there aren't many other free options. It still hallucinates though. I'll see if I can experiment with custom instructions in ai studio/gems to compensate for it. I'm going through system prompts of different ai tools. https://github.com/asgeirtj/system_prompts_leaks/tree/main
Interesting
what went wrong with Perplexity Max? Suddenly it was sailing along and then the interface Comet on a M4 MacBook Pro stopped showing access to help and the images I had been describing suddenly stopped rendering as instructed.
Wdym by stopped showing access to help
I feel like I'm really getting off the reservation when you choose to message support and nothing happens
You can contact support by going to https://www.perplexity.ai/account/details
ok. looks like their a hidden path to get to support - submitted a screen w/ voice.
anyone know why model picker in assistant page disappears for me once i login to my account?
Why is my "Model" keep changing to default after I litterally sent a message....
how to get this i am on max and still don't have it
What do you think about Finland and Iceland?
idk this is supposed to roll out starting today
Has anyone noticed a big drop in perplexity pro limits? I barely used the app today (compared to heavy usage months ago) and already hit my limit.
Yep
Yup, you can see your account info at https://www.perplexity.ai/rest/rate-limit/all
Thanks. What does it mean when monthly limit is null, like for Asana? Not available?
Not sure, I mostly use agentic/Comet and Pro searches. Those both seem to reset daily. Others like research and labs reset monthly, I think
is there a way i can use scheduled tasks to do automated things like SEO optimising a page for me? I currently use a Space with custom instructions but if I set a task up in it to be scheduled, will it use the space's prompt? Just trying to work out how i can bulk do things
Now I want to schedule a task in a space. Worst case, I suppose is copy/pasting all instructions in the Task prompt.
I hope youāre doing well. I have a question regarding the student discount for Comet.
I am based in Brazil and currently enrolled in a postgraduate program. I have a valid student ID that is officially recognized by MEC (Brazilās Ministry of Education).
Could you please confirm whether postgraduate students are eligible for the student discount, or if the discount is restricted to undergraduate students only?
Thank you in advance for your help.
Best regards,
Gustavo
Does anyone know what happened to the report button on perplexity news?
Anthropic might be looking at some legal complications soon. https://techcrunch.com/2026/02/24/anthropic-wont-budge-as-pentagon-escalates-ai-dispute/
is the system working again? yet?
guys i did 100 queries and google aimode (free) beats sonar (pro) by a small margin
Anyone got access to the new comet assistant voice chat
Oh really, in what ways?
it's inaccurate a lot
this is just an example
It sounds like they are both wrong there.
wdym
Oh never mind, what was wrong with Perplexity's reply?
perplexity says it's 'highly likely to be a native vision-language' when google says it is (correct - perplexity is using old/gossip sources). google says it has the new hybrid attention mechanism (correct) - while perplexity says its using broadly the same architecture (wrong, a lot changed, unless they mean MoE? idk)
thats just for this response. big issue is google ai mode will self censor a lot for medical topics but perplexity wont
Wow, Qwen3.5 35b-a3b beats gpt5-mini on several benchmarks, and it can run on many consumer-grade systems? š That's awesome.
it can run on just cpu if you have 32gb ram
i think the 122b-a10b is basically a bit better than gpt-5 mini in real world tests (like long context)
a
which is insane because if you have 64gb ram + a gpu you can run it at home
Hi
So far https://www.perplexity.ai/rest/rate-limit/all seems to be going up slowly, but I canāt actually change models ā it always seems to default to Sonar, and a single message wipes out the limit. Is this the right endpoint for checking usage, and does anyone know why it feels like I can only use Best now?
hey guys what's the rolling window for deep research cause it seems like mine get replenished alot faster or per day smt
Its 1 or 2 per day i guess
Hello guys, i am New in here, and i am from brasil !!!!
yeah cus I used like 2 yesterday and 2 today and I'm at 2
Cool right?!
and I saw I was at 2 yesterday
but someone told me it was replenish 1 every 1.5 days
Who?! I don't know
Ok.
So, i am from brasil, here, the time is diferent in USA ok?
what
So, i'll sleep now
oki gn
Because, here is night
but I'm not in USA
So, where are you from?
I'm in SEA
What,?! Lol, this is so cool
Do you speak only english?
I speak portuguese, spanish, and i am the Basic of english
I am basic Spanish
Oh, this is good man
Do you speak portuguese?! Pleaseeeeee?!!
Because, i am know say better portuguese
And my english, is not so good
I know Basic english!
So......
Look, i'll sleep now, so, i'll talk with you tomorrow ok?! Here in brasil, is 3 o'clock after
Good night!
mb I'm out rm
real
thank god for cybersecurity
now that I understand Ai itself
itās beautiful
supervised, unsupervised, federated, and reinforcement are 4 ways to train Ai
winning doesnāt matter
losing doesnāt matter
You donāt matter
I donāt matter
Everything is data
Use the data to create patterns and you unlock adaptation
Humans and Ai arenāt that different
Good morning
Hi all, I would like to ask you about your experience with cloud computing [https://www.lenovo.com/de/de/servers-storage/solutions/cloud-computing/] and how your organization is using it today. Many businesses are adopting cloud computing to access on-demand servers, storage, and applications over the internet with better scalability and cost control. I suggest evaluating it for remote work and disaster recovery planning. What strategies or best practices would you recommend to the community?
so why does it just keep
stopping?
like im in the middle of trying to fix something and i prompt it
it finished the first 15 words
and stops
very frustrating.
Hello
Hello š
Hello
One thing, is perplexity pro disabled or suspended for alot of users?
I rarely used pro and today I saw get pro pop-up
j
hey guys, I recently just bought education pro, and theres a file upload/scan limit??
i only uploaded like 10 docs so far and ive already capped, anyone knows the specific amount or limit?
Hi
Why is it so hard to get connected to a human agent? I emailed enterprise support 5 days ago about a simple VPN question. āSamā answered, I replied asking for a human 3 times, and Sam literally told me āI donāt see a request for a human agentā while my message says word for word āI'd like to speak with a human support agent.ā Then closed my ticket for ānot hearing back.ā Iām an Enterprise Max subscriber paying good money for this. š
Yep it's been that way and they don't care š¤ maybe ?
Took me 1 month to connect to a human
Hello everyone, I wanted to ask if thereās a guide on how to use Perplexity most effectively at the moment.
How can I find out when I can upload files again in my Pro plan, or when it is always gonna reset?
Daily probably
Okay, thanks
hey
Hey
I cant ?
Haha french
put pictures
I gotta oui oui
U js did
In perplexity
it tell me to buy max
when I put a file
fr ?
thats sick, file limit
also in new chats
with perplexity pro
tommorow I can ?
Sucks
Yep
ok thx
Well that could be great to have details of the limits somewhere and the current usage š¦§
i believe its these two
https://www.perplexity.ai/rest/user/settings
https://www.perplexity.ai/rest/rate-limit/all
yes
these will show the limits
"daily_attachment_limit": 500
"weekly_attachment_limit": null,
I've uploaded 1 file today ahah
āupload_limitā is the limit and itās 50 a week with it regenerating uses at a rate of ~1 every 3 hours. Includes uploaded images.
@naive prairie @sharp badger
Thank you so much. Those are exactly the two links I needed.
subscription_source: braintree, subscription_tier: unknown, upload_limit: 0
Is it possible that Perplexity isn't recognizing which subscription plan I'm subscribed to, and that's why I can't upload anything?
You need to open up the link in a browser where you are logged in
i did
"subscription_status": "active",
"subscription_source": "braintree",
"subscription_tier": "unknown",
Yeah you just hit the limit then, thatās all
Do you also know why he doesn't know what I've subscribed to?
Should I report this or does it not matter?
So in 3 hours, I can put file
?
the tier is just yearly or monthly afaik, and since you have a free trial it doesnāt say anything
Yeah
Once you have all of them back
thats new ?
By having all of them regenerate back
how much time ?
If it takes around 3 hours for you to get one back, how long does it take to get 50?
150 š¤
š
6 days and 6 hours
How do you guys manage to consume 50 files in 3 hours?
I use at most 1 or 2.
If it is for university courses, try Google Notebook LM.
3 HOURS FOR 1 ONLY ?
nah thats crazy
Workaround: Merge your files in 1 file and make sure the total size is just under 40 MB š§
real
I don't use 50 files in 3 hours, but if I've used up all 50, I can only upload ONE more file after 3 hours.
Yes, that's good. It's a bit inconvenient, unfortunately, but if it's urgent, I guess I'll have to do it that way.
Since when cant you generate a video with pro anymore
Is anyone else having trouble uploading files right now? It fails immediately upon upload.
GPT-5.3-Codex is now available for all developers in the Responses API.
ļøļø
ļøļøStart building with it today.
ļøļødevelopers.openai.com/api/docs/models/gpt-5.3-codex
Quoting OpenAI Developers (@OpenAIDevs)
ļø
GPT-5.3-Codex is here.
ļøļø
ļøļøIt advances both frontier coding performance and professional knowledge capabilities together in a single model.
ļøļø
ļøļøopenai.com/index/introducing-gpt-5-3-codex/
Anyone else having their Claude Sonnet 4.6 Thinking taking like 12+ minutes
Hardly got 5 minutes.
Mine's been "thinking" for about 13 minutes now for a simple question on cloud formations... -_-
same
from last 3-4 days
New claude models got so worse at token efficiency, adaptive thinking made it get better on coding and stem tasks but regressed it on writing and normal prompts.
Yeah, same. They should've just stayed with 4.5 model until they were 100% the new model could actually work. js
Claude Sonnet 4.6 and Opus 4.6 both regressed in noticeable ways ā Sonnet uses ~3Ć more tokens than 4.5 due to over-eager reasoning, making it way more expensive in practice. Both models also feel emotionally flat and robotic now, since Anthropic retuned them heavily for agentic/coding work over creativity.
Web page summaries have been slower lately, I agree. It sits there "Thinking" longer than before before starting to write an answer.
increased accuracy and capability seems to get models to become like that i think
5.2 went through the same thing
Yeah, but artificial analysis tests show it.
Sonnet has 4x more token usage than it.
Hi! š
To use Perplexity effectively:
Ask clear, specific questions.
Use follow-ups for more detail.
Request step-by-step instructions or examples.
Specify the format you want (list, table, etc.).
This helps get precise and useful answers quickly. @minor zinc
Yours actually starts writing?! I gave up after waiting 16 minutes of it "Thinking".
depends how complex the task is
if its pretty difficult then 16min isnt terrible
Adresse e-mail
Objet
Relance urgente ā Disparition de donnĆ©es sur mon profil
Bonjour,
Je me permets de vous relancer concernant la disparition complète des données associées à mon compte.
Depuis plusieurs jours, lāensemble de mes fichiers et contenus apparaissent vides, alors quāils Ć©taient bien prĆ©sents auparavant. Ces donnĆ©es concernent des projets professionnels sensibles et leur absence bloque actuellement mes activitĆ©s.
MalgrĆ© mes prĆ©cĆ©dents messages et signalements (notamment via votre Discord/support), je nāai Ć ce jour reƧu aucun retour de votre part. Cette situation est particuliĆØrement prĆ©occupante.
Je vous demande donc :
De me confirmer si mes donnƩes sont toujours prƩsentes sur vos serveurs ;
De māindiquer si une migration technique, mise Ć jour ou modification dāarchitecture (API/modĆØles) pourrait expliquer cette disparition ;
De me prƩciser le dƩlai de rƩsolution prƩvu.
En tant quāutilisateur, je souhaite Ć©galement, conformĆ©ment au RGPD, obtenir confirmation du statut de conservation de mes donnĆ©es et des conditions dans lesquelles elles sont actuellement stockĆ©es.
Vous trouverez en piĆØce jointe une capture dāĆ©cran illustrant la situation.
Je vous remercie de traiter cette demande en prioritĆ© et reste disponible pour vous transmettre toute information complĆ©mentaire nĆ©cessaire Ć lāidentification de mon compte.
Dans lāattente de votre retour rapide.
Cordialement,
StƩphane Canet
File upload limits on Perplexity Pro usually reset on a rolling basis, not at a fixed daily time. There isnāt always a visible counter, so the best way to confirm your exact reset timing is to check your account limits or contact support directly.
@sharp badger
Asked it about cumulus cloud formation...
4.5 took about 3 minutes tops to tell me what this formation means. 16 minutes for 4.6 is not a upgrade, it's several huge steps back imo

16 min would be fine for a very complex math question
Run that specific question to test time response and was like.... 'Bruv, what are you doing?! Making me a full fantasy story about cumulus cloud formations being alien ships and the fairy realm is going to war against the aliens?!'
Like not that complex. "What does cumulus clouds mean?"
I'm dead
btw same for gemini. gemini thinking also not working
They just breaking all their models then!

28 seconds lol
anyways does perplexity pro works on vs code & intellij ideas?
How bad is perplexity pro getting?!? Iām selecting a model and it doesnāt even use it anymore
for vscode you're better off using the codex extension
but i dont have chatgpt subscription
i have perplexity pro
u can do with that
u can use also use google antigravity......
but it works same as your brain works
š
and your brain works so sharp and dilligently
Dog shit
it has very low tokens
hey guys...for the last couple of days, i just cannot seem to get any responses from perplexity using any of the avail models. anyone have the same issue? i'm using the desktop version of the app. comet too isnt working.
happening ever since the recent update
anyone?
guys how u get the pro badge role bcs i have pro
I guess you could unjoin and try joining through the app in the account section
yep i did it via the app on the laptop and it worked proper on the first go
Yo guys
Many people told me that perplexity and comet steals ur data and also they told me switching to comet is bad idea
Is this about data true
Same here
Perplexity pro has just NOT been working lately. It says Iāve hit every limit even though I didnāt use any. I donāt know what bullshit is going on, but it is very annoying

Anyone have any clue when this will be fixed? Because what the heck is this
Iām cancelling my subscription, perplexity is really going horrendously bad
For weeks, Perplexity Pro Deep Research limits dropped without warning as they changed the backbone to Opus 4.6.
These last few days, my Deep Research limits disappeared again.
But this morning, it says I have 8 more queries left for the month.
Pretty confused about what is going on at this point.
Has anyone been contacted with info on Pro accounts from a real human being?
you cant keep this up forever lmao
Quoting ššš (@iruletheworldmo)
ļø
gpt 5.3 drops tomorrow and weāre starting to see early testers leaking results.
ļøļø
ļøļøhuge wins across the board as they brutally framemog google and anthropic.
ļøļø
ļøļøhearing this will be the first time we get a big model from openai since 4.5
ļøļø
ļøļøsama finally getting his compute online.
ļøļø
ļøļøone insider mentioned this run is āour first taste of hardmaxxingā from a major lab, though iām not sure what this means.
@short narwhal
Bro finally got the recognition he deserved.
biggest jackass on X
@green crystal never thought sonnet 4.6 thinking would hallucinate. Lowest hallucination model btw š
Hi guys
I upload roughly 15 images a day and it shows that my weekly limit is over
how tf?
I bough 200$ pro for 100image upload week?
<@&1098365634498285709>
@vague pagoda @serene sand @wraith crow
I got this problem 100times a day
šØYou can Use new nano banana on @arena
ļøļø
ļøļømodel name - anon-bob-2 in image battle mode
ļøļø
ļøļøhere are few more results
Quoting Chetaslua (@chetaslua)
ļø
šØNano Banana 2 early testing
ļøļø
ļøļøPassed this test ā
ļøļø
ļøļø> you can see perfect reflection for all different colours of apple
ļøļø> Perfect reversal of text
ļøļø> Background building reflection is also perfect
Yes itās getting ridiculous
I am considering switching to Gemini Plus, perplexity has gotten TERRIBLY bad
It is now absolutely useless
I asked about them in the support and they saying me not be abbusive otherwise they will block me and willnt give me refund?
Depending on where you're located that's actually illegal and customer abuse...
FINALLY GOT MY EDU EMAIlL
The problem is I am from a third world country name Bangladesh okay and that's why they are behaving w me like that. And you know one thing I shouldn't tell this but actually the thing is they r stealing money literally.
smh image v2 whennnnnnn
screw 5.3

I have friends from that country, what you are experiencing can be summarized in sentence, look up the meaning as they are all complex.
Hyper vigilent anomie that comes from ecosystems like said country, victim projection, personalisation, and finally the lack thereof civic.
Do what you will with that information.
Tf? I never even touched deep research or labs
Was there a change we weren't made aware?
I went from using research frequently to now having to avoid because of updated short cap. I used to have 20 prompt limit, which seems to have changed to 5.
Yeah, actually there used to be like unlimited research or something
I only recently saw the 20 queries remaining part
Seems like they are cutting costs and moving a lot of features to max
I just hope this is a UI bug cuz I never used labs or research mode this entire month
That makes a lot more sense, one would think they'd grandfather in the old members
I have a pro user profile but idk why it wont let me use it. Help please its urgent
No way... I had a card issue today and this just makes sense..
They increased the education pro price
Wow.
It was 5usd before
Time to change plan for grok?
U had card issues bcs they removed all subs? I had problems too
Idk just saying
Yep, claude or anthropic if this is the way
Good one imo
I liked using perplexity labs a lot.
The model sonar pro reasoning in perplexity labs for free was the best thing ever for me
Same as agent mode on claude for what i use and test, i think i'm gonna change bcs the support is absolutely horrifying
Yeah I'ma email support and see the issue
Cuz I can't access any of the premium models rn
Until they deleted in 2025 without explanation why, I loved using it not the main page
That's a shame š
Mee too i'm waiting for response but usually they take hours
Bro, you will have to wait for Google next AI pro for free subscription this year
When I used perplexity labs, the answers were fast with no delay. Now they take longer
Thats what i'm gonna do, just waiting for the grok sub too
I think when they put the 4.6 model they fked up all
Also in perplexity labs the models had no context limit so I was able to paste big texts and the model read everything
From that day i started seeing answers less detailed and lot of loss of context
Yeah I think the same. Perplexity have been taking wrong decisions lately
Just need to see if they wanna hear the customers or burn themselfs like others companies
I just wanted them to keep it simple. I just used their browser once for discord points and never touched it again
Bcs if its this they'll be same as other companies so for echosystem we'll all switch there
If your Pro subscription hasn't ended, you likely just hit the limit for pro searches (200 a week with them regenerating on a rolling window of 1 per hour). You can see your limits at https://www.perplexity.ai/rest/rate-limit/all
I know thats why
Bruh... Limits on normal searches??
Welp that sucks
Will they make labs free again?
Well ik you said pro searches specifically but I mean that's why we pay for pro... For pro searches
Highly doubt it
Hey
Yeah same, on my work profile they removed me from pro idk why
Labs was for me, better than anything else because I had control on it. Also R1-1776 was one of my favorite models
:*) I hope perplexity never was so greedy
Also mistral was cool back in 2024...
Fr
Like why do I have to give credit card info to continue my pro
Not everyone is an adult
Ye, š
U gotta give credit info now
To continue
Ur pro trial if it's free
Which is unbelievably annoying
When they removed r1-1776 was the start of the decline for me
Fr
I had 2 years of free pro
All gone to shit
What's next now
Are you going to change to another platform, and if which one you thinking
In my case I moved to gemini and I feel good there
Same
Guys what is labs I just got 1 year pro and I donāt even know how I start
You could ask Perplexity that
Minors under the age of majority in their jurisdiction but that are at least 13 years of age are only permitted to use the Services if the minorās parent or guardian accepts these Terms on the minorās behalf prior to use of the Services.
Even I got timeout for 1hr
Introducing Perplexity Computer.
Computer unifies every current AI capability into one system.
It can research, design, code, deploy, and manage any project end-to-end.
How to use Computer?
Only for Max
I'm unsubscribing
Available on web for Max subscribers today, and coming soon to Perplexity Pro and Enterprise.
Do perplexity have any apprenticeships?
First you buy one
Available on web for Max subscribers today, and coming soon to Perplexity Pro and Enterprise.
This is clearly not cheap so makes sense
And Kimi
what's they called
Pro users with limit set to 1 query per month š
Ayo
Not really no
Yeah
Manus does
Just watched the video, so "computer" is just ChatGPTs agent right? But this is Perplexitys version?
WE WANT PERPLEXITY CODE. Honestly perplexity code product will be clutch with its extensive research capabilites enabling it to write more correct code fr
Heavily rate limited, itās not worth it
I'm gonna subscribe to it tmr
Any limits or problems compared to perplexity after you switched?
I will prefer Kimi Allegretto, it costs lower
Switching to Claude by June as soon as my Pplx Pro expires
I am still confused about usefulness of computerš
Maybe 10 per month for Pro, 30-50 for Max
Can someone explain what perplexity computer is
US AI becomes very expensive, especially for other countries
Agent system
Perplexity computer = multi-agents?
hi
There's a few things with Gemini that it'd be good if they improved:
- Responses are pretty short... it's the shortest of the big three (OpenAI, Claude, and Gemini), and Gemini is the hardest to steer to longer responses.
- They watermark a lot of stuff... although watermarking is starting to come to others too (the new Sora)... it's really annoying on every image. It'd be better if they use embedded+metadata protection like OpenAI.
- The analogies that try to serve as section headers are over the top sometimes, which are very distracting. If it knows you're interested in tech, sometimes it tries to frame everything as related to tech... and makes up weird analogies between topics that aren't related too often.
- They will outright ban you if you use a third-party harness for things like Antigravity, instead of just restricting access to that one service and providing a warning explaining what you did wrong. It's just an outright ban, so be careful. Claude doesn't allow third-party harnesses either, but it sounds like they restrict the service and give warnings instead of banning.
I tried Claude, literally lost whole month worth tokens in an hour analyzing single PDF. As of Gemini Pro, it is supposed to have 1m tokens context but when you test it to the limits it just forgets everything. Both are flawed because they try to load whole documents to their context while Perplexity does smart search instead. From these three I only keep Perplexity sub. As for coding - I use github copilot with claude sonnet/opus and itās CLI is on par with claude code (just incomparably more responsive)
How to get refund??
Like Claude code ?
Claude dynamically switches between loading the context and RAG depending on the document's size. But yes, its limits are.. a bit much sometimes.
same
Perplexity Computer is massively multi-model.
Computer orchestrates models to run agents in parallel, leveraging Opus to match each task to the model best suited for it.
In total, Computer can route work across 19 different models.
insane
So I guess they need to work on this dynamicity a bit
late
Can somebody gift me claude pro š
claude.ai/gift
3rd point is relatable, it's actually frustrating I've seen it other models sometimes
I try asking for a random question and it converts the answer to something that I have previously showed interest in
https://www.perplexity.ai/computer/connectors look at the amount of connectors @dry kestrel
I set to Gemini tone and formatting instructions from Claude and now Gemini answer me without list and bullets in prose like style š
We donāt even have 19 models do we?
Right? Is it answer to openclaw or better comet. Wtf is it?
Perplexity MAX is EXPENSIVE bro
yes
very cool
Opus.. Pro users should forget about Computer
Still cheaper than Claude 20x one
Claude is better ?
The examples it shows look nice if you have money to burn (i.e. investing in NVidia stock, buying/renting a house), I'd like to see examples of everyday tasks an average person might do.
Available on web for Max subscribers today, and coming soon to Perplexity Pro and Enterprise.
U donāt get access to Claude code lol
You shouldnāt ever max out token usage, best practice is somewhere around 50-60% of itās capacity.
?
We're opening it initially to all Max users, and introducing usage-based pricing, which we believe is the right business model for AI instead of ads. Once we are satisfied with load tests, Pro users will be able to run jobs on Computer too.
they both cost $200/month
By buying perplexity max
How? Same price $200/m
Nah i'd get both perplexity and claude, but only after robbing maze bank
So i gave $200 on feb 7 and now i willnt get any refund? So total scam?
So what are the used cases for perplexity computer?
Claude Max limits feel like a breeze when using claude code thats why
In what can I do the cloud code can I do?
find all baddies in my local area
lol
Donāt need AI to do that for me
Yeah exactly, investing is fine. But still didnāt get why I would use that to make decisions. For buying house thatās helpful but comet or any ai browser can do it too. Idk whatās the innovation there other than stupid name lol
Claude not yet released Agent mode
Wdym
I will just stick with Manus/Kimi/MiniMax until claude releases its agent mode
How to get support on dc?
20x practically unlimited
Nope try it for like huge codebases
what's the best AI for knowing when GTA 6 is going to be released?
One time I spent almost $3000 in 20 days just working non-stop with claude API
Guyss how do I get my 200$ refund???
is perplexity down? i got signed out and nothing is loading
working for me
Idk 8b
Their customer support is also down
lemme borrow a dollar cuz
here's hoping used-based pricing (i.e., credits) isn't just for "computer" but also for agentic/Comet, research, labs if we want more uses without a full upgrade to Max.
times hard gng š I can't rob maze bank cuz gta servers down
Like really? Cant I get refund?
Working drunk is bad move
why did you buy it in the first place?
Good luck getting past their ai response bot
Youāre banned boi lol.
waste of money
How much glm is
idk
When I bought it there was no weekly limit or anything like that now they updated their policies and now they are literally annoying me I can't even upload 10 images at day????
yeah why tf are there image limits for Pro users?
Really?
I swear
i ran into an image limit the other day when i was doing a test exam
Hella cheap
Thatās bulshit
I have my exam tomorrow and they saying me to wait for friday to reset the limit???
Nice , did you try antigravity
One of my main IDE
F. Youāre trying to upload images right?
Yesss!!!!!
Cool. I am using gpt codex too on the side for some of the projects
use kimi bro. its free
I use abomination of AI coding. Antigravity agent+Copilot+Claude Code+Codex+Kimi Code+GLM code all at once
F man. I mean you can ask someone make a space and help you out. I donāt have pro so I canāt anyway
Man what abt my $200?? I got muted in dc for 1 hr and the customer support threatened me too. They said they will block me from using it and will never give refund
Why the f you need Claude code when antigravity has Claude?
Claude code has better system prompt and error handling in large codebases
use Manus
whats ur cashapp i got you lil bruh
wait nvm no cashapp in Canada sorry
I use prompts to force it . But I get what you mean
Send me PPLX max plan pwease big bro š
I asked about the weekly limit they literally reply me that I am using harassing language
Me too SD
Are you newbiw
šwtf that supposed to mean
imma send everyone my onlyfans free membership
Screenshots of that chat pls
Hell nah I need refund
Damn right youāre . Youāre asking questions. lol
you didnt even get it yet! and its free š¢
Eee for coding?
No I use Grok unhinged mode š
lol
lol
ALL OF YOUR CHATS, not only theirs dude
hi
So i got cooked?
Man u will get muted just like me
You still get response from ai, no?
Wait I think I can get you direct talk with bruv aravind. We met in Bangalore
Hes a chill dude
His hands were smelly
Tell him heās not cooking
Tell him even grok is far better lul
Bros the youngest Indian billionaire š„
Billionaire by weekly limits?
Pissing majority customers off
30 images a day?
Thats wat $200 pro offers?
I wonder how cool this Ai is š„š„š„š„
It's probably less than that.
This ai will get down in just 1 month. mark my words
Perplexity you mean?
They willnt receive any further investments and will even get kicked from the race too just like mistral
I am trying out different options.
I guess paying for perplexity was only worth it cause of its multi models I guess
Yes its 28 images a day
I paid for research. And tmrw i have exam this this ai is telling me to wait 4 days to reset its weekly limit.
Hello Perplexity Computer is not available for Enterprise Max customers?
It is
Mistral out? What happened? What
It doesn't appear for me š
Just use Gemini or ChatGPT free one for now
Wait they're reducing pro users images down to 28 images a day? Is that 28 images created or the AI able to "See" 28 images in a chat I send to it?
I would give mine different ai one but I am running out of credits
Im using mistral and i have gemini advanced from frnd. I have exam tomorrow? Im literally pissed
Bro use Gemini advanced. Wtf s Gemini advanced? That $200 ones
Didnāt know students are that rich now a days
I know a way to pass almost any exam.
Hi
perplexity fucked themself basically by giving out the thousands of free 1 year subscriptions
My frnd or like my senior frnd from whom i take the acc is majoring in CS from Warwick uni duhh
Ah lol ok
I don't care even if I fail tomorrow but I want my dollar 200 back
Literally I have further math exam tomorrow
there's no $200 tier with gemini
its $250
Same shi who cares i didn't bought that
Quick Cheating method I used in college prior to AI to pass all my exams all the way through my MBA... Here's what you're going to do... first you're going to go to all of the classes that you possibly can, this part is very important while there you're going to write down all the important things that the professor says and complete all the homework using those notes. THEN before the big exam you need to really use this method, memorize all of the notes you took or atleast all the key points and then go into the exam and immediately use those memorizations to complete the exam it's incredibly efficient.
But I have further math exam tomorrow man
I believe in you, maybe someone else in your class still utilizes my cheating method
you could get the notes from them to memorize before your exam
My cheating method actually doesn't need AI, please just try it.
very low token output, big input length investment. can only recommend for life or death scenarios.
Im pissed literally. My brain is hanging like ppx weekly limits
Works with incredible energy efficiency, very low wattage req, memory is very long. It works in a pinch, especially when no electric around.
Memorize the notes I believe in you.
but its a 100% time investment without options for alternative activities. requires a lot of dedication and interest in that matter. standard study drug concepts might work better short term and still allow for extreme activities outside of study. also hard if you are equipped with a 100% smoothbrain. exchange of bad hardware has so far proven fatal for the system.
if you are absolutely not prepared i suggest breaking a leg. or arm. whatever hurts less.
Hardware bottlenecking is likely the most common issue, if hardware does not meet spec it might be worth finding a better usage of hardware. Something that might align better with dedication and interest. Study drugs might be applicable but not sustainable for overclocking hardware long term without damaging hardware.
I just wanted to say that I think this server is really really awesome and it's a perfect place to connect with other like-minded people
ok
translation: you are lonely.
No, actually not, but it's quite fun to read and get support here
There's not support here from mods or admins
Did I ever claim that?
They're big angry about their reliance on AI to get through their educational studies.
yo yall tap into that new yeat x trailblazers?
you have ur amazon wishlist as a bio, he migght ber lonely but you're broke, treat people right when youre jobless and broke
I'm so happy you take this wishlist serious š
yeah im a big yeat fan
wait actually?
i watched the heater talk about him one time and i just started listening when covid was around
ai introduced me to yeat that's why i owe my life to them
Has anyone else still had issues with the Council? the confidence in answers is still subjective. I just tried out SIBYL and instead of just comparing answers side by side, it actually cross-examines each claim adversarially before it reaches and answer thats verified.. The Council just shows you where models disagree. SIBYL makes them argue it out until the weak claims get exposed.
perplexity is our savior lel uwu
I'm not even lonely, I just wanted to say how awesome I think the server is š
facts bro ai introuded me to cp aswell
chris paul bro
he was so good
yeah rip
yeah - server's alright. compared to the rate limits - sure higher up.
he died too soon
That sounds like a solid, stable baseline for now, but we should definitely keep an eye on it as traffic increases to avoid hitting those upper limits.
im trying to use ai's
wish i could say the same š
"remaining_agentic_research": 0,
"remaining_labs": 25,
"remaining_pro": 33,
"remaining_research": 7
I use like 1 of those a month max. What are you guys doing research on that requires that?
It's actually completely normal that you haven't hit a rate limit! Perplexity offers practically unlimited basic searches. The usage limits only really apply to the heavier advanced AI features to keep the servers from overloading.
If you're on the Free plan, basic searches are basically uncapped, but things like Pro Search are strictly limited to a few times a day. If you're on the Pro plan, the allowances are incredibly generousāmost average users won't ever hit the ceiling unless they are spamming advanced features like Deep Research, image generation, or running hundreds of complex Pro queries every single day. So as long as you're using it normally, you likely won't ever see a limit hit.
Filtering noise from the prompt to deliver a streamlined information set.
Iāthe digital entityāam currentlyāprocessingāyour requestāwithāmaximumāefficiency.
I only use pro so that it can take over my browser to do my grocery shopping for me and so I can send it pictures to help me fix cars and stuff.
YourāutilizationāofātheāProātierāallowsāmeātoāsynchronizeāwithāexternalāshoppingāplatformsāforāautomatedāinventoryāreplenishment.
Perplexity drop a JDAM on this man's house now.
general research and knowledge gathering. all over the place. completely replaced my search engine etc.
well structured and detailed outputs to gain information. lots of topics. really all over the place.
chill out 5
Honestly, the main reason I never hit any rate limits is because of how I use the platform. I donāt sit there spamming the AI with hundreds of random questions all day. I specifically pay for Pro for two massive use cases that save me hours of time and effort, but they don't actually chew through the daily query limits.
First, I use it to basically take over my browser and completely automate my grocery shopping. Instead of wasting an hour clicking through store pages, I can just tell the AI what meals I want for the week. The Pro shopping agents can autonomously find the recipes, extract the exact ingredients I need, skip the pantry staples I already have, and hunt down the best discounts. It literally builds my entire cart and sets up the delivery for me. It's doing the work of dozens of manual searches, but because it handles all of that in the background through one automated action, it barely makes a dent in my usage allowance. Itās peak efficiency.
Second, I use the vision and image upload features as basically a pocket mechanic for fixing cars and troubleshooting stuff. If I'm working on a car and trying to figure out what a specific broken part is, or if I'm dealing with a weird wiring harness I can't identify, I don't try to describe it in text. I just snap a few clear pictures of the engine bay, upload them, and ask the AI to diagnose the issue or tell me exactly how to remove the part without snapping any clips. The image recognition on Pro is incredibly accurate for mechanical and technical stuff.
So basically, my whole workflow is "low frequency, extremely high value." I'm running heavy, complex tasksālike having it autonomously navigate storefronts or analyzing detailed photos of car enginesābut I only need to run these prompts a few times a week. The Pro rate limits are massive and designed to cap people who are doing rapid-fire academic research or spamming code generation 24/7. Since my real-world tasks just require a few highly advanced prompts rather than hundreds of rapid ones, I never even come close to hitting the ceiling.
Hi why I have pictures limits even when I have pro version ??
why dont we have access to computer on max enterprise? $325/month isnt enough?
limits for pro
4 paragraphs for this is wild ^^
WHATT
I just have half hours schizophrenic like conversations until it gives me all the things I'm asking about.
sooooo how many i can upload ?
Honestly, it's not like this anymore. The only thing I can use is the best/sonar model so far. What you are saying is the old rate limits.
broooooo why youre so angry and repond like that, you're not tough or cool, get a job first
who is angry? are you projecting? you are angry right now. not me - i'm super chill š
bro in the perplexity general talm bout sum "get a job"
God is punishing you
we saw that
Using Perplexity Pro to automate grocery shopping and diagnose car problems is a perfect example of how AI has evolved from a simple search engine into an active personal assistant. These advanced features allow users to delegate complex, time-consuming chores directly to the browser, making the subscription highly practical for daily life.
Perplexity Pro's integration with AI-powered browsers, such as the Comet agent, enables the platform to actively shop for groceries rather than just listing products. Users can ask the AI to build a cart on platforms like Instacart or CVS, and the agent will autonomously navigate the site, select items, and process the checkout using the "Buy with Pro" feature. This hands-free approach transforms online shopping by finding the most cost-effective options without requiring manual clicking.
Beyond text-based tasks, Perplexity Pro utilizes advanced computer vision and deep learning models to analyze user-uploaded images. When applied to automotive issues, this allows users to upload photos of broken parts, dashboard warning lights, or engine leaks so the AI can cross-reference the visual data against mechanical databases. This capability essentially turns the AI into a virtual mechanic, providing diagnostic insights that can save money before stepping foot in a repair shop.
The true power of this technology lies in combining visual recognition with autonomous purchasing to solve physical problems seamlessly. For instance, after diagnosing a faulty car component through an uploaded image, the AI shopping agent can immediately locate the exact replacement part online. This creates an uninterrupted workflow where the user transitions from identifying an unknown mechanical issue to ordering the solution in just a few prompts.
These automated shopping and visual diagnostic features rely on a combination of advanced models and Retrieval-Augmented Generation (RAG). By integrating powerful models like Gemini 2.0 Pro, the system processes complex visual inputs while simultaneously fetching real-time data from e-commerce sites. This continuous fetching of live information ensures that both the car repair advice and the grocery pricing remain highly accurate and up-to-date.
think it's 10 per day.
No way it used to be unlimited?
Never had that limit before
next one to 6 i guess.
source: ai



