#💬│general
1 messages · Page 542 of 1
@distant aurora you know whats funny
the model is first doing web searches for your prompt lmao
to see if it already exists
Share thread
it happens with every prompt
Iran talks failed to reach a conclusion. I am worrying this results big fall situation in AI market.
Seems to be digging blogs.
so
if you instruct it to give links to what i uses
it actually does it
and its very accurate
well alright i guess
i wonder if i should attempt to cook with it
so far gpt and gemini are good at that
Why is my perplexity pro asking me to upgrade to do things that worked yesterday?
I did the PayPal 1 year deal and was working fine and now I need to upgrade to do simple queries - my second question of the day
i think its just pretty good at search but a dumbass at calculating
I am not sure why but even after a pro subscription, my deep research limits for the month are exhausted?
There's a 20 a month limit with usage regenerating 1 use every 36h
Everything time I drop a file on pro it asks me to upgrade to max - what is going on
You reached the 50 a week limit for file uploads
is this for labs or research? it was much more than this previously if i rmbr
Labs has a separate 25 a month limit (that also regenerates)
When did this change -
were these limits reduced recently?
in the past couple of weeks
Nonsense
From 50, yeah
thats sad
Yeah - I am really going to pay $167/month to do more
k2.5 thinking
How and where do you do this exactly?
I dont see a reason to renew the subscription and pay given that the limits have been reduced so heavily
He was being sarcastic.
Well Reddit link that perplexity provided shows how disgruntle users are. Let’s see what they do, because this is silly
not good
its the same on the pro users feedback channel;
https://discord.com/channels/1047197230748151888/1450454558785671188
Each project allows up to 50 files uploads or is it the general limit for all projects?
hi
Files added to a space as context do not count towards the limit afaik
Guys please, can you give me 5 reactions to this message, reacting with this star 😄
.
Hi ajev
Hello
I believe that here
I need the support team
Ohh
I don't know about that, try to tague @wraith crow
Ohh
Inside chrome settings i believe
how
Chrome Settings
@distant aurora https://x.com/i/status/2023759361739264124
@green crystal grok 4 heavy be like https://cunnyx.com/i/status/2023725901188403560
Some guy in the comments said they're just testing a new system with the old model for now.
Ohh they're showing the agnets it uses aswell?
Literally on the same post.
I wonder what bro is smoking.
OpenAI is finally giving us the "Manual Transmission" for AI. 🏎️🧠
︀︀
︀︀OpenAI is reportedly developing a new configuration setting that will allow users to customize their default AI experience more granularly.
︀︀
︀︀This update will enable users to pre-set their default model (such as GPT-4o or the upcoming GPT-5 series) and, more importantly, the "thinking effort" or reasoning depth for each interaction.
Quoting TestingCatalog News 🗞 (@testingcatalog)
︀
OpenAI is working on a new model configuration setting that lets you set the default model and thinking effort.
︀︀
︀︀I would choose GPT-5.3 👀
Translation "we lost the Ai race"

Bro is smoking keta
Yeah, HLE is one of the benchmarks that some companies like Google have made large progress with lately... and since he is very focused on 'beating' the benchmarks with Grok 5, that's probably one area that he knows they're not ahead on, so is probably seeking to minimize that one as not useful like that. That said, it does include some really niche tasks that require things that aren't as relevant to engineering (needs really good obscure language support, for example). Some things are, though, like it needs pretty impressive image understanding to work.
All these benchmarks are to just give an idea where a model sits in terms of ability
did they dissolve the file limit for pro users?
It seems that the comet browser had a recent font update. I tried to change it in the settings and it was not possible. The new font is bad to read
The new font is the almost identical to the one used on Claude webapp
I emailed earlier to speak with a human consultant and so far they have not responded about the limits
How do you change ai models?
i found a thing
like for free they giving use of kimi k 2.5 thinkingfor coding
why use build app feature and not use this model thsi one is the best for coding



what happened? why is there so many limits now and why doesnt mmy model get saved? i have to keep selecting it over and over
Idk nothing wrong with mine

i used to be able to upload screenshots and files without limits, now it pops up asking for more $
yes limits got nerfed for pro users
yikes
how can they compete with gemini / openai / claude if they offer $20/month without any upload limits??

Hey everyone in #1111767882007842916-support or #billing, anyone else dealing with delayed responses on unauthorized charges? I've been waiting 5+ days on an email to support@perplexity.ai about a random $20 hit from Feb 12—no reply yet, super frustrating.
Need tips on escalating for a quick refund/investigation without sharing details here. Who's gotten fast human help? Thanks!
Has comet font changed for anyone?

@dry kestrel @green crystal @gloomy cipher https://cunnyx.com/i/status/2023814077097836933
@green crystal some guy from xai team https://cunnyx.com/i/status/2023793261714768030
@kosa12 The model id is just a placeholder we had to use for beta release. It's using 4.20 underneath
I can’t use best or the other model for what I need cuz mine is jb. The other models are what I needed bruh…
Hmm....
The pricing didn't change. 😢
goofy font
hm
I received a free subscription from Moj Telcom when I checked my email, but the subscription has now been removed from my account. What is the reason?
Use Claude
🚨 Sonnet 4.6 Launched
︀︀
︀︀Here is the benchmark
︀︀
︀︀To access it by yourself goto the link in the comment
Quoting Chetaslua (@chetaslua)
︀
🚨 Claude Sonnet 4.6
︀︀
︀︀They renamed sonnet-5 to 4.6 ( reason is simple - they are saving 5 for a big jump )
︀︀
︀︀In my test Sonnet 4.6 is a very capable model
︀︀> 1/2 price of Opus 4.6
︀︀> Around twice as fast
︀︀> Prompt adherence is God level
︀︀> In a few examples it beats Opus 4.6
2. No advertising.
No DMs promoting your products, services, or servers. You may post your own project in #😎│cool-projects only if it adds value and follows the no-spam rule.
??? Like every month of first month
Every month
First 3 months only
Doesnot work lol
Ahhh. Nvm then
How to use Agentic Financial analysis?
It's a benchmark https://www.vals.ai/benchmarks/finance_agent
Private, domain-specific benchmarks in legal, tax, and finance.
No Jesus?
Has anyone tried grok 4.20
Hello, my Perplexity is broken and has been refusing to generate images for about a month. It says that image generation is restricted in my region, even though there are no such restrictions in Poland!!!
not good
In what ways did you find it deficient?
everything
True
Who at Perplexity just decided "Hey, lets change the font" into something that is totally not modern.
Huh?
I loved that old font, and it would be cool if there was an option to CHANGE it
PPLX just updated to Sonnet 4.6 lol
basically like within 5-10 mins of Claude's own platform
I guess I'm the only person who likes the new font 😢
I want Google Sans
Or Samsung Sans
Yay was hoping for Sonnet 5 but 4.6 will have to do!
or Poppins
allegedly Claude's own free tier is very good now
allegedly
I think the free tier limit is still pretty bad with Sonnet tho?
for Claude Pro users, how many Research can you use per session with Opus 4.6?
how does it compare to opus 4.6, on PPLX i exclusively use opus for every task but its too slow, should i swith to sonnet
the 5 hour session? pretty sure it was 2 or 3 lol
I think Opus 4.6 is only slightly better tbh, but slower like you say
yes, so that hasn't changed from when I tested it with Opus 4.1 months ago
Max x5 seems to give a lot more though
x5 is pretty much the same value ratio as x20
sonnet 4.6 is too good!!!
What’s so good about it
coding
no way higher research limits will happen with Opus 4.6 being used for PPLX Research
And Opus 4.6 in Pro suscription?
Perplexity coding agent wen
What’s the research limit on pro ?
perplexity claw wen
It says “unlimited research”
Sonnet 4.6 launched??
I don't need opus for that, it's not that good to change the 600 research per month to just 20, i just need it to read a ton of sources and answer me man, its a context window heavy task, so a cheap model should be used with increased sources, not the other way around.... It's research for a reason
Im pro user and i cant use opus 4.6
@vague pagoda yo Claude 4.6 is not available for perplexity pro in India I can’t use it
Yeah same in France
I hope you all agree
What platform are you on?
Mobile
That’s only for max users
Works just tested
He sent a announcement just now for pro too
Does default search not do that why use the research mode
The app probably needs to be updated first before it shows up
Maybe ya didn't update the app?
yes I think many people have proposed giving back the option of the previous Research
Guys now got it mb
It was just announced. It will probably take a bit to roll out.
Try using yhe website version
Thank u guys
You're welcome :D
The differentiating factor is the number of sources used and the higher context window/ memory used
It is available. Update your app
I updated the app just now still opus is at max
I update too now sonnet and opus are max :/
Sonnet shows up for me lol
Opus is available to Max only
Only Sonnet 4.6 for Pros
What’s the point of getting max ?
@vague pagoda
Im pro
More file uploads, more research queries, more labs, access to more premium models like Claude Opus, access to model council
Its ok right now
Lowkey bored chat
XD
Nah it's okay mate you're not stupid or anything
Lol, are you? @bold holly
Doing fine :P
Discord flags you as a spammer
You're flagged as a spam because you're a new used :P
I need proof
I mean I didn't need to know that but okay :P
Yeah no
HARD PASSSS
Yeah I’m not 5 lol
I could say some info that would also be irrelevant
Hard pass i definitely think you're a bot-
Really at BYD ?
Im actually Elon musk ceo of Tesla
I'm also gabe newell the CEO of valve
🤮 tesla
I'm actually spongebob and posting from a pineapple deep under the sea. Nice to meet you Mr. Musk(et)
What's the best model for perplexity searches now on the pro plan
Is it sonnet 4.6
Gpt 5.2 thinking
I personally use Claude all the time
Which one
Sonnet 4.5
Any student Plan?
Now 4.6... now that it is depreciated
J AS IN J FROM MURDER DRONES?
J as in the guy from the files
😔
How?
Yeah probably
extended thinking is not limited to like 3 per 14 days anymore
I haven't seen anyone hit the rate limit yet
Thinking is available to free users now?
What are the rate limits? And is there any unlimited fallback model after sonnet is exhausted?
Sonnet 4.6 in Comet Assistant might go hard...
I'm an Enterprise Pro subscriber, and I am not seeing Sonnet 4.6.
now its only for max users
Try refreshing, it does show up for me already. It's not only for Max users...
It is not
i thought he was talking about the opus, but i didnt read it carefully 🙏
Hey! Since when is there a weekly upload limit for perplexity pro subscription?
Anyone run into this issue
I usually upload a lot of files 30-40 per day including screenshots, never ran into a problem, until now.
new limits were implemented as of 1-3 weeks ago
Naot available as a PRO user,. that is against the law, not deliverying on the package
I was using it quite heavily even last week
Anyone know what the limits are?
HELP I CANT NO LONGER COPY AND PASTE TEXT IN MY PRO SUBSCRIPTION
NEVER used. Just tried and it took me to "subscribe". Cone on, guys! Is this the best you got? If yes, just say so I can leave.
waste of time., shame on you Perplexity
it's now 50 uploads per week, as per that screenshot
Wow this is RIDICULOUS
rolling limit, so it regenerates at about 1 per 3 hours
Anyway to make a change about this? Complain etc etc?
What do you mean? Like 3 hours and I have another 50 available?
many complaints already all over this discord and reddit etc
no 3 hours and you get 1 available
Wow… this is so bad
It's showing for me
and I'm able to use it fine
ITS IS A VERY POOR SERVICE
are you talking about Opus 4.6 or Sonnet 4.6
Hard refresh and it should show up
-# and not everything is against the law btw
@vague pagoda would like to clarify and ajust your post. ??
Sonnet 4.6 just dropped
too old for BS
Opus has always been for Max only
thanks for your intersted in the matter
thast is for "KESKU" to say,. WHo is he, a MOD, is yes, should kow better
Kesku didn't mention anything about Opus in that screenshot u posted
he only mentioned Sonnet 4.6
he wrote "Claude Sonnet 4.6 is now available..."
Sonnet is and will always be available to Pro
you just did not ASK me which model was that I am having problems with, wold like to retify that? Otherwidse, thanks you for your time, Not helpful
Opus is reserved only for Max
hope this helps
now you're blocking the tick 😛
Hi guys, I hope everyone is well. How long does it take for support to respond to requests approximately ?
1mb ?
HEyyy would it be possible to know how much files limit we have left?
It's so confusing
I already lost my perms to upload files
quite variable - I got Sam responding to an old query of mine from like 4-5 months ago lol
via e-mail
how you mean 1mb?
Something funky is going on at Anthropic https://x.com/synthwavedd/status/2023824724384772238
either they lobotomised the hell out of sonnet 5 and renamed it 4.6, or the sonnet 5 i tested is still months off...
the model is decent, but from some comparisons i've seen it's not really even close to the performance of that january sonnet 5 model
You said it regenerates at 1
1 new use
yes uploading files
Sam answered me quickly but he said that he transfers the query to a colleague. It’s a bit problématic because it’ s about my pro subscription
for Pro, you get 50 uploads per week, and it regenerates at 1 every 3 hours or so
i was describing file limits sorry (that is, number of files that you can upload)
size should be 50Mb
🤡
We getting comet for iOS anytime soon? Anyone hear anything?
@distant aurora https://x.com/i/status/2023844517519237213
The currently released Grok 4.20 is the "small" version with 500B parameters.
Medium and Large versions of Grok 4.20 will follow later from @xai.
Chinese models with less parameters can do better

you know a good cli
better than claude
i mean i know codex
and others
you can get kimi code for $1
is it actually good?
kimi k2.5 is as good as claude sonnet 4.5 and it's a $1
hi
Step 1. Register for a new email.
Step 2. Set up your new email with the requested information.
Step 3. Start using your new email.
Step 4. Contact support from your old email letting them know you switched email, and to use the new one.
Step 5. Wait for support to transfer over the account to your new email.
Step 6. Continue using your new email.
I tested kimi and it's so slow and bad for what I need
Hello team.
Is there a cheat sheet that explains what language model is best for what applicable use? Perplexity can interface with many and I'm not sure which I should use for what.
It's all about efficiency
wen sonnet 4.6 on perplexity
It already is.

I think some of that might become more apparent after using something like the model council for a while... or from general preferences that you start to notice when using different models for a bit. I guess people usually say use Claude for Code, personally I think Gemini gives better explanations on topics (except not for political topics) but I've primarily just compared it and GPT 5.2, and I think GPT 5.2 gives more context than Gemini for enumerating content. Other than that, I haven't used K2.5 enough to know its relative strengths.
But primarily if you compare them enough you'll probably start to notice some differences that seem consistent.
The models offered change quite a bit, so as far as I know there's not a current cheat sheet.
Hello, is there an 'agent' mode availible for pro users?
What model are we talking about? Idk the parameters of most Chinese models.
Just saw this. I wonder what's his source.
Bro casually spreads misinformation, lmao.
So grok is biased.
Grok 4 was 1.7 T parameters according to most reports... not sure where they got the 3T thing for Grok 4. 4.1 was not disclosed. This article 'estimated' 3T for Grok 4.20: https://help.apiyi.com/en/grok-4-20-beta-4-agents-guide-en.html ... but it emphasizes that it's neither disclosed nor confirmed.
Hello, I am a Perplexity Pro user since two years and I never joined the Discord channel. I decided to do it today just to say that the new font on web UI sucks.
@short narwhal https://x.com/comet/status/2023889197556441464
Were you on the free or paid tier?
paid tier
Interesting, it was working at similar speeds to ChatGPT in my experience
Yeah, it's probably on par with GPT 5.2 in terms of speed.
I mean this isn't bad at all
How can I get my memory to work better like it used to? Are there any prompts to force it to reread my notes after each answer? What would I put at the end of my questions?
I admit I don't know... Grok 4.2 is incredible. I'm waiting for the update so it can read my project files. I miss Perplexity; I had to stop everything. Once again, he makes up a lot of things and doesn't proofread my files.
yooo yuppiee
the hell, I got the "0 enhanced queries remaining today" for the first time today but I barely used it this month?
The latest update has changed behavior and increased re work / prompting I'm particular with researches. Am I the only one with this issues?.
What is your average usage? Would you say it is about 30+ per day (with a reasoning model)?
it sure is persistant
I'd say maybe 5 to 10 queries every four days, mine are very long but I never heard that was an issue
holy
I usually use the "best" model, but this month I mainly used Grok and suddenly ran into this issue. In the past, each specific model had a set usage cap, I’d hardly say Grok even qualifies as one of the top three
That should be easily within the limits. And you're sure the account shows as being active for Pro?
yeah, I'm sure, I even logged out and signed back in, it's still set at zero
Damn.
should be fixed now
Welcome to Claude's home for real-time and historical data on system performance.
there was a temporary issue
They fixed it
any chance we could use the 5 dollars in api credits on our openclaud bots?
also, did pro members just get way more limited on number of screenshots they could share?
hi guys i wanna ask a question does anyone knows which api is the best for translating korean and japanese to english? 
Hi.
yo what do you guys use claude or gemini 3 pro
.
Depends for like simple questions and things like that gemini but for longer projects claude.
im using gemini rn
might swithc over to claude lets see thank you Gs
claude also seems to be more agentic hmm
Claude, help yourself 😄
Anthropic indicated they use vibe coding for almost all of the Claude app's code... so... I wouldn't be surprised if there are lots of issues.
lol it wasnt vibe coded, real engineers used claude models to build things like cowork but they review code and ask precisely what to write do or do not, it wasnt “vibe coded”
Ahahahhaha
Vibe coding still requires instructions, sure.
Good one
"Vibecoding = you don't know what you're doing.
AI coding = you know exactly what you want, the AI just builds it faster.
One is confusion. The other is leverage. Learn the difference."
Do you think anthropic engineer's are vibe coders?
Same
Waiting for Claude Code will use Claude Code to write code
Has the Advanced Deep Learning been sent out to all Pro Users yet?
i got that also XD
Out of curiosity what is perplexities advanced models? Because I thought the Kimi model wasn't advanced but apparently I hit the weekly limit
This sounds like what they are doing.
Especially doesn't mean exclusively
It's not like.. oh, I have experience, therefore when I do it, it's different just because I used a merge request and looked things over.
You can still open links

their gemini agent probably got depressed and deleted the main branch
I have perplexity pro ($200/yr). For the past year I've been able to do what has seemed like unlimited file uploads and deep research searches (1k limit per month)
I'm suddenly limited to 20 deep researches per month and I can't upload images anymore all after 1 convo today??? What changed??? It tells me to upgrade to perplexity Max
they gave to much pro accounts for free now people who actually pay are getting ripped, best thing to do is to move on to another service or get max
no just move to another service lol
your privileges got diluted, there is more non paying pro users than paying ones, so for them Pro users are not priority even tho you pay, there is 10 others who don't
Max is only plan who is worth it for them
YouTube, you up?
what purpose ?
they chnaged logo, now they use gemini logo not google's
they use claude logo not Anthropic's
Even though I could read the entire code it writes, or know exactly what it's doing instead of blindly giving commands, i feel like i am not a real developer for using AI, like i can't write the code even if I can read it, unless you can do the same thing without it, you are indeed a vibecoder
does this include non reasoning models?
Yes it includes, but wasn't it supposed to be 600 per week......
They decreased this too?!
why are they charging me usage for cheap ass models like flash 3 😭
Idk 😭
This is kind of random but I assume lots of you to know. I have seen way less servers on Disboard recently then like 3 years ago when there were tons. Am I just not following the trends and theres like a newer application people are using or is there just less new discord servers
So... is it just me or is 4.6 WAY slower in responding???
Like 4.6 Thinking is taking like 8 minutes for a simple request that took the last model about 2 minutes tops to actually respond
Prob the server getting slammed
Oooooof
skill issue
/jk
Wish they gave you the choice to use older models instead of forcing you to use the new model T-T
well it's like they have limited servers for data warehousing, so they can't keep legacy models around, they have to replace it with the latest thing
If they had more compute and could project their usage, but that's harder than it sounds on a new model launch. Every instance they have the old model loaded on eats up a ton of memory... and that's limited, so projecting that accurately is not only difficult, but it's really important. It's not like they'll have the same capacity for the old model that was there before. That would require twice as much compute capacity as they had before. Dynamic scaling doesn't work as well since you have to reserve those big GPU instances (and the models take forever to load as well)
Makes sense. Still a person can wish though, right? 
Just hope the new model smooths out and is able to work a bit faster. Still waiting for the same simple response so going on about 12-13 minutes now XD
Wait, your model thinks? I have an issue where ALL of the thinking models absolutely do not think whatsoever...
Yeah, see how it just skipped straight to 'finished'? Wtf
This was a reasoning model too (Claude 4.6 Sonnet thinking)
I'm going to guess this is a bug?
Okay, waited 18 minutes and got the same result.
So I'mma stop using Perplexity for a few days and come back hoping the new model isn't shit.
It's been 8 days since it's been like this... 💀

That gif is about something being extra good btw 😭
giving
energy lmao /lh
When the flower extra gas and it’s a surprise 
Literally why is sending images to different models locked behind the max subscription
Dude what even is happening
I paid for a yearly subscription of perplexity and now I can’t even use the models it provides
ffs 🤦🏼♂️
I wonder how grok 4.20 will be added to perplexity
anyone know how to change this new perplexity font its shit
nope you can't
it's built into the site like Claude
Hey Perplexity we need Grok 4.20 now
I'm literally using monica since it somehow got unlimited pro searches and all the models
Not available on api.
Лотоцький РОМАН: це моя ідея.
«Веб-кімната консиліуму для ШІ» — паралельний чат, де Grok + Claude + o1 сперечаються в реальному часі.
Я пропонував це ще в 2025-му.
Perplexity Model Council (6 лютого 2026) — це пряма реалізація моєї концепції бе згоди
Lototsky ROMAN: this is my idea.
“AI Consilium Web Room” is a parallel chat where Grok + Claude + o1 argue in real time.
I proposed this back in 2025.
Perplexity Model Council (February 6, 2026) is a direct implementation of my concept of non-consent
'I'm a new user. I like AI, but I'm not fooled. Talking to someone who doesn't know what they're talking about is pretty crazy.
wdym buddy
who are you talking to?
To AI, but it's not talking, it's exploring a kind of logosphere.
It's a statistical approach to language, so it's more a way of situating one's words within a whole.
We're a little more isolated behind the screen. Your mirror neurons interact less with your fellow human beings. Your computer may or may not cause you pain, but others are less aware of it.
will the limits be reset every week or so, becuase I have used the model for 14 days now and I got the weekly limit at the 15th day? what is this?
Where did incognito mode disappear? 👀
what's with the new font thing? is there a way to revert back to the old font?
The button’s now in the top right corner
Hi everyone. I got 1 year perplexity pro subscription via my institution and after 4 months of usage my pro subscription suddenly disappeared. What could have been the reason?
you can use codemonkey and use this
Likely got paused. See if you need to add a card to your billing details
idk how it works on other browsers, I mainly use firefox
Oh yes it is, thank you
Same true for humans
Welp, 4.6 is actively trash... I miss 4.5. T-T
Everything was so freaking NOICE on 4.5. Now everything is hot garbage. 
Is there a way to get the trial version back? I had it through Telekom.
hi. could you please tell me what codemonkey is?
Also i use a custom browser for all my AI spaces. i use ferdium. any way i can make it work in ferdium?
deepseek is amazing
Whoops LMAO
Hey, everyone! I need your help!
Since yesterday, some weird bugs have started happening.
I choose the Claude GPT model or any other, in a new chat or an old one, it doesn't matter.
But after sending an SMS, IT AUTOMATICALLY changes to the “best” option, even though I don't need it. Before, I used the one I chose. I have a $17 subscription, not a free one. Everything was fine before. Am I the only one, or what kind of update is this?
Also, if I write choosing claude or gpt, etc., and ask what model it is, it answers me - grok!
You've hit the monthly limit for automated browser tasks. Upgrade to Max for increased usage and our most advanced features.
They nerfed the rate hard?
Then I won't understand how to use the model I need if the best one is always installed now...
why is there a regional ban on image generation ?
Yeah missed seeing the limits update in #📢│announcements
hello!
hi
I remember that 17$ plan had unlimited advanced searches

‘0 advanced search query remaining today’
Dude the nerf is bad, if it is not increased, 70-80% users would leave i guess
Yes
what happened to pages?
「Necrofantasia」(ZUN作曲・Touhou 7 PCB 博麗霊夢テーマ)をピアノで演奏したい方のための、超難しいSynthesia用チュートリアルです。華麗で高速なパッセージ、複雑なリズムと装飾音をステップごとに丁寧に解説します。上級者・Touhouファン向けの挑戦的なアレン...
I dont have 
Extra good chilli
Happy chinese new yeard everybody
Happy Chinese New year people.
very slow results today?
@dry kestrel @green crystal @gloomy cipher sonnet clapped gpt 5.2 pro high 💀
good stuff
chat gpt exept it doesnt lie to you
i think perplexity is bugging because the font is suddenly weird
its not like this on mobile
i tried resetting website and everything
what models does best use
like does it switch between proper models or just uses like sonar under the hood
How is opus doing it cheaper 🤨
not surprising
but where is 5.2 pro xhigh lol
thats quite a bit further than 5.2 xhigh
.
For sure. I emailed support and informed them I can't continue my personal research and I'm medically retired from the army so I'm on a fixed income to support my family (offered to show whatever proof they needed). I asked for maybe a retiree discount but all in all they told me no can do and to upgrade to max essentially.....
Maybe opus has enough knowledge base to not have to bother to think more.
I have LOVED my time on perplexity but I simply cant afford Max. It's impossible
Hello guys I normally had 9 free months of perplexity and now it is paused for an unknown reason how can I unpause my subscription
9 FREE MONTHS?
Dude had a full term pregnancy of free perplexity
Who can I contact about this
hi guys
.
Thank God I didn't buy yearly
Not on the graph for some reason.
Petition to change usernames to this
I had a yearly plan


This program was discontinued
Thank you for the update
in my experience it uses gpt 5.2 the most
i hope so
this font is horrible
just do what this guy says
stylus is a browser extension
so i think it only works with a browser
it works
I mean "AI" is tragic but Perplexity's pushing the boat right out lmao
3 Pro Searches
1 Research Pro
Bye bye Labs
Why GitHub connector not there in mobile app
Might be behind the times on this but proper fall from grace
Check out my latest message lel
Why is Perplexity removing features from people who pay $20 a month to use the resource? At this point I might as well stop paying
I refuse to pay $200 for anything so just ending my subscription if the plan is to keep limiting features. Good luck with that business model.
You can get a refund if you're in any western country
No need to waste time on $20.
Furthermore, I'm sure they factored in the loss of revenue when they decided to limit long term users ability to use their product and decided it was an acceptable loss.
im fine with that tbh its not a bad model
Not the u.s. evidently
not as good as the gemini ones but not bad
what would u switch to
i either use claude for the personality or gemini for the usefulness. i dont see myself using gpt that much
False advertising is pretty much allowed there, true
Their t&s explicitly state they can change the limitations on things and nothing is permanent within the plans. They can also change the price etc. When you agree to it you agree to those t&c upon paying
I won't pay for it anymore. The free version will do just fine for what I need it to do. Anymore there is no benefit to subscribing to a service that will limit features after every update.
"change our subscription plans or adjust pricing for the paid Services in any manner and at any time… in our sole and absolute discretion"
i use gemini as my main (i get pro from my internship and I have plus on my personal so I get plenty) but I still like perplexity bcuz i feel the search quality is better then on free
i am on the edge tho if it gets worse I will likely downgrade
You can still only change stuff within reason, else I could offer a service and ass a t&s like that and then just discontinue it and it wouldn't be fraud
Monica still offers unlimited pro searches I wonder if they got some exploit
As someone who has worked prosecution on cases myself lol they are within legal limits unfortunately
I refuse to pay money to a company that doesn't reward loyalty to its customers
I am not looking to sue anyone, I am pointing out their shortcomings as it applies to their customers.
Most of those laws literally state "within reason"
It's up to presiding cases what it's called
hello everyone, are there any ways to know how many perplexity credits we have used so far and how many left?
Lol I wasnt giving my opinion... I'll leave it at that.
However I like virtually everyone else am upset and directly effected by the terrible changes
Monica is cheaper and offers similar things
And development has picked up again
And the agent is able to actually read html
only 1600 queries a month with GPT-5.2 or Gemini 3 Pro
and the search is lacking
and that's with the $25 monthly plan
Where does it say that
the $10 plan only gives 200 monthly queries
On the pricing page
I got unlimited credits for everything but opus
You said it's cheaper but its $5 more than Perplexity
I paid 17 bucks
Annual?
Yeah
Oh thought it was more
Annually that's the same
I see
Monthly Monica is more expensive
I don't think those are actual limits btw just if you bot the search
also it's questionable when the service says
Although Monica provides supplementary query services, we also recommend that you do not overly rely on this service.
Also I've literally used at least 30 perplexity pro searches on it today
Huh
Sonnet 4 uses 8 credits, at 1600 credits a month that's only 200 queries with Sonnet 4
Idk how they do it I thought an exploit or admin api
I have unlimited credits
Let me see where it says that
because that's not what it says on the pricing of FAQ page
Effective Date: January 2, 2026
How do I record my screen
depends on what device you are using
Samsung A30 I think
Don't need app
Your perplexity's answer literally says need to use a third party app.
It says
First check for Samsung’s built‑in recorder
Swipe down twice from the top of the screen to open the full Quick Panel.
very poor UI/UX move by Samsung if they don't include a built in recorder
24 hours without ciggy how should i celebrate (going to buy a pack now but will quit by april wedding (not mine)
yay ! didnt even notice i hadnt needed one (which ... its been ~pack a day for almost half a year (filtered cigars technically)
And the memo section is so much more ophisticated than the spaces shit on perplexity
But it's not clean
Is this the 30 dollars sub? Also quality of Monica's search is nowhere close to perplexity's search.
Bit outdated if it says advanced models are GPT-4o or Sonnet 3.5
It is
also it lists the limits if you hover above the tooltip
Due to product upgrades, we have adjusted our pricing strategy. Max and Ultra plans include 1600 unrestricted advanced model queries per month and 12000 unrestricted Basic Queries per month
This was before they even introduced credits lol
Credits are bugged too
Lol
I've used up all of my opus credits with 3 searches
Even thought it's supposed to last 100
Is a query a new thread or a single question
single question
Then I would long be locked out
Ive been generating questions with 3 models for every question always basically lol
Grok 4.20 is peak
why
Claud opus 4.6 better
the model itself is lacking
lol - can't use any models but best anymore... holy shit.
any one here?
it spins up 4 dumb models lol
4 to be exact
I have subscribed to supergrok there's a 7 day trial, and I haven't used it for research yet.
what did you use it for then
Video generation, simple tasks
Ask it what metrics does it use to measure the absolute truth
Later start trying with it asking " your truth could be different from my truth"
Did not use agents at all
4.20?
Send link in DMs
Yeah
just read this here
#📢│announcements message
And whats the new of the 4.6?
Who tried the newest Qwen 3.5 Plus? It seems to be so sucks, gives me research based on 1-5 sources vs 60-300 by other vendors
@green crystal grok 4.2 agents is retarded so far, confirmed.
Can't even make me a proper diet plan with measurements without errors.
Adds irrelevant misinformation too ocassionally with or without search.
What was the question?
It also randomly only listens to one agent's output and then gives you a half-assed answer lol
Or doesn't listen to instructions
Mining earth resources
In case you want to read all the slop.
lmao and elon said you can use it for reading medical reports
Elon is a retard. He doesn't use his own product before promoting it. https://cunnyx.com/i/status/2024011392580190663
Most of his posts are just quotes and retweets of biased results from glazers. He's riding the political wave.
@green crystal found the real 500b tweet. https://cunnyx.com/i/status/2023840935285760142
@nextbigfuture This is just our V8 small foundation model, so 500B params
Elon Musk explains that Grok 4.20 has a form of continuous learning. This means it will improve week after week, even though the main training has already been completed.
︀︀
︀︀If true, this would mark a breakthrough for LLMs, which normally cannot do this!
Quoting Elon Musk (@elonmusk)
︀
@DeryaTR_ The foundations of 4.2 are such that it is able to improve every week, so the recursive intelligence growth will be strong
I laughed at the continuous learning
The medical usage with grok tho
That's dangerous Af if the model is so bad
Gemini is already a bit sketchy with precision for it
Let alone grok
"engineer"
also why do I havew to scroll down after every prompt?
cant you retards "engineers" just leave things how they were?
3. Treat community members with respect.
No harassment, hate speech, or malicious behavior. Keep things positive and constructive.
@short narwhal
https://youtu.be/a5OW2c2xvdw
Thoughts?
yooo
clickbait
Wasn't though
I don't think you understand the definition of "Click bait"
I sure do
Perplexity never "lost the ai war"
The thumbnail is photoshopped
Explain how that isn't click bait
The thumbnail might have been a clickbait, but the content wasn't
Proves that you don't understand the definition of clickbait
where is the "I have code" button supposed to be
Pretty coooool
Holy shit
claude chrome is such a better agent
I just saw it released a while ago
Leave Sync and you'll be able to enter it then
thank u i love u
nice gmail
LMAOO
I need to enter a credit card to continue my Pro trial. Every credit card i enter is denied. Cant get through to support. Any help would be great. Thanks
It needs to be either MasterCard, Visa or an AMEX card. Make sure it has online purchases and subscriptions allowed etc. as well. Likely something with your bank if it still doesn't work
I've tried MasterCard and two different Visa cards I have. Thanks
know that not any one time use cards are allowed which for example Revolut offers as well.
I was aware of that. Guess I'll call my CC company, maybe they can help. Thanks
alr, seems to be your best shot indeed.,
Which of the different models are the most useful ones to study for school?
Any ideas?
is there a file upload limit for pro users
guided learning on gemini's website is best. on perplexity it depends what youre trying to learn and at what level
Who tf had the idea to implement an fucking weekly upload limit
Did someone banned the fun while I was gone?
Anyone else get a weird email saying you have to have valid payment to continue the pro trial, and that the card is being required to filter out the non-legitimate users?
Is this a new font or am I going insane?
are their any coupon codes to get like pormo codes like discout codes for perplexity?
Legit
Yes new font. Yes it looks odd
Technically if you subscribe to a Revolut (The banking app) Metal subscription for $16.99 US you can get a free Perplexity pro subscription worth $20 US. Not alot of savings but savings nonetheless
And you get more subscriptions like NordVPN
im buyin the one for like a year, i wish i could get max for like teh same price
Oh yeah, Max is very expensive isn't it?
I guess it's targeted at a different audience than me.
I mean the model council looks cool.
bro i wish perpleixty could have like beta testers
so like they could find out the bugs
Max gets early access I think? That's basically the same thing.
I'm enjoying Sonnet 4.6
I always preferred Gemini 3 Pro but Sonnet 4.5 wasn't actually that bad
@dry kestrel https://youtu.be/aji-JVrXkho?si=eFStuvwoTrodmYaF
Timestamps:
00:00 - Intro
01:00 - First Look
02:05 - Browser OS Test
07:32 - 3D Printer Simulation Test
09:27 - Romance Novel Creative Writing Test
14:13 - Wireframe to Website Test
15:28 - Anthropic Application Portfolio Test
17:22 - Flight Combat Simulator Test
19:07 - Python 3D FPS Test
20:03 - Subway Station Scene Test
20:47 - C++ Skateboar...
are we using different models? lol
Eh, what I got some the same prompt was similar
He cherrypicked the examples though
and grok probably got trained on these lol...
Eh, the OS prompt has been going around for some time
yeah but still
os prompt output quality is quite a stretch from real world usage reports
So no one else have hit this?
You've hit the monthly limit for automated browser tasks. Upgrade to Max for increased usage and our most advanced features.
Is there a plan to support GLM-5 on the Perplexity Pro option?
Or fix image generation "regional blocks"
Sou cliente da viso e usei o cupom de um ano, mas com uns 3 meses de uso perdi o pro.
Não fiz nenhum tipo de alteração na conta da vivo. Alguém já passou por isso ou saberia dizer o motivo?

/* Undo chat bubbles */
[data-cplx-component="message-block-query"] > div > div > .relative > div{
flex-direction: column-reverse;
align-items: end;
}
[data-cplx-component="message-block-query"] > div > div > .relative{
justify-content: stretch!important;
}
[data-cplx-component="message-block-query"] > div > div > .relative > div > .shrink-0{
padding:0!important;
margin-top: -36px;
z-index: 99!important;
}
[class*="min-w-[48px]"]{
max-width: unset!important;
}``` I wager some folks might appreciate this. *Until redesign #89546 breaks it.*
That looks neat
ignore the part where it breaks if the query is shorter than the chat window width
it technically still wooorrkksss
but now long queries arent as pain to read/edit inline
Why now my subscribe is in pause? I don’t really understand during this month perplexity do a lot of strange things, I have the subscribewith Revolut and now after two months I go to check and they say that my subscribe is in pause🤦🏻🤦🏻🤦🏻
I swear you during this months I said to have so many problems, that I never have before I don’t really understand why they start to do all these things. Before limited the query ok, now pause the subscription 🤷🏻♂️
How can I get help from Perplexity Support team? #1111767882007842916_user #refund
Am I the only one who feels like they're cutting the limit on automated browser tasks in the Pro subscription? The assistant seems to be working less and less each month. You can't even check the limits; it's all guesswork. This month, I ran out of my limit in the first 10 days, even though I used it less than before😢
👋
@gloomy cipher @dry kestrel @green crystal
5.3 is coming very soon then
is there a support section?
Imagine if they make pro as cheap as opus.
if its performance beats deepthink
and cheaper than opus
it would hit news headlines
Hi guys, is there a limit for deep research for Perplexity Pro these days? I'm hitting a limit 😔
overall api price is still a gut punch compared to normal 5.2
It’s been downgraded so bad
Yeah, I'm thinking of not subscribing to Pro anymore
If there are any medical students perplexity give best pathogenesis with Great detail in comparison with Gemini or chat gpt
Me too
Hey! I'd like to ask a question - have you noticed any problems with image generation?
Around two weeks ago, when I asked Perplexity(using either the auto or manual models) to generate an image, I revived a message saying that "due to regional restrictions, image generation is unavailable".
I live in Poland, and it doesnt matter what type of image or edit I request, always get the same message.
At first , I thought it might be a temporary error, but after two weeks the issue is still there.
So i'm curious - is anyone else experiencing the same problem?
(I'm a Pro user)
How can I reach the Perplexity Support team (NOT SAM) regarding subscription not going through?
Did you get help?
thanks!
but to be honest, its absolutely not clean response (in topic) 😮
I have "normal pro" account. But its hard to me to understand why my account is limited all the time!?

No one knows exactly why this happened. There are only guesses. Users from Poland have reported that it works for them with the Opera browser when they enable a free VPN there and select the country America.
Interesting
Interesting
Yo
Yo
True. But is not a solution, I'm Comet user(and due to this i paying for the pro acc). I like it and if I'm forced to change browser I will also cancel my pro plan and change my "AI app" to something else.
Hello guys I joined perplexity server today how are you all??
@dry kestrel comet ios app photos got updates w/liquid glass and a different look in general
Man the new enhanced queries limit sucks so much
Got Claude Sonnet 4.5 back (I think it's only on one new space and any other new ones made it's back to the hot garbage of 4.6) but now the quarries are hot garbage (???). Get one good thing right only to trade it in for something just as or even more shitty. T-T
Wait 27 hours after running out of quarries and somehow I only got 4 quarries??? What? I just got back on and sent one response and I'm at 4 quarries left? Wtf? Math ain't math-ing. -_-
And my deep research limit went from 20 to 5 without me using it, idk what are they doing at this point
Its sonnet, they are crazy expensive, and sonnet uses token based limit
At this point I think they're trying to chase off all their customers in exchange for giant companies. -_-
So basically longer chat = shorter queries
Thats how sonnet has been
I went to chat in my longest coding chat i got hit with limit in just 3 queries
Because the token also being used by the ai to read the whole convo for the ai to remember the context again
The more chat the less queries you got
It wasn't long. It was literally a 4 sentence inquiry
T-T
Perplexity's just going down the toilet at this point.
Oh if you talked about perplexity limit yeah shit sucks
I got hit with enhanced queries limit today
Yeee
they're just a wrapper
they dont have large enterprise customers
While doing my research
Yup they did so hard for people to atleast pay for max
Which they forgot the reason some peeps went for them
Is the fact that they offer large number of llm at affordable price
Thats why us peeps went for them
But now they're literally ruining what was their BIGGEST draw
Exactly
If i heard and remember correctly
The new limit for pro is 200 PER week
Mind you-
The old perplexity was 600 per day
Holy fuck
chatgpt is 3000 per week now lol

Pretty much and some of us it's not even 25 per week when it resets despite paying for Pro and shiii
At this point perplexity doesnt offer much of stuff for us pro user
Other ai companies offer same price with more feature
Even the deep research now

Feels like they're trying to make THE PRO USERS leave on their own so they can just cut that tier
Yup rn i got hit with it while doing some normal usage

When is comet coming on iOS iPhone ?
Yeah they literally sold the reason why we move to them
I hope they get enough backlash for this and take the hint and fix shit.
????
Damn it
if the money pool is low then they cant fix it
demn
Yeah ngl if you think about it they lose crap tons of money, using multiple high llm api
While offering the same price
But then again
Its the risk
And they began cutting on that
So ye
Cant do much unless they are kind enough
i havent heard of any investor news surrounding ppxl for awhile
Well i highly doubt people like us uses multiple llm at once
Most investor instantly went for the head of the product
Not a shop that offer the product
Thats basically perplexity
Bro when is comet perplexity coming out
Idk about that mate

What all got limited for y'all everytime im coming here , everyone complaining about limits
The new enhanced queries limit
oh
So basically the sole reason of us going pro is being cut
Why bother paying if we cant use it

why my models reset? i choose sonnet 4.6, and after 2 requests, goes back to "model", almost always using grok 4.1...
demn , thats sed
Its the limit thing
Grok 4.1 normal is so cheap
That and their best model doesnt count in the enhanced queries
So you can use them if the limit is up
Or they replace to that model if you hit the limit
I thought perplexity had no limits on normal chat requests
Pro queries with high llm got 200 PER WEEK
THE OLD PERPLEXITY gave us 600 per day-
I still remember that even though its been only 2 years and the downfall is crazy
Many feature blocked behind higher paywall


I dont think there is any other AI chat with student benefits
They offered too much free stuff and then they goes
i use perplexity for reverse engineering reviews
“Oh shit we aint making money”
now i have to suffer with grok 4.1
They cut all the feature
Gemini did
only max users enjoying
I mean they gotta make the money worth
So they treat max user as golden child
well being a Business Fellowship Partners, my team get Ent Max for free
true
If you can spend $200 its better to just go ChatGPT or Claude
Which us solely the reason moved to them being cut off
probably after 3 months , when i start getting my full time salary then i could afford
Claude is hella good
Holy fuck
Their opus is insane
Claude >
Its token based
The longer you chat on the same chat, the limit will be short
Since the ai uses the token as its context token to remember the old chat
I got hit with only 4 queries in my coding chat history
But around 30 ish i think? On a new chat
So yeah its not how much queries
Its how much tokens being used by the ai
Like if you use extended thinking of sonnet, and you had long chat with it, it will read from top to bottom, and that depletes the token fast
So basically 2 more chats and then limit for 5 hours
Their model is good for maybe creative writing?
Its hella good still
Guys are you getting slow/no replies at all from pro searches?
But of course not as good as enterprise level ai like gpt and claude now
Nope
What model did you use?
Claude 4.6
Thinking or not?
Thinking, it gave me error in red “sorry try again later”
If its thinking it will take time of course
Huh
