💬│general

[rocky yarrow]

Where's the team at perplexity?

Are they going to address this publicly?

I'm mostly just super sad because I've gotten so used to using it and I don't want to unsubscribe to Pro but feel like I don't have a choice now

[naive sage]

They didn't post in #📢│announcements about the new limits yet, but maybe.

[rocky yarrow]

I don't understand how they can just decrease my ability to use the app by like 95% and not announce it

I know things degrade but it almost seems like a mistake or something to do it that much.

[vagrant scarab]

Those who had scheduled tasks and were using advanced automatic search were severely disadvantaged; in addition to not receiving any warnings, the speed decreased ridiculously.

[rocky yarrow]

Kind of feels like all the AI companies are gonna just collapse under cost pressure soon.

[dry kestrel]

They upgraded the capability of Deep Research with a better sandbox infrastructure and Anthropic models powering it

https://x.com/perplexity_ai/status/2019126571521761450

Perplexity (@perplexity_ai)

We've upgraded Deep Research in Perplexity.

Perplexity Deep Research achieves state-of-the-art performance on leading external benchmarks, outperforming other deep research tools on accuracy and reliability.

Available now for Max users. Rolling out to Pro in the coming days.

X · Feb 4, 2026, 7:11 PM

[vagrant scarab]

In my opinion, the research has gotten worse; I complained about this in the latest updates.

[dry kestrel]

Interesting, I found that the search was more expansive than the previous one, better than other competitors at least like ChatGPT

[vagrant scarab]

I don't use GPT chat; I'm currently subscribing to Gemini PRO, which I find better, and the usage limit is much higher.

In addition, it generates an audio response for convenience.

[dry kestrel]

Gemini trains and views on your conversations though, and the Deep Research takes so much longer

[vagrant scarab]

Yes, because it uses far more sources than Perplexity.

[naive sage]

Yeah, I know it's still pretty often in some other providers where I go to the citation and can't find what it's referencing... I see that it referenced it, but the fact that it referenced doesn't appear on the site. With Perplexity, it is more accurate in that sense, it's much more often that I follow through and see the reference like I would expect. In my use cases, the reply is important but following through to actual documentation is very important too. I can use the latter for reference in design docs and tickets, if it contains the information being documented, but I can't use the LLM reply as a reference lol.

[dry kestrel]

more sources ≠ better quality answer

[dry kestrel]

I agree that it would help with clearer documentation

They do clarify a bit here why limits were reduced
https://www.perplexity.ai/help-center/en/articles/13600190-what-s-new-in-advanced-deep-research

Usage limits have been adjusted to allocate more computing power per session, so you get deeper insights and higher-quality results.

[dull path]

Metric,Managed Service ($50/mo),Local Compute ($600 GPU)
Logic State,Opaque (Black Box),Transparent (Weight Inspection)
Availability,Server-Dependent,Always Online
Censorship,High (Provider Policy),None (User Control)
File I/O,Upload Limits / Retention Risk,Infinite Local Context / Privacy

[shy blade]

anyone else getting a error where screenshoting sometimes doesnt work or is that just on my end?

[vagrant scarab]

But I had no complaints about Perplexity until these latest updates. I noticed that the search function has become poor; results started appearing differently than before the update, especially in chats I typically use when they get full and I have to redo them.

Furthermore, the search style limits have worsened across the board, and PRO users have only suffered losses because of this. Before it was wonderful, now it's a disappointment to maintain the plan.

[dull path]

Final Determination:
Paying $50/month is economically inefficient. By redirecting the 1-year equivalent ($600) to a Radeon RX 9070 (16GB) or RTX 5070, you acquire a permanent asset. This asset eliminates "Stall," allowing your usage frequency to increase well beyond what a $50/mo tier would permit. The efficiency gains from this unbuffered throughput effectively "pay back" the hardware cost within the first operational quarter.

[vagrant scarab]

I don't remember saying that it was =

but I still think it's better

[dry kestrel]

more sources doesn't meen you get a better answer

I've had gemini spend 30+ minutes pulling outdated sources when searching for model pricing information

and it ended up not using a ton of sources it did search

[vagrant scarab]

I didn't see it, because what I found most here was complaints about these limitations; I didn't see anyone as satisfied as you are.

[dull path]

even github copilots orhcestation/pluggins pull bad code

it can pull demo thinking its real

[dry kestrel]

Well, that's why I wanted to post it here. I agree that Perplexity should update their documentation to be clearer regarding these new changes.

[vagrant scarab]

It's working for my needs, so I don't see a problem.

And Gemini's answers are more detailed than Perplexity's. The point is, I'm not complaining about the response at any point, and I don't even know why you're bringing that up, considering I complained about the limit in the first place. Do you work in support?

[dry kestrel]

People have different needs, if you found a product that you are satisfied with, use that instead.

[dry kestrel]

I don't work at Perplexity

[vagrant scarab]

If I'm paying and receiving support, do I have the right to complain or not?

[naive sage]

While I agree with this in principal, especially if local models get better... the models you can run on those cards aren't state-of-the-art models like any of the ones that Perplexity hosts. You can get something 'decent' for a good number of queries and use cases, but I don't think LM Studio supports web searching well, and the accuracy/quality of the responses from the SOTA models will continue to be better. A local LLM might be able to handle the 80% use case for basic questions, though... which can offset usage limits from other providers. But that's the same way that the non-reasoning models ('Best'/'Auto' mode) tend to offset usage limits for the more basic questions. So at the end of the day, I'm not convinced it's better to run locally, especially if Perplexity's search enrichment/references are important... maybe local search enrichment will improve at some point, though.

[vagrant scarab]

So I'm forced to accept any changes from the company and have to keep quiet, is that what you meant?

[dry kestrel]

No?

[naive sage]

Now for privacy, it's hands-down, local is better... but in general, ehh.

[dull path]

the models are not what makes them "state of the art" its the orchestration

[dry kestrel]

Would be great if you posted or upvoted other posts regarding this issue in #1111765634267742408

[dull path]

the model can get fucked, the agent is what counts for searches, only thing model does there is make search script and it takes several terms to do it

[old summit]

If you paid for a year subscription you have to accept these limitations although you used only 1 day of your subscription let's say?

[dull path]

turns^

[vagrant scarab]

I've already done that, I wasn't going to stop talking about it.

[dull path]

in a month nobody will js

[naive sage]

Perplexity does allow you to request a refund, if you're not satisfied, contact [e-mail].

[dry kestrel]

https://www.perplexity.ai/help-center/en/articles/10354288-refunds

[vagrant scarab]

I don't know how it works. I paid monthly, but what about those who paid annually and signed up with the old limit, and it was changed? Would that be considered a problem?

[old summit]

Exactly

[vagrant scarab]

Because I saw complaints on Instagram and X from people who paid for the annual subscription about the limits.

[old summit]

Smart move level 5000

From the company

Even in my Minecraft server I used to run as a kid I had more respect for people who paid for ranks and advantages

[naive sage]

Generally any substantive change to the terms, under contract law, allow you to be able to request a refund. I would think an arbitration judge would consider drastic changes to usage limits as a qualifying case (even if the original terms says those limits are subject to change, if the change was drastic). But... if Perplexity's policy is only allowing annual subscriptions refunds for 72 hours after purchase and doesn't allow partial/prorated refunds later on in the term due to changes, then you'd have to go to an arbitrator... and that is a big hassle. Though you could probably try to convince a third party payor that the terms changed and what you bought wasn't as advertised, which can let you get a refund for up to 60 days.

[vagrant scarab]

Here in my country, this kind of thing isn't legal, and it can lead to legal problems.

Especially if it's an annual plan.

[old summit]

Of course, it's not legal in Romania, where I am from either.

Apparently perplexity never heard about consumer rights and company reputation

😂

[pearl nexus]

I hope that they will at least revert some of this extreme throttling, giving us at least 10 Deep Research a day and maybe like 50 Pro Queries a day... that would seem more reasonable taking computing costs, but also competitor offers into consideration. 20 Deep Researches a month and 200 queries a week is criminal. Or, they could at least offer separate Deep Research tiers - the old and new ones separately.

[dry kestrel]

The updated limits show up to 200 Pro searches a week for context https://www.perplexity.ai/enterprise/pricing

[vagrant scarab]

I'm from Brazil, and here they're even overly picky about this issue; companies are very careful with plan changes and notify users in advance.

[pearl nexus]

Yea mb, I meant 50 pro queries a day, so 350 a week.

[dry kestrel]

All good, I just wanted to put that page out there since it's harder to find

[vagrant scarab]

The Deep Research queries I received here came up with less than 50, I barely used them this month and I'm at 18. Furthermore, the company didn't even notify the user by email or anything like that. Usually, other companies, even when the plan is free, notify the user by email.

That's true even when the plan is free.

[pearl nexus]

At least you get 50 😂 we get 20

[dry kestrel]

Hopefully this changes or is made clearer in the near future

[vagrant scarab]

HAAHAHAHHA

[pearl nexus]

Yeah, the lack of prior announcement and communication from the company is just criminal

[vagrant scarab]

Thinking about it that way, I think I came out ahead.

[old summit]

Brazilian Privileges

[vagrant scarab]

It must be an effect of communism.

[pearl nexus]

Like, I understand they were losing a lot on power users, but this much abbruptness in changing limits I think will do more harm than good to Perplexity, and it's sad to see as I really like(d) it

At least I hope a good competitor with the same business model comes out now 😂 maybe we'll get something out of China

[vagrant scarab]

The limits, and mind you, I barely used them this month, but since they didn't warn me about the limit decreasing drastically, I had scheduled tasks, so you can imagine what happened.

[old summit]

Yeap

[sonic owl]

I haven't even used perplexity today. But when I see it says I cannot use research for today. Pls help.

[pearl nexus]

Yup, I used it primarily for academic and business research, gg to that now

[old summit]

They also capped the amount of files you can attach

[vagrant scarab]

I'm thinking of going back to Deepseek or some other Chinese service, which at least is free and unlimited there.

[pearl nexus]

Yeah me 2 to be honest 🤣

[vagrant scarab]

But this is the first time I've had a problem with them. Before this, honestly, I had nothing to complain about the company; I really liked them and recommended them to all my friends. This is the first time I've been disappointed with them.

If it were just me complaining, that would be fine, but then go on social media and see the people on the Pro plan commenting on it.

[old summit]

I guess part of their decision is because Amazon AWS Cloud Service raised the prices of their services by 15-20% this month

But still from 600 to 20 queries is huge difference

[naive sage]

Ugh, I didn't hear that yet.

[old summit]

😂

[pearl nexus]

Me too, I really gave a lot of free marketing 😂 I think I reccomended it to 100+ people, for real

[naive sage]

Oh OK, it's primarily for GPUs and EC2 Compute

[old summit]

I think it impacts Perplexity because they use AWS

[naive sage]

Yeah, I don't know what Perplexity uses, I was just worried it would impact other stuff too.

[pearl nexus]

Now, from some limited looking into this problem of alternatives, I've found Felo AI and Genspark to have sort of similar interfaces...

[tame current]

just accept that the usage limits were unsustainable and they were eating the cost

[pearl nexus]

Yeah, but AWS increased prices 20%, and Perplexity reduced limits over 100 times overnight, no notice, and that's quite unacceptable for a paid service tbh

[old summit]

I agree, but it's too drastic still, a price going up 20% and you reduce over 100 times overnight

[vagrant scarab]

It could also be that I didn't use up the entire limit in one day because before I renewed Deep Research daily, and I hardly used the lab.

However, before it was 300 or 600 daily limits, I don't remember exactly, but I have an old screenshot. Suddenly my limits disappeared, I was even scared because I went a long time without using them in January because I was on vacation, and I only noticed it today.

[pearl nexus]

Ik that, but the cut is too abrupt, too big, and no notice for it

[vagrant scarab]

Look, the government raised taxes a lot, but they gave warnings beforehand.

[pearl nexus]

I calculated that they were losing up to 200 dollars per month per pro user with moderate to heavy usage :))

[old summit]

Maybe in Brazil, here they raise them month to month too

[pearl nexus]

It was 600 daily, which was quite absurd to be honest 😂

[vagrant scarab]

Antigo limit, still bem that salvei provas

[pearl nexus]

But it is more absurd to cut from 600 a day to 20 a month 🤣

[iron shuttle]

how to track usage again please?

[pearl nexus]

For regular pro queries, you have no way to do it, for the rest of them, remaining queries just show up in the interface selector

[iron shuttle]

there is a link last night

[vagrant scarab]

For you to see that I am not claiming anything haha

[dry kestrel]

https://www.perplexity.ai/rest/rate-limit/all

[pearl nexus]

Me and a friend from the same region have different limits on the same plan - I have 20 deep research a month and he has 20 a day. Does anyone know why that might be?

[vagrant scarab]

Yes, but even so, it used to be 600 daily tasks, now it's 250 monthly, and without prior notice, with scheduled tasks. I don't consider myself a coward.

[iron shuttle]

how do you know if its daily?

[vagrant scarab]

Because I checked the renewal every day, I think I have more screenshots saved besides those from consecutive days.

And it speaks to today, not weekly or monthly.

I put it in English, now it's better, I had forgotten and sent it in my language, sorry.

[coarse girder]

is deep research now a paid feature?

[shrewd steppe]

Where can I contact support guys and who is moderator here

[dry kestrel]

[e-mail]

[shrewd steppe]

Oh I tried there an ai bot same replies all the time

I want to talk to human support

Where to contact to human support guys

@vague pagoda sir I need help I tried emailing but ai response all the time it's been 5 days my payout is stuck on pending my leads are processed but payment awaited from program in pending I asked the dub support they told me to ask perplexity and when I ask through email ai bot reply to check can u please tell me how much more time it takes to move from pending like it said payment awaited from program so when will next batch will be released thanks

[eternal pike]

The only person who replied to my Twitter comments asking about this was actually an investor, no one from Perplexity. His comment was good, people using deep research poorly with no followup. So they felt like most of the Deep Researches were being wasted while others abused it. So if an investor is the only one who responds you know it had to come up in the meetings. :0)

[naive sage]

That is good information, thank you for sharing.

[whole sundial]

Idk how my pro version got removed

[subtle pier]

Hey, can someone help me, i'm straggling with an issue.

[split dome]

hey we need GPT 5.3

[subtle pier]

The border option and the copy option disappeared when I want to copy something for vps.

Someone could help me, please, what is wrong with.

[eternal pike]

I also spent today refining my prompts for Pro search (since it has been updated too) and it now gives me the same quality of response as my Research calls. I just had to work on improving my prompts.

[lusty coyote]

model council on freaking max plan yet again

its always the rich peoples who enjoy these stuffs

[abstract plank]

where is deep research?

[gusty forum]

in the plus sign

[winged plinth]

Can anyone tell me why I am seeing this when try to upload file here

[hybrid grotto]

Basically means you hit the limit

For file upload

[winged plinth]

But for pro user is there any limit for file upload?

And what is the limit policy like weekly or monthly

[hybrid grotto]

Yeah there is a limit and idk the exact number, I don't think it's even documented

[wicked berry]

#1463724190191321109 Weekly

[gusty forum]

they took away your pro

[wicked berry]

50 is limit

[gusty forum]

the limit was 50 per chat

[hybrid grotto]

Hmm

[austere monolith]

[winged plinth]

Then what is the use of pro if still there limit on file upload and only 50 per week

[coarse girder]

lmao its quite easy. AI isnt as profitable as it was thought to be

so they charging u for everything now

[winged plinth]

I agree with you.

[gray osprey]

[near chasm]

i miss research - im getting this now also

[pulsar nacelle]

There is zero impetus for democratization in all of the big LLM giants including Anthropic (there was nice discussion in the first talk here: https://www.youtube.com/live/HOzc4MaRqI0?si=YQgpuNiMHI_DgbDU)

[solar venture]

Honestly people will just go to the Chinese models. If they don't watch out.

[old bone]

Hello everyone ,

I'm Nameesh
our tech community is looking out for event sponsers does anyone know how can i contact server's admin regarding this

[nimble quiver]

Well, in the morning it said I had 1 Deep Research request for today... And now it says I have only 1 left till the end of the month as a Pro user. I want my money back 🤬

[coarse girder]

https://giphy.com/gifs/cortisol-goatis-spike-2yKdOu82dVygJ7S4N5

@NETANYAHU GIVE ME BACK MY DEEP RESEARCH CREDITS

[next dune]

Model Council greatly reminded me of the Pewdiepie video

[rocky yarrow]

Is perplexity fixed yet.

[ancient inlet]

🙂

[glossy sequoia]

Used to be able to do research for every prompt and now it's like barely any

[hard turret]

Yeah, its a pretty drastic cut and without warning. I understand that if they're losing money its not sustainable, but at the same time its not really worth the price without the research mode.

[novel arrow]

i promise you that council isn't that great, it's useful but not a breakthrough, you can open 3 tabs ask 3 models and then ask another model to compare answer it's same thing, i tried coucil and research is still way better+ council answers are too slow because all models need to finish. Also it's hella expensive for people who got pro for free or got it using some promo code

[glad blaze]

how do i even tell i got the new research, or how many research credits i have left

[short narwhal]

It should say “advanced” next to research once available

(Supposedly)

[quartz wave]

Oh this is fun. I get to take legal action against Perplexity now. I mean it'll cost me a lot more money than it saves, which is what they rely on when illegally denying refunds. But it's still fun.

[limber kraken]

um

suddenly my pro is gone

but the deadline is august 2026

[sly badge]

Hi

[short narwhal]

Add a card to your billing details and it should be back

[novel arrow]

should look like this

[rain barn]

hey um i was wondering why the fucking hell yall keep lowering our pro limits? i never had reached any limit while using perplexity for almost a year before now, i didnt change anything in how i use it, and now i reached upload limit? what the actual fuck? didnt it say "unlimited uploads" before? thats basically the main reason i had the sub

[glad blaze]

and how do i tell how much research credits i have left

[rain barn]

nah fuck yall perplexity, paid for a yearly sub and now this service is useless. nobody is paying $200 a month for your service, there are way better alternatives for that price. perplexity was ok for $20, but now it seems yall are downgrading the ai models themselves, because i started getting way worse answers a couple of weeks ago AND yall limiting us WHEN WE ARE PAYING YOU. never gonna use ts again, good job perplexity👍

[austere monolith]

[wide veldt]

cool

[vagrant scarab]

Welcome to the world of PRO users who are missing out on their benefits.

[quartz wave]

99.89% reduction to Research and there are still people like "It's not that big of a deal"

[vagrant scarab]

Old limit before this awful update.

[cyan fractal]

Does pro include unlimited use

[quartz wave]

We really need to get the term "Perpocalypse" going. For awareness. Certainly not that I just wanna start a thing.

[quartz wave]

No. But it was kinda close before. Now it gets you almost nothing

[vagrant scarab]

Unless the person is one of their employees, look at the comments on social media; what I see most are complaints from PRO users, who apparently make up the majority of subscriptions.

[cyan fractal]

Damn. Luckily i got pro from my revolut

[quartz wave]

Nah I have seen people going "I have pro and I don't use Research or Comet so I don't mind"

[vagrant scarab]

So what does someone use the PRO version for? It's better to use the free version; paying for it doesn't even make sense.

[quartz wave]

We went from 600 research uses a day to 20 a month in just a couple of weeks. 99.88% drop

[vagrant scarab]

I loved this company until someone decided to do this.

[quartz wave]

Less limits on the search. They said search was usually better than research. Which is not my experience with it at all tbh.

[vagrant scarab]

What's worse is that they thought nobody would notice this mess.

[quartz wave]

Oh yeah. They intentionally hid their changes. And then denied me a refund so I am now taking legal action. I will actuallh lose more money this way but consumer law takes priority over their weak ToS

[vagrant scarab]

I disagree. I almost never used the search function; I only use advanced search. I generally used search for silly, uncomplicated things, but now that I use it for work, it's always advanced search. However, this has completely harmed me because I lost what I used most.

And the comments from PRO users on their social media prove that this is not true.

[quartz wave]

Even if I did not use Research I would still cancel. They do not care about us at all.

Plus it still nerfs Comet which was my main browser

[vagrant scarab]

Here, consumer law is actually pretty good; they usually give a good refund if you win. Apple itself is paying several users here for illegal actions.

[woeful flare]

Guys, I can't upload pics in perplexity anymore

what's the problem

[quartz wave]

Corporate greed

[woeful flare]

It's asking me to improve my subscription

[woeful flare]

https://tenor.com/view/comaru-gif-5015684624479161574

[vagrant scarab]

I've already uninstalled Comet; I won't be using it anymore because I plan to stop as soon as my plan expires.

[vagrant scarab]

There are several

[vagrant scarab]

News

[woeful flare]

where is it

[quartz wave]

Perplexity be like

https://tenor.com/view/pokimane-broke-boy-gif-1631750273434798155

[woeful flare]

Will it change?

[vagrant scarab]

Apparently, the PRO user no longer exists for them.

[vagrant scarab]

Judging by the company's stance and comments on social media... No.

[quartz wave]

Like a week or 2 ago it dropped from 600/day Research to 25/day and got so much backlash that it moved up to 200/day. But that was a test. They changed it to 20/month now (no, not a mistake). Which is in their blog and documentation now.

99.89% reduction

[woeful flare]

We're living like in the book 1984...

[quartz wave]

Yes

They tried to be sneaky and secretive about the exact changes too. And now after those undisclosed changes they are denying refunds. Which is highly illegal here. They just think nobody will take legal action when it will cost more than they are being charged.

What they didn't consider is that I don't care. And will lose money to help ruin their scam.

[vagrant scarab]

Unfortunately for them, in my country there's an agency called PROCON, they handle all the paperwork for free, the government pays for it.

[quartz wave]

Please pursue that. It is important. And they are much more likely to give refunds if they are risking their own ability to sell their service in your country.

[vagrant scarab]

I'm just waiting for support to respond because I need a protocol number to file a complaint with PROCON (Brazilian consumer protection agency), I will definitely do that.

[quartz wave]

Nice

[coarse girder]

100% maturing is realising that paying for ai is basically paying to give u wrong information almost every single time

[vagrant scarab]

Apple, a few years ago, tried selling iPhones without chargers in my country, lost the case at the consumer protection agency (PROCON) and was forced to reimburse everyone affected, and now has lost another case at PROCON against Epic Games.

[coarse girder]

Doubt anyone gonna do a long lawsuit for perplexity.

[vagrant scarab]

My country is becoming a European Union. HAHA

Go out on social media and complain, because without that, I highly doubt they'll take any action, especially given how it happened.

[quartz wave]

https://tenor.com/view/the-boys-homelander-i'll-lose-everything-i'll-have-nothing-to-lose-gif-14868566852172046992

Also tbf "long lawsuits" are not usually how these things are handled. It wouldn't make sense. Yes it will cost much more than the subscription but I can make that work.

[vagrant scarab]

But usually when you win, the losing company ends up covering the costs.

[quartz wave]

It depends. Maybe. But you still can't rely on them losing because... they don't always. And you should never put yourself at financial risk for it.

[vagrant scarab]

I, at least, am relieved because I know that many PRO users are complaining about this scam.

[quartz wave]

Yes. It genuinely is a scam.

[coarse girder]

I have pro though?

[quartz wave]

Yes. Pro's research use was cut by 99.89%

Cut, not cute. But it is cute that they think I wouldn't win a dispute against them.

[true sphinx]

https://tenor.com/view/jgmm-monkey-think-monkey-meme-ponder-monkey-idea-gif-6401133862108294696

[plain grove]

I don't understand, why are the limits for sending photos to Neural Network so low? Can you tell me what the limit is?

[tranquil adder]

But why they did this change?

[quartz wave]

There is this wonderful thing called capitalism...

[tranquil adder]

lol

[fickle sorrel]

DEEP RESEARCH on and this is the result? business pro account according to my boss

[wanton crest]

There's also this wonderful thing called "false advertising". "Unlimited deep research." Report them in your region.
https://www.perplexity.ai/pro

[green crystal]

thats not perplexity?

[quartz wave]

Oh I am doing more than reporting

[green crystal]

looks more like chatgpt without deep research being toggled on

[wanton crest]

It's probably also breach of contract, possibly fraud, depending on where you are.

[quartz wave]

Yes and yes

[plain grove]

Can you tell me what the current limits are for sending photos to the chat on the Pro plan? I've sent about 10-15 in the last 2 days, and the limit has reached its limit.

[fickle sorrel]

it is on

https://i.imgur.com/FePj3mA.gif

[green crystal]

oh thats bugged lol

because it should first ask you questions

and then you'll see it take its time

[fickle sorrel]

right, i remember

[fickle sorrel]

i think the business pro account is the issue

i might got scammed

[vagrant scarab]

rsrsrs

[nocturne socket]

Hello, I'm having a problem with Perplexity when I try to upload images to the discussion. I have a pro subscription to Perplexity, but nothing changes.

[vagrant scarab]

This is the first time I've seen a limit on image uploads in an AI; if it were generating images, I could understand.

[nocturne socket]

I've never generated an image; I only upload images to get a prompt.

[vagrant scarab]

This new update is truly devastating; it has harmed every possible PRO user. The genius who designed it deserves praise.

[nocturne socket]

Is there no solution to avoid this message or this problem?

[plain grove]

How much can you send anyway? I still don't understand, I ran out too.

[vagrant scarab]

Well, up until this new update, I never had any problems sending messages, I never hit a limit sending or searching for anything, but now after this update I only see people complaining about practically everything being limited on the PRO plan.

[warm reef]

"You've reached your weekly file upload limit." Should just downgrade to Free and enjoy same shit xD

Time to switch to ChatGPT or whatever external AI Agent. I think Claude's better..

[vagrant scarab]

I'm testing Brave Search right now; it's free and has a premium version for the browser.

[gritty root]

😒

[vagrant scarab]

I subscribed to Gemini PRO; the limits are higher for text and image generation.

[nocturne socket]

I've uploaded 7 images since this morning and now I'm blocked.

[warm reef]

I just need one for coding.. Been using Opus 4.5 for coding but now it's better to switch to Anthropic itself xD

[fervent zealot]

What is the purpose of this action?

What is the purpose of this action?

[weary surge]

Hello everyone, I have met a issues that my education pro is not working till today. However, I had signed up as a university student to get a free 12 months of pro version of perplexity in December 2025. I just sent a email to the support team and I would like to seek for any assistance from here🙏🏻

[mental dew]

Wild 😭, guess it’s time to switch to another service

[vagrant scarab]

Good luck

[short narwhal]

Check if you need to add a card to your billing details in https://www.perplexity.ai/account/details

[coarse girder]

same

i added my card and nothing happened

[warm reef]

Looks like PPLX is looking to copy Claude's pricing?

Wasn't PPLX more expensive before?

[short narwhal]

It's $17 a month when billed annually, which rounds to $200, which is the same as it has always been

[warm reef]

Yeah I'd rather switch to AntAI now

[weary surge]

Thank you

Let me try it

[stone lark]

I'm an professor at Montgomery College and my students are getting errors trying to get signed up. We have install Perplexity and Comet on a 40 machines in six different computer labs at different campus locations to enable students to access Perplexity and Comet. Students are not getting the education license screen and are not being directed to the SheerID site to verify their student status. I will be teaching six sections of my Intro to Generative AI courses this term at the different campuses. Please assist.

[fallen tundra]

is this like a temporary thing? cause i ran out of both deep research and file uploads...haven't even used it that much

[fallen tundra]

it was better value before, used to get actually unlimited file uploads and like i think like 600 deep researches limit per day

[nocturne socket]

You're subscribing for nothing in the end 🥲

[fallen tundra]

i got it free through airtel

but still disappointed tho...was really planning on purchasing it for college

[distant aurora]

@green crystal it's happening https://www.vxinstagram.com/p/DUafwYijUbL

[young hollow]

Why Would A Ai Pay for That

💀

[distant aurora]

Bigger question is who's paying the ai .

[young hollow]

Probably Some Company

A Public Stunt

[distant aurora]

100 dollars is a lot anyways.

[fallen tundra]

it's probalby one of those stupid comapny pr stunts....kinda like that vending machine that cluade made..but that was actually cool tho

[distant aurora]

For just holding a banner for like a day.

[young hollow]

Not for The Big Players

[green crystal]

Lmao

[woven jolt]

Bonjour Bonjour j’ai un abonnement payant chez vous mais je suis limité pour envoyer des analyses de documents pour quelles raisons

[vagrant scarab]

Bonne question, personne ne sait pourquoi ils ont décidé de rendre le plan encore plus mauvais.

[distant aurora]

Dry mouth is awful.

Can't even enjoy chicken nuggets without sauce.

[indigo quarry]

Bonjour j’ai eu le même soucis. Option affichée illimitée dans l’abonnement, mais en réalité il y a une limite…
C’est illégal, ça s’appelle de la publicité mensongère.

[edgy nebula]

I have a question for you what is the best for deep research Gemini pro or perplexity because yk after this update i don't want to use perplexity

[untold flume]

I have both, for now perplexity is better for deep research

Like much better

[edgy nebula]

broo i cant use deep research i depend on it so much but now after this update i cant do deep research

[hazy saddle]

Hi Folks, how many deep rearches have perplexity pro and perplexity max?

[chilly stump]

ive seen some people use system instruction prompt in gemini to get better results in deep research

also you can try notebooklm for deep research

[desert canyon]

So has perplexity released any kind of statements this week about the changes to terms? Or they are just avoiding it?

[edgy nebula]

Everything would have been better if perplixity didn't do this update

[hazy saddle]

It looks like Perplexity Pro lowered its deep resarch limit, right?

[desert canyon]

Yes dropped lots of limits; labs, research, file uploads, etc

[hazy saddle]

but why?

[pliant solstice]

lowered everything 💀

also the amount of files that you can also upload if you're pro

[hazy saddle]

what reason of?

[untold flume]

lack of token

That’s the basic point

[pliant solstice]

we don't know either, there is no public statement

[hazy saddle]

but I pay the same price for Pro?

[untold flume]

Gemini downgrade a lot as well, maybe its the electricity

[hazy saddle]

what do you mean with lack of token? token?

[untold flume]

Lack of token generating capabilities

[hazy saddle]

token for what?

[untold flume]

For everyone to use AI?

[hazy saddle]

is it a temporarily problem or for ever?

[untold flume]

Every question you asked need to be answered with token generation

It can’t be permanent

With upgraded AI model and Hardware

[hazy saddle]

shall it be clever to change to Perplexity Max?

[untold flume]

Depends what you need

I assume it’s not about coding

[hazy saddle]

more seep researches I need

[untold flume]

Well in that case

Perplexity max or Gemini ultra

Both are good

[hazy saddle]

but I never know how many deep researches I get

[untold flume]

Have you tried Kimi agent swarm?

[hazy saddle]

what is this

[untold flume]

That’s the best AI for research

Just Google it

[humble lichen]

Hello

[untold flume]

Perplexity has Kimi 2.5thinking but no agent swarm

[hazy saddle]

can I get 40 deep researches per day mit perplexity max, what do you think?

[untold flume]

Damn I don’t know but I think Gemini ultra can do 40 per day

[hazy saddle]

okay okay

[untold flume]

Wait

Don’t you have perplexity pro?

[hazy saddle]

I have perp pro

for long time now

[edgy nebula]

Bro kimi is more expensive

[untold flume]

$200 per month?

199

Yea it is

But if you willing to pay $200/mo you probably should subscribe to Claude opus 4.6

[edgy nebula]

im student so i just pay 5-10$ in perplixity

[untold flume]

That provide same features as agent swarm

Anyway it’s hard to say which one is the best

[hazy saddle]

hm

what? kimi agent swarm looks similar perplexity in browser

do they use the same surface?

[untold flume]

probably not

It’s whole different thing

[hazy saddle]

I mean this website surface looks very similar

[green crystal]

they're copying most of chatgpt

[hazy saddle]

Kimi Agent Swarm needs much time for one run, seems to be a clever thing

[olive gale]

Guys, does anyone know an AI that can replace Perplexity? it's impossible to use it anymore

[untold flume]

so is it good?

[hazy saddle]

yes, it's impossible, I'm actually trying Kimi 2.5 Thinking and it's very good

[untold flume]

Great

[hazy saddle]

My first trial with Kimi Agent Swarm was fantastic

@olive gale Try Kimi

[untold flume]

I have two instruction sets I found very useful

[olive gale]

Well, I don't even know. I have a bad attitude towards Chinese AI. I used a DeepSeek for a long time and it's just disgusting.

[untold flume]

I can show you if you want

[vagrant scarab]

Gemini's research has more content, while Perplexity's is more concise. If you don't mind reading, Gemini uses more sources and conducts more extensive research; in addition, it creates a research plan beforehand, which you can edit or request changes to before it begins.

[untold flume]

These instructions must be followed from now on:1)Act like 2 persons: Person A is the Prompt engineer who will build a good prompt based on the user's demands. 2)Person B will execute the detailed prompt built by Person A to fulfill the user's demands. 3)Do not output Person A's works to User. 4)Always use web search.

One more instruction: For all answers involving facts, data, or news, strictly implement comprehensive, in-depth verification and fact-checking. Adversarial searches must be conducted to eliminate outdated information, the latest timestamps must be forcibly checked, and skepticism should be maintained towards single sources. Do not perform linear extrapolation; multi-source cross-validation is mandatory.

See if you needed

[olive gale]

@hazy saddle Can you please tell me more about Kimi?

[untold flume]

100dudes work for you

That’s how it looks like

[hazy saddle]

but these dudes are no real guys, right?

[untold flume]

They are not

[hazy saddle]

100dudes seem to be agent swarm only

[untold flume]

Yep

[hazy saddle]

I don't see them using Kimi 2.5 Thinking

[untold flume]

They are separate models

[hazy saddle]

yes

[untold flume]

Kimi 2.5 thinking is open source model which can be found on perplexity

But that agent thing is unique

[hazy saddle]

Ouch, Kimi is a Chinese bot?

whoa

[untold flume]

Don’t worry it’s Apache 2 licensed

[hazy saddle]

how does it work with restricted Chinese internet access

[untold flume]

I have no idea

As long as it’s working

[hazy saddle]

if it's Chinese the bot seems to be one of the best. Typically Chinese

[tardy vessel]

tf do yk

[untold flume]

Claude opus4.6 comes up with same features

Pretty much the same

[tardy vessel]

i see

guys ai is gonna eat us lets become ai

[tardy vessel]

lol this word means worm in my lang

[untold flume]

What

[tardy vessel]

kimi

that name

[untold flume]

Oh

[tardy vessel]

man i heard the antigravity ide is soo good

[untold flume]

It is

Those AI are open to each other if you paid $200/mo

[olive gale]

Has anyone used Claude? A lot of people recommended it to me, but I'm still unsure if I should use it.

[lunar vigil]

hi

[past horizon]

Model council stopped working for my Perplexity Enterprise Max. Anyone else?

[untold flume]

Can be switched on perplexity

I subscribe to perplexity for scrolling news, very useful

Kimi for office work and Claude for Coding

[olive gale]

I just need to replace Perplexity with something because I can't use it anymore.

but I don't know for what

[untold flume]

Well if your budget is around $200/mo i highly recommend you get all of them

[young hollow]

200$ is probably his monthly expenses

💀

[untold flume]

If it’s $20 you should go for Gemini pro

[olive gale]

Well, no, $200 is too expensive for ai for me

Perplexity was perfect but due to the new policy I can't use it.

[golden cargo]

Hlo

How can I persue cs engineering

But the problem is

M in biology

In

Usa , canada, uk , Germany, australia,

Anyone

[olive gale]

Unfortunately, I can't use Gemini either due to restrictions on my Google account.

[chilly stump]

Did the upgrade the deep research yet?

[untold flume]

What new policy ?

[young hollow]

Yes for Max Users

[tardy vessel]

lmao thats what i get paid

[young hollow]

I thought so

[untold flume]

Liar

👀

[tardy vessel]

I am doing freelancing rn so ye only around 400 usd monthly

imagine paying all that for ai

[young hollow]

💀

0 net profit

[austere bone]

Hello everyone

[olive gale]

Well, in short, they reduced the number of available queries in Deep Research and Labs from 600 to 25 and from 50 to 20, respectively.

which is incredibly little for me personally

as I was told (I haven't noticed yet) they have reduced the number of requests in comet assistant

[wise nova]

Check you Claude's usage page
https://claude.ai/settings/usage

[wise nova]

It's now paid Free plan

[olive gale]

Guys, can anyone recommend an AI to replace Perplexity?

[lapis tide]

buy a gpu and run it locally

[olive gale]

no i dont wanna local ai

[lapis tide]

then claude

[sly python]

Not working for me either

[warm reef]

Depends, for development use Claude

[final patio]

lazy mother....

I seriously told Comet to take over browsing and fix citations on a Notion page.
Comet opens Notion page and goes straight to the AI agent in Notion and writes:"fix the citations." and waits until the work is done.

[onyx pine]

same

[proper oracle]

AI outsourcing its work to another AI, lmao

[stuck shuttle]

Bruh why is the card getting declined

I have tried 2 cards and they both were declined

what could be the reason?

[short narwhal]

@twin grail @gloomy cipher one of you infected my PPLX 😭

[gloomy cipher]

https://tenor.com/view/stretching-warriors-song-lets-do-this-preparing-myself-ready-to-play-gif-25793593

[violet peak]

Thats a big L

[short narwhal]

Is your card either, Visa, MasterCard, or AMEX? Are online purchases and subscriptions allowed on the card?

[violet peak]

Wanted to extend my subscription but because of this im thinking of going to another one since the new policy

[stuck shuttle]

idk but it accpeted one of my card

SBI Prime somthing

😭I am so happy

[violet peak]

Man

Its Ai why the limits bruh

[glossy sequoia]

I feel bad for putting people onto Perplexity

[wise nova]

It's Claude, most limited AI

[violet peak]

It was amazing, now idk but I still have to use it

[violet peak]

What do you recommend to use ?
I just used perplexity and didnt try anything else

[viscid hawk]

Just curious how do you guys blow through all queries available. I run a business and mostly use perplexity to study, analyze marketing data and even generate css for my site and i don't even use it that much

[wise nova]

Gemini AI Plus/Pro, ChatGPT Plus is enough

[wise nova]

Pplx won't be able to retain users unless they offer more than Google and OpenAI for same or less price.
Google offered family 5-person Gemini Pro and shared 2Tb for just $20. No one gives the same options

[icy parrot]

True, but also google is trilion company they have more money they can throw aways then pplx

[silent ocean]

@short narwhal when will we see sonnet 5?

[icy parrot]

Sonnet 5 isnt even out yet

only opus 4.6

[silent ocean]

yeah right but when do you think we will see it

[icy parrot]

If you look at time difference between Opus 4.1 and sonnet 4.5 then i would argue that it would be same

[short narwhal]

Never at this point (could not tell you nor has there been any rumors)

[icy parrot]

Since anthropic did that in summer too they released opus 4 then opus 4.1 then sonnet 4.5

[silent ocean]

appreciate it

[bitter grotto]

they basically made the Pro plan ($20) be more or less on the level that the Free one is, excl. some minor limit increases + more models

[hushed urchin]

fr

what do people think about this idea:

ability to highlight sentences / words from an output, and then have a really lightweight model built in to quickly explain it without makign a new thread.
It is for researchers and students afterall..

where do I present this idea to them

[hushed urchin]

rumored next week, I'll even bet on it

[livid apex]

Hi everyone, ouch, I’m finding out the hard way that the behavior or usage limits have changed. In my case, trying to download a file shows a message recommending that I switch to the Max option. Can anyone tell me more about this? Or how long it takes before the service is restored?

[hushed urchin]

we're all about to protest

;p

[raven marlin]

Fun Perplexity use I've found is having my wife Send me her weekly shopping list on Discord then copy/pasting it into the browser and ask it to set a pickup time at walmart for that list and to take over the browser and it just adds everything from the list to the cart and picks the best time then sets it up in my schedule to go pick it up.

[hushed urchin]

okay so now the fucking bot comments come

[bitter grotto]

I'm quite surprised all this hasn't somehow hit the news still lmao

[hushed urchin]

yeah

[raven marlin]

Eventually my wife is going to catch on and add something expensive for herself...

[hushed urchin]

@vague pagoda tell the team we're bout to protest for perplexity pro

[hushed urchin]

bro what

[livid apex]

oh.. I wish the best of luck to those who are going to protest.

[raven marlin]

Fun Perplexity use I've found is having my wife Send me her weekly shopping list on Discord then copy/pasting it into the browser and ask it to set a pickup time at walmart for that list and to take over the browser and it just adds everything from the list to the cart and picks the best time then sets it up in my schedule to go pick it up.

In regards to this^

[hushed urchin]

yeah it's just out of the blue, but cool dude

good positivity now that perplexity has made pro like the free version

[raven marlin]

yeah, not sure what the benefits are of having pro if the free version does basically the same thing

[livid apex]

I would’ve thought that making the terms of a paid plan the same as those of the free one would be more of a bug than a logical decision.

[crimson wyvern]

Whether protest or not, I don't think Max membership will be as popular anymore, because they stopped offering unlimited use of Max membership.

[hushed urchin]

problem is that it's a standard tactic to downgrade their features slowly if it isn't sustainable to them, so we're afraid it's just gonna keep coming

[hushed urchin]

holyyyyy even max is getting reductions?

[raven marlin]

I'm not a super heavy power user I don't think so maybe it's time to drop the pro if I can just get the same thing from the regular?

[bitter grotto]

https://futurism.com/perplexity-ai-ceo-denies-rumors

"Our goal isn't to save money and scam you in any way."

LMAO

old article but still

[crimson wyvern]

Yes, they have already done that. It happened at the same time as they nerfed Pro membership, but the difference was that Pro members noticed it immediately and Max's documentation was updated a week or two later.

[hushed urchin]

wow.

I always loved perplexity

;(

[raven marlin]

I like perplexity for comet tbh

[green crystal]

the llm wrapper got wrapped up

[hushed urchin]

yeah thats good

[livid apex]

Guys, I hope you weren’t doing any major work that depended on access to the Pro option. How long have you been noticing the issue?

[bitter grotto]

From what I see/hear, they introduced this final nail in the coffin yesterday

[livid apex]

oh thanks @bitter grotto

[crimson wyvern]

What happened yesterday?

[glacial geyser]

bad maj restrictive i guess

[bitter grotto]

the file upload limit got introduced (at least for me and some other people on Reddit)

[crimson wyvern]

Didn't this happen weeks ago?

[bitter grotto]

Well I haven't hit it before and my uploading habits have always been the same sooo.. not sure; it's scummy anyways, especially since they conviniently 'forgot' to update the upsell page and it still says 'unlimited'

[hybrid grotto]

Did I read that wrong?! Not only did they restrict labs to 25 per month, Research is now 20 per month instead of 20 per day?!

[bitter grotto]

yes

[crimson wyvern]

#💬│general message

[glacial geyser]

After an in-depth search, I was limited. How do you know what my limit is? (with pro plan)

[bitter grotto]

Yeah they seem to have updated their self-help page, which is not the thing you see when you go to see the upsell

[hybrid grotto]

Now who would use Perplexity with 25 Labs per month, 20 Research per month and 600 searches per month/week/day idk, and image and video generation restricted in india too

[bitter grotto]

https://www.perplexity.ai/rest/rate-limit/all

[glacial geyser]

thx

[hybrid grotto]

They literally killed it

[livid apex]

would it be a geographic restriction?

[bitter grotto]

Nope its global

[hybrid grotto]

Yeah image and video one, it shows due to indian govt orders....... It is restricted

[bitter grotto]

Image/video is global too

[livid apex]

(greed too)

[hybrid grotto]

The image video restriction is global?

[bitter grotto]

Yep, I am in EU and don't have it

[bitter grotto]

So I guess it might be this + them nerfing it just like the pro plan itself

[hybrid grotto]

And that 600 pro searches is monthly or what?

[bitter grotto]

the seraches I think not, I think they are the only thing that remains as 'daily reset'

[hybrid grotto]

Hmm

[north steppe]

I got a tip for the people here that use revolut, you can get a 12 month pro plan if you have revolut premium,

[bitter grotto]

or 24 with metal, yeah;

[north steppe]

yeah really handy

[hybrid grotto]

Everyone knows that, PayPal and airtel and what not, more than 95 percent of pro plan is free users, maybe that's the reason, though now their paid plan is worse than the free ones of others

[bitter grotto]

seems like their problem lol, should have stopped offering that long time ago

[hybrid grotto]

We are getting 1.5 years of Google Ai Pro in india with just an indian sim card by JIO ISP, it gives us 2 TB of cloud storage and the plan can be shared to like 5 other people too

[north steppe]

I read that most companies are gonna downsize ai anyways, cause it doesn’t make profit

[north steppe]

thats crazy

[bitter grotto]

let's hope this happens fast and we get normal RAM/SSD prices back

[hybrid grotto]

And all ai benefits and stuff, the plan gives out a lot

[hybrid grotto]

Yeahh

[north steppe]

I mean the bubble is gonna burst anytime now

[glacial geyser]

2027

[hybrid grotto]

That's true but i feel like ai is revolutionary, I do want the prices to stabilize, but i don't want the "destruction of AI"

[north steppe]

im almost 100% sure ai will keep on growing, but only if we find other ways to “feed” it

cause the way 90% of the world uses it, is just stupid.

[hybrid grotto]

I as a developer made many automations and systems using it, I manage servers and websites and what not, i learned all of that using it, even used this free google to generate videos and images for some of the businesses

[north steppe]

Thats actually the way it should be used

[hybrid grotto]

Yeah

[hybrid grotto]

Yeah

[north steppe]

so im sure it’ll keep on existing, but probably overhauled so its more cost efficient

[rocky yarrow]

Did they fix PPLX yet?

[hybrid grotto]

Hmm

[hybrid grotto]

Made it worse

[rocky yarrow]

Mine is still broken

Can't do deep research

[hybrid grotto]

..

[rocky yarrow]

I'm really hoping for a response from the team soon. I can't do my job rn and I might need to switch to someone else for deep research.

I'm so annoyed that I have never gotten Pro for free and all this happens after they give it away to everyone.

[north steppe]

Why cant you do your job?

[hybrid grotto]

Grok offered a really damn good deep research, these days, if you really wanna overdo it, like the deepest of deep research, use google research

[bitter grotto]

tbh idk what other way to feed it data there is, apart from it training on its own generations

[hybrid grotto]

Gemini deep research i mean

[north steppe]

i meant feed it as in the way to keep ai alive

Burns through a lot of resources

[glacial geyser]

Honestly, I'm disgusted by what they did.

[north steppe]

Ai data centers like the one elon has are CRAZY when it comes to how much energy they cost

[glacial geyser]

I will upgrade my claude subscription

[hybrid grotto]

If they had to take such step, I am sure Perplexity is about to be bankrupt or really low on resources, or who in their right mind would take such a step for their pro plan

[livid apex]

Guys, I don’t know what you think about marketing. I can accept that it makes us pay a bit more once we’ve built up a habit with the product, but I’m more puzzled if it’s about setting invisible traps… and reusable ones, because nothing guarantees that upgrading to a higher plan won’t cause the same problem again.

[glossy sequoia]

Research was effectively unlimited. I used hundreds a day on pro before

[hybrid grotto]

Hmm

[glossy sequoia]

It would say 600 a day, but then never go down as you used it

[hybrid grotto]

I connected a US vpn and then it worked

[bitter grotto]

Ah so they left it as US-only; then it might be something law-related

[livid apex]

oh, it's a geographic restriction

[tawny lichen]

time to go to gemini

what a stupid decision to limit research that much when its the only reason to use perplexity

[glacial geyser]

Perplexity just advised me to cancel its own subscription.🤣 almost something work

[livid apex]

I'm speechless... Are you in the US, @glacial geyser?

[hybrid grotto]

Try to scroll to the bottom of this document and you might hate Perplexity a little less

[glacial geyser]

EU

[livid apex]

oh you too

[marsh dirge]

Hey, how can I contact support directly cause they wont answer to emails

And they’ve been sitting on my money for months

[hybrid grotto]

This is the Research Pro version for this prompt, 77 pages

This is the deep research/ web version, 104 in this one

[livid apex]

Hi @marsh dirge, I wrote to them, and even though their reply encouraged agreeing to the Max option, the guy was very polite. Try not to show that you’re projecting your frustration onto them — they’re human beings, they’re just doing their job, and they don’t want to lose it.

[marsh dirge]

Thank you, what’s my next move?

[snow ridge]

hi

why is my pro plan cancelled and suddenly its asking for billinf details?

[livid apex]

We'll find a solution in asking the community, see, a lot of guys are reporting comparable issues

[fast pasture]

Hello folks

How do we add perplexity ai bot to the Discord server

@tranquil orchid

@slim gull

[short narwhal]

It's paused, not cancelled. It'll continue after you add a card, which is there to prevent fraud. You won't be charged or anything till the end of your billing cycle

[short narwhal]

Not possible :(

You can make your own bot though

[fast pasture]

How about I use perplexity api keys to make my own bot

[naive sage]

Would be curious if you could share your prompts, perhaps in #1124060874785685554

[livid apex]

Maybe it'll be helpful to show you my last message, currently unanswered:
Thanks for your reply. I understand that my Pro usage is restored on a weekly basis, and that you recommend either upgrading to Max or being more “thoughtful” with my usage.

My difficulty is that I have no visibility into the concrete thresholds I’m crossing, or into any usage meter in the product. Without that, it’s impossible for me to consciously adjust my behavior.

Could you please clarify:

which type of usage I exceeded (e.g. Pro Searches, Deep/Research, Create files & apps, advanced model calls), and

whether you plan to expose a simple usage indicator in the UI so Pro customers can self‑regulate before hitting invisible limits?

Right now, I feel I’m being asked to “be more thoughtful” without being given the basic information needed to do so.

[glacial geyser]

They just want you frustrated and dependent. For me, the fact that they removed scientific search and added excessive limitations, without any communication, fully justifies looking for solutions elsewhere.

[naive sage]

They still have "Academic" as a source, was there a separate "Scientific" one in the past?

I think the only one I don't see anymore is "Financial"

[glacial geyser]

No sorry academic, you right. But i didn't have it anymore.

[naive sage]

On the web UI, do you see it here:

[livid apex]

you're wise and right, @glacial geyser
I hope your research tasks haven't been too dependent on Perplexity

[glacial geyser]

i use it a lot 😅

[zealous forum]

Ici on parle anglais déso

[naive sage]

I wonder if that's because it's in an existing thread - does it still show up for new thread view?

[mortal jewel]

havent used perplexity since the upload limit the other week, is it fixed?

am pro user

[glacial geyser]

it's because I was using Complexity | Perplexity AI, but the sudden update made it obsolete… The good news is that I do have access to searches again 🥳

Thanks, I wouldn’t have understood without you.

Oceanic

[naive sage]

At least one issue solved!

[livid apex]

oh I see @glacial geyser

[bleak lichen]

Did the Comet Assistant feature get a model selection option? I hope it's not a UI glitch like before, but a real option I've been waiting for this for a long time.

[haughty jasper]

who makes the videos for the updates?

[ionic stone]

Opus 4.6. Build me an app. Make me money. Make no mistakes.

[shell mulch]

I'm on Pro but just found out there is an Education Pro plan that I'm eligible for. However, when I go to change my plan it's not listed as an option. Any ideas, please?

[twilit crown]

Why no Opus for free users?

Apps like Yupp literally pay me to use Opus

At least give a couple prompts per day

[dry kestrel]

they view and train on your data to make money

[dry kestrel]

Costs a ton compared to other models

[twilit crown]

What is it the chart showing?

It could possibly be money generated per prompt

[dry kestrel]

Cost to run artificial analysis benchmark

[twilit crown]

Grok 4 is above

We already have Grok 4

[dry kestrel]

And?

[dry kestrel]

Grok 4.1 is not Grok 4

[left jasper]

So how is Model Council even a "feature"? As far as I gathered, perplexity does the source gathering, then sends the data (more or less identical) to 3 differnt LLM models and asks for a summariztion. The result should be within margins of error from them all.

It would be different if it would run the Gemini 3 search + summary, GPT search + summary and Claude search + summary but that's not what's happening, is it?

[twilit crown]

We had Grok 4 when it launched

Grok 4.1 was just replacing it

I get like 100 prompts for $10 in Github Copilot for Opus

And it is just per task

So I could ask for 100 things in a single prompt

[dry kestrel]

At the time when it launched Opus was still more expensive

You can't expect Opus on a $20 subscription - it's not sustainable

[left jasper]

and if you reuse the same session you can run more prompts. copilot (unless they've changed it) counts a session as one premium request, not every single prompt

[dry kestrel]

If you want to try out Opus you can get a Claude Pro sub

With extremely harsh weekly and hourly limits

[twilit crown]

That was the $75 per million output token model

4.5 is like 3x cheaper

[dry kestrel]

Hence At the time

[twilit crown]

Google subscription gives me Opus

[dry kestrel]

Actual results prove otherwise

[left stream]

this world is dark

now that I have the knowledge I wanted in cybersecurity I regret it

[dry kestrel]

1451 compared to 1866 is not a 3x reduction

[left stream]

this world is doomed

[bleak depot]

🙂

[dry kestrel]

https://x.com/AravSrinivas/status/2019862450401956186

Aravind Srinivas (@AravSrinivas)

We're going to figure out ways to provide access to Council Mode to our Pro users too. There will be rate limits of course since it's expensive to run this, in which case it would make sense for Pro users to upgrade to Max in that case. Stay tuned.

X · Feb 6, 2026, 7:55 PM

[left stream]

too bad everything is always free for the eyes of the knowledgeable

see you guys again in a couple of months toodles

[dry kestrel]

lol cya

[naive sage]

👀 What do you mean by that? What does he mean by that? Knowing about stuff doesn't make it magically free.

[south lance]

Try google antigravity, it's kinda new so the limits are generous. As a Pro subscriber, I don't think you can physically hit a limit while using 4.5 Opus (I assume 4.6 will be added soon)

[short narwhal]

Woowie

[twilit crown]

I use Antigravity

It is great

It replaced Perplexity for so many research things

[obtuse coral]

guys what is the maxiumm files size you can upload with pro account

i legit just jumped on today and already hit a weekly file size limit wtf. i havent used prop since last week

is it even worth upgrading

i legit have a sports model i need to fix and the files error is a big set back

[night isle]

.

[ivory zinc]

Hi

[austere acorn]

Hell yea first time ever getting an ai to curse, all natty

[silent ocean]

@short narwhal something is happening

white colored pro?

and theres no education pro available right now

so either they're updating something or they're sunsetting the education plan

[short narwhal]

Nah mine is white too now

RIP the blue color...?

nvm it's back

nvm it isn't

[distant hare]

Hi

Anyone knows how to check if my subscription is through my Revolut card?

[wide bluff]

guys when i try to send my files to perplexity chat it says thinking and doesnt answer me i have 23 files and they are maximum 9KB what can i do

can someone help me please

[austere acorn]

reduce file size

[novel arrow]

opus 4.6 isnt working for max users, i pay max and dont het opus , what is perplexity doing

its stuck for 10 minutes

[stark hazel]

F hate this ai response from the support team, get me to an agent but no, it wouldnot

[languid skiff]

tengo perplexity pro, cuando lo adquirí generaba imágenes, ahora me dice que en mi region no es posible. Mi plan es de un año, no entiendo porque las condiciones han cambiado sin previo aviso.

[novel arrow]

try vpn usa

[naive sage]

Opus 4.6 Thinking must be slow.. does it show anything if you drill into the Opus reply that's in progress?

It looks like it started running, I wonder if the API timed out or something.

[manic night]

so how do i access the council

sorry if thats a dumb ahh question

oh nvm

its for max only

and i checked on my phone

now im on my laptop, and i see, its max only lol

[naive sage]

Yeah, they're thinking about bringing it to Pro with a few uses to try it out, but right now it's only Max

Basically, a fun feature you can use a few times, then it serves as an ad to upgrade to max. 🙂

[manic night]

idt its just fun

it'd be really cool to see what a few ais discuss with each other before coming to me

oor chatting with them

i wonder if an oss exists

[naive sage]

The summary response that evaluates them is neat. There's a lot of tools that run parallel replies, but I don't know of any that provide good summarization/judging of all the responses to identify similarities, differences / unique insights, etc.

[manic night]

might as well make my own

[robust terrace]

Seeing a lot of new stuff for Max subscribers

is that common or is it a good mix of max and pro

[dry kestrel]

https://x.com/buzinover/status/2019896626567848447

mb (@buzinover)

We upgraded our memory agent for max + pro users

Expect Perplexity to be better at catching key details, reasoning through time, and pulling needles from the haystack of your past chats

▶ Play video

X · Feb 6, 2026, 10:10 PM

[short narwhal]

Is that why it's pulling "you play league" from one single question I asked 3 months ago 😭

[dry kestrel]

AHAHHAHA

[naive sage]

Yeah, that's essential to building advertising profiles for people, to either sell to data brokers, or to better monetize ads through specific demographic targeting.

That is the direction that all of them will go - you already see ChatGPT monetizing ads (with some of the highest ad premiums on the market), Google is... Google, and Perplexity has made no secret about building user profiles for stuff like that. Anthropic (Claude) is the only major provider that said 'no' to ads... not sure if that will change, as it seems inevitable.

[main carbon]

What’s with the new limits? I’m in the middle of an annual subscription and logged on to see I have 7 research queries left for the month?!?!?

[naive sage]

The limit is now 20 per month for Research. Is it possible you used 13 over the past week or so?

It is the new normal.

[main carbon]

It was 200 when I paid for my annual subscription

[north olive]

hewo

[main carbon]

So how can they just cut it in the middle of a subscription?

[naive sage]

Maybe they will bring the old model back with increased limits if people complain a lot, but somehow I doubt it... it sounded like it was more strategical. One person indicated someone said that lots of people were using research for basic questions, and that seemed to be wasting compute.

[tranquil orchid]

3. Treat community members with respect.

No harassment, hate speech, or malicious behavior. Keep things positive and constructive.

[naive sage]

ChatGPT only gives 10 Deep Research queries per month with their good model (and 15 additional ones with the lower tier model).

[regal gyro]

why ther daily file limit is 10 on PRO ?

[main carbon]

Are any Perplexity humans in here? The email bot is useless. My issue is that I paid for 200 a month and I got cut to 20 with no notice. Actually, even had someone bothered to send an email, it’s not legal to claw back. Ew subscriptions or renewals, yes. Existing, no.

[naive sage]

I think Google's AI Pro plan with Gemini is a decent option if you need more,... 'up to' 25 Deep Research reports per day isn't terrible (which is what Google's plan provides), even if you effectively get half of that it's still more than the main competitors now. I'm not sure what the research limit is for Anthropic, but given Claude's tendency to be more expensive it's probably not a lot. I'm not sure about the Grok limit either.

[main carbon]

It’s 20 a month on my Perplexity Pro now, not 20 per day.

[serene sand]

https://x.com/perplexity_ai/status/2019126571521761450?s=46

old DR is dead

you get new SOTA DR

Perplexity (@perplexity_ai)

We've upgraded Deep Research in Perplexity.

Perplexity Deep Research achieves state-of-the-art performance on leading external benchmarks, outperforming other deep research tools on accuracy and reliability.

Available now for Max users. Rolling out to Pro in the coming days.

X · Feb 4, 2026, 7:11 PM

[naive sage]

If Perplexity's new Deep Research response quality surpasses even OpenAI (like that shows), that still might be a compelling value. I do acknowledge that Perplexity's very high Research limit was probably a selling point for a lot of people though.

[orchid trellis]

Yeah there's no humans at Perplexity. I also got an annual plan because of the research queries. It's a 99.89% drop in query limits without warning. And instead of mentioning this, they just keep posting about max tiers new features

I guess for 1 Max user, they make the same as 10 pro users, and let's be honest, it's a good system so the lock in is real

So they're unlikely to car about anyone in the Pro tier

https://www.perplexity.ai/help-center/en/articles/11187416-which-perplexity-subscription-plan-is-right-for-you

This page indicates that even if you have a Max plan you will have significantly less than 500 a month

They only specify a limit for Enterprise Max of 500

But normal Max is unspecified as well.

So even for 200 euros a month, you'll still get probably 200 queries a month if I had to guess and be generous

[orchid trellis]

Bait n switch, free lunch is over tactic

But yeah 2400 euros a year, I'd rather finally get a beast of a machine at home on a 200 euro per month loan or something to run the open source models

[main carbon]

Well, I might have been convinced to go MAX because Perplexity was a real value tool for me. But they just lost my trust.

[orchid trellis]

I mean look if you can afford it, they are doing great work.

But this is some proper low tier customer treatment

[naive sage]

Yeah, I think ~200 Deep Research queries per month is probably about what the Max plan offers now... extrapolating from the ent pricing page. But that's just a guess.

[orchid trellis]

Which doesn't beat just using Opus from Cursor in my own managed file/code base

[gilded plinth]

Hi, has the Perplexity Student Pro plan changed recently?
I currently have the Student Pro plan, but I’m no longer able to upload or attach files.
Is there a new limitation or upload cap for student accounts?
Could you please clarify what the current file upload limits are and whether this is temporary?

[main carbon]

You can tell a person’s (or company’s) character by how they treat the little people. 200 a month would be very tight for me. But like I said, they lost my trust now. It will be a pain to move everything g to a new tool, but I guess that’s what I’ll do. I’ll just have to eat those last few months that I already paid for.

[naive sage]

200/month used to be car prices for me, it's not anymore given inflation, but I still associate that in my mind kind of.

[orchid trellis]

Yeah you don't have access basically any more

99.89% quota drop for Pro users

[gilded plinth]

OMG

[rocky yarrow]

Guys did they fix Perplexity yet?

[naive sage]

The service is operational I think

[rocky yarrow]

Mine is broken and won't let me use deep research anymore

[gilded plinth]

But when did they change that? Because until this afternoon I had access, and now, what's going to happen? Do you know how many attachments are allowed now? They're not responding.

[rocky yarrow]

I think they updated something and broke it

[orchid trellis]

Yeah they reduced the quota for Pro tier from 600 per day, to 20 per month

[rocky yarrow]

What???

Lol

Is the company shutting down?

[orchid trellis]

Oh no, the opposite

[serene sand]

#💬│general message

[main carbon]

Thank you everyone for your help. I guess I’ll pop in here tomorrow and see if anything has changed.

[rocky yarrow]

Yeah I'll just check in everyday I guess to see if they pushed a hot fix and fixed Deep Research

[orchid trellis]

You need to pay 200 euros a month now instead to probably get 200 research queries a month

[rocky yarrow]

I use it literally everyday for work.

[gilded plinth]

Was it only in the attachments section, or are the templates also limited to sending messages? (for the student plan pro)

@orchid trellis

[orchid trellis]

Everything my dude

[rugged scroll]

Started the move to Gemini last night. Getting used to Gems to replace Spaces

[orchid trellis]

Yep, Pro tier is kinda disabled

[rose bone]

*you used it literally every day 😄

[orchid trellis]

Ahhh, having luck with that?

[rocky yarrow]

How are they gonna stay around as a company if everyone paying unsubscribed and all they have is free pro plan people draining tokens?

[rugged scroll]

Yes. Once I wrapped my head around it. It's not bad and I think their Deep Research gives better results.

[rocky yarrow]

I feel like they have to be going out of business soon or something.

[rugged scroll]

I don't think anyone that has to depend on paying others for actual model access is going to make it.

[rose bone]

Anyway, as of now, pro is completely useless. The search mode that's left is a joke. And they definitely won't see me paying 200 plus VAT. 😄

[orchid trellis]

1 max subscriber is 10 of us plebs, and this pushes folks to make Work get the enterprise packages, then likely API as well

The economics are likely sound

Also Opus 4.5 costs a hell of a lot more than previous models

[gilded plinth]

What are the chances of them reversing this and going back to how it was before? Given that, for the community, this will not be accepted.

@orchid trellis

[rocky yarrow]

Yeah but the thing is, my work already pays for Google Gemini and it also has Deep Research

[naive sage]

Perplexity has a few models they host themselves, and their own search index/crawlers. So they're not entirely dependent on others. They could theoretically nix all third party models and still have a product to sell, it just wouldn't be as compelling.

[rocky yarrow]

So like what even is the point? Lol their deep research isn't absolutely gimped like this either and they have image gen and coding stuff

[orchid trellis]

AI is expensive, unlikely they'll care if I'm honest.

But also, we shouldn't be so locked into a service we're we are genuinely trying to figure out how to afford 200 euros a month

[rugged scroll]

I was already paying Google $10/month for 2TB on Drive/GMail. For $20/month I get Gemini Pro and 2TB. so it's actually cheaper.

[orchid trellis]

Like we need redundany 😅

[naive sage]

Their own model enriched with their search capabilities is what they're back to emphasizing as the primary selling point, b/c you get far more usage with that than the third-party models to begin with.

[serene sand]

#💬│general message

Same as OAI bringing GPT 3.5 back

New models new rates, that’s how it goes

[gilded plinth]

But 200 euros in Brazil, where I live, is 1233.81 reais, which is complicated and makes it expensive.

[orchid trellis]

Like, I have not seen a human post from Perplexity.

You can't even do due diligence with a human support person for doing consumer commission complaints because they use AI for the customer support 🤣

[orchid trellis]

Yeah man, I'm from South Africa and that's a ridiculous amount for my friends back home

[naive sage]

(This is the future I think, most places will be going this way. I tried contacting facebook a couple years ago b/c someone was posting my phone number, and although it might've been a real person - not sure - they just used canned replies and didn't even address my issue)

[gilded plinth]

Thank you very much, reflecting. @orchid trellis

[orchid trellis]

It makes the case stronger, and more likely to turn into a class grade issue, but if they don't reply to lawyers or aren't in your jurisdiction (Perplexity is in the US for example so being in the EU the chances for complaints working is a bit wobbly)

Tbh, the fact that this is the discussion, happening in general, I highly doubt anyone who works at Perplexity is in this discord

[rose bone]

Anyway, what alternative agents with deep research of similar quality to old perplexity research would you guys recommend?

Is gemini any good?

[rugged scroll]

I think Gemini's DR function is better.

at least in my first testing over the last day or two

[rose bone]

How are the limits? Perplexity states 20/day for pro, which I could live with, even though I would prefer higher monthly limit instead.

[naive sage]

I have not yet compared it to the new Deep Research model (it's rolling out to pro users), Gemini's seemed comparable to the old Perplexity Research responses. However, they both ended up having some good points that the other didn't when I compared them. I wrote up a bigger analysis of it here: https://pastebin.com/k99sB8g3

[short narwhal]

I personally don’t like Gemini’s deep research. For me even Pro search gives better answers than it

But maybe I’m not researching hard enough

[proper oracle]

Back when I reported a few bugs in #1140622008086970420 they'd be responded to by employees

[proper oracle]

Discord's support in a nutshell. They don't even understand the features of their own app sometimes and give you template replies that are completely unrelated to what you asked for

[rose bone]

Will probably have to test it myself than 😄 They seem to offer free trial, so I will probably give it a try if this pro mess does not get resolved in few days. 🙂

Imho they simply gave out too much year trials and are pulling the panic button.

[frail crane]

Hello, Is Perplexity limited the Deep Research queries for 20/month and it's as it used to be ?

[supple fiber]

I’m sorry if this has been asked before, I’m new here. I pay for the pro version but got told I reached my daily upload limit, even though pro is suppose to be unlimited. Is this a common occurrence? The unlimited upload was the only reason I paid for the account

[naive sage]

The limits for file uploads do appear to have been recently tightened up a bit, I believe the new limits are 10 uploads per day, and 50 uploads per week. Connectors to things like Google Drive may allow you analyze more, though... (the global file limit for things like that is, I think, 500, according to what the API shows)

[visual lodge]

plot twist perplexity is vibe coded

[orchid trellis]

Only one way to find out...

[supple fiber]

So “unlimited” is false advertising then. Cool cool cool.

[orchid trellis]

600 per day 🥹

[meager vault]

.

[distant aurora]

@green crystal opus tripping

Intro paragraph contradicts last paragraph.

[green crystal]

Agi canceled

[vagrant scarab]

Has it gone back to the way it was before?

[autumn raven]

Said 20 queries remaining today 3 days ago but still not reset the limit

Ridiculous

[devout sonnet]

buy max

[ionic swan]

Hey there, do you also have a video generation problem with the Pro subscription? Thank you

[hoary python]

for all my homies that like spaces

[sharp frost]

There is max only model council huh

[sharp mantle]

hey all

anyone have any tips or tricks to utilize p for stock research

[glacial anvil]

You can try my prompt (V3) if you are a Pro user🤔 https://discord.com/channels/1047197230748151888/1469266245890212084

The downside is that you need to manually perform the operation four times (A longer execution time for each stage), whereas the official implementation uses a parallel search architecture, while mine can only simulate it sequentially. The official version may have a higher search quota or backend optimizations, and the synthesizer might also have the capability to perform additional research.

[midnight storm]

what is the file upload limit in pro subscription and where can i see how many uploads i have left?

[glacial anvil]

https://www.perplexity.ai/rest/user/settings
Look at the upload_limit item.

[midnight storm]

thank youuu

wait theres a "max_files_per_user": 500?

so users can upload only 500 files?

[glacial anvil]

By the way, from what I’ve seen, it looks like one credit is restored every four hours. You can test it yourself to be sure.

[glacial anvil]

Actually, I'm not sure. It always shows 500 in the stats🤣

[midnight storm]

ahh

i see

ive hit my upload limit and im strugglinggg

[gilded ice]

Has Perplexity changed its Pro account features? I remember it being very different before.

[midnight storm]

IKRRRRRR

[glacial anvil]

Wait 4 to 5 hours and check if the upload limit has increased on the stats page, you might regain one quota to upload the file again.

[midnight storm]

somethings weird with pro

[midnight storm]

ahh, is there a way to delete existing files so i could upload more files lol?

[glacial anvil]

It seems like there's a limit of 50 uploads per week rn, but as I said, you might get one slot back after a few hours. If you want to test your assumption, you can delete a file from your space to try it out🤣

[midnight storm]

yeaa im about to

loll

[glacial anvil]

Please tell me the result after you test🤣

[midnight storm]

defooo xdd

[untold flume]

How to set instructions to perplexity?

[glacial anvil]

#1447189884329791680 message
I even tried using a prompt to merge multiple images into one so I could ask different questions and save my quota.

[glacial anvil]

Actually, you can just ask Perplexity. A lot of simple questions can be solved by using AI rather than waiting for others to answer.

[topaz panther]

I am a pro subscriber, but can't create an api key, do I need to pay separately for api token funding??

?

[grand dagger]

When is comet coming to ios

[crystal jackal]

which is the best model to select for general prompt? i don't understand how different models work? can anyone help me understand

[grand dagger]

Choose best

[naive sage]

I agree. Because you have extremely high limits with 'Best', it's usually best to try that first, and if you're not satisfied even after some followup questions, you can try a reasoning model (those are limited to up to 200 per week on the Pro plan).

[crystal jackal]

I am not too concerned about the limit, rather I want a well researched response

[naive sage]

I've mostly been using Gemini 3 Flash with Thinking lately. Even though GPT 5.2 seems to give longer answers I've noticed, I find Gemini 3 Flash's explanations of things more approachable and understandable. Kimi K2.5 is a newer one that I haven't used as much, but people have been saying it's pretty good for thinking as well. Claude's models tend to be good for prose (so when I'm exploring different lyrics or styles I sometimes prompt Claude), or for code... but I don't use perplexity for coding too often, and when I've compared in the past I actually like 5.2 for code. That's a high level rundown of my current thoughts on them.

[sacred gale]

I have a noob question, did they change the personal pro package? I can't seem to use the Deep Research mode now. It's prompting me to do a Enterprise plan. Or I just hit the limit?

[naive sage]

Yes, Deep Research was recently limited to 20 per month (somewhat akin to OpenAI's limits), but is also getting a much improved model to compensate: #📢│announcements message

[crystal jackal]

seems like that took some effort to put down, thank you for that. Just to take this line of thought further have you found that when you choose "best" model it chooses the models you mentioned for the tasks you mentioned or have you not thought of this. just asking

[naive sage]

I am not actually sure how to tell what model was selected in a response from that mode.

I've seen it perform reasoning, though, sometimes - so I know it can fluctuate.

[crystal jackal]

same even i have noticed

also do you have some sort of prompt engineering that you use for specific tasks?

[naive sage]

Kind of... I think I type more than the average person, so I don't have set templates that I use for that - I mostly do it naturally, following a 'background', 'question', 'examples' pattern where appropriate. But I do tend to ask another model to generate a better prompt before I run Research queries (because they primarily benefit more from more complete prompts). Some of those prompts have been pretty long, but I don't use a set template for every one.

So I make sure to ask another model to construct a better prompt suited for Deep Research. I usually ensure I ask for additional insights and stuff like that. It usually fills out more aspects of what I'm asking about, which can help Deep Research be more thorough when it plans the queries.

[glacial anvil]

But you need to confirm that using the 'Best' model doesn't eat into the 200-per-week reasoning quota first🤣

[crystal jackal]

also do you mention the type of sources that you want the model to look into. for instance when i want to find genuine reviews of a product that i want to buy i ask the model to look at reddit or twitter. But i can't seem to find the perfect prompt for it like i want the model to look at the most top voted comment and then analyse that with other sub reddits you know like how we would research. Do you understand what i am trying to say

[naive sage]

I don't think I've ever done that, I like when it references official documentation from companies, but it seems to do that properly on its own pretty well. I don't think I've ever prompted specifically for reddit or twitter.

So just keeping 'web' on and asking the question naturally seems to work out pretty well from my experience anyway

[crystal jackal]

the limitation with 'web' option is that it limits to mainstream articles which are mostly all partisan. Therefore, not giving an objective perspective

[naive sage]

This probably depends on the search index itself. Perplexity's 'biases' from search would be different from other providers (they don't just 'search google' like some people suggest online, that's not how Perplexity works). One way to combat that is to challenge it with prompting for opposing views and adopting a debate-style format to respond, or compare and contrast pros and cons of different views. It is important to be cognizant of how many people hold a view, where the views are prevailing (it can shift from region to region), and try to understand why the view is held as well. Though there is a tendency in society to be overly-politically-correct in a way that tries to acknowledge every view for the merit of basically just having a different view. Being informed is one thing, but trying to shoehorn views from one region/context/situation to every region/context/situation is not really an advisable thing, unless those views are actually relevant/realistic/informed.

So at the very least Perplexity can give different perspectives on answers than one might find on Google, as a basis for comparison. But I do think prompting is really good at enumerating alternative views, with details, when prompted specifically for that.

[rocky yarrow]

Did they fix PPLX yet?

[paper marten]

I have just signed up to Perplexity Pro but it is prompting me to use Max at £200 per month if I want Deep Research

Why is this?

[quasi wolf]

I just paid for pro and I'm getting only 20 deep search queries per month is that normal?

[naive sage]

Yes, Deep Research was recently updated with a new limit to 20 per month (somewhat akin to OpenAI's limits), but is also getting a much improved model to compensate: ⁠#📢│announcements message

[quartz lark]

So i should just use Google, got it. Later

[young hollow]

Gemini is Calling Me

💀

[fossil wyvern]

did perplexity add mcp to windows app ?

[sharp frost]

Hey anyone know if I can just use labs/create for deep researches?

Since I rarely use labs

[vagrant scarab]

I have QWEN, I'm using it now and it's good too, it has different search styles, it even automatically generates PDFs on the first try with in-depth searches.

[sudden nova]

just gonna post this on general for more awareness

This isn't a bug report, rather me complaining about a supposed 'change' in the pro subscription.
I'm extremely confused by how I seemingly ran out of image/file uploads under a Perplexity Pro subscription. Under a Pro subscription, isn't one of the perks 'Increased file and photo uploads'...?

I've been subscribed for 2 months now, and have never ran into this issue previously, even though I have been seemingly able to upload 'infinite' files before. Not like I can even confirm this, because rate limits/file limits are not disclosed due to 'operational reasons'.

I've read some recent threads regarding new updates, one of which alleges a recent degredation of the limit of 'deep research'/queries you can send in a given week/time period. This is pure shady behaviour, at least communicate to a paying customer about changes in the services you provide ahead of time so that a customer can either accept the changes or cancel their subscription before the changes are rolled out.

Furthermore, when prompted to upgrade to a max subscription, you advertise a '$167/month or equivalent when billed anually' - although immediately after show a prompt for $220 a month..?

Additionally, you advertise 'unlimited file uploads' on iOS/app versions. Pure false advertising.

Why are your policies and perks for subscriptions so vague? I can't even get into contact with support even through emails other than a AI chatbot that spews the same information. You're only losing trust in your business by doing this.

[vagrant scarab]

Try using qwen, it's free.

[sharp frost]

Perplexity is a small company /s 😜

[solar tulip]

I like perplexity a lot...but within 10 minutes of searching my next prompt was to search for alternatives. Not that I'm lazy ore search always for the ceapo's. Not at all. But in this age of time communication is key. It gives me the impression that there are enough users to "pull some of these stunts". Besides that... give the counter on labs back. Again... if this was all on purpose to be vague as a strategic marketing campaign... job well done!

I will stay with Perplexity for now... because I also believe in "learning on the job". I'm also not perfect. But I will spent 60 min more to plan my exit strategy if there is no clear information coming back from all these feedback I'm reading. If I missed something... and it was not vague. My bed... still these people like me should have the opportunity given (read: also serve the simple user not reading every update that is communicated) to find clear info (if any).

Perplexity... keep up the good job.. I really like the product (no sarcasm!) and love to continue... because every tool has it pros/cons. That is part of life.

[paper marten]

I have just signed up to Perplexity Pro but it is prompting me to use Max at £200 per month if I want Deep Research - there is no support for Perplexity Pro users - just an AI robot. Sadly I am cancelling my membership. Their service for Perplexity Pro is very misleading

[sudden nova]

ik ur joking but there is no excuse for companies to engage in blatant false advertising and deceptive practices

[paper marten]

I am cancelling my subscription

[sudden nova]

i've noticed issues with tons of other ppl as well with pro subscriptions as well

and i will be reporting this to my governments consumer competition and to apple directly if this isn't resolved in a timely manner. seeing this many people upset is unacceptable, especially for something this expensive.

[sharp frost]

Go do it, maybe it qualifies as a bait and switch scheme or something.

Not everyone lives in countries that has robust consumer protection laws

[young hollow]

Qwen is Best Free Chinese Ai

You can also Try Kimi, GLM, Deepseek, Mimo and Stepfun

[vagrant scarab]

I've used Kimi and Deepseek before, and I liked them, but Qwen is really good with the new features; when I first used it, they were just starting out.

[sudden nova]

i'm australian so i have the ACCC and fair trading to report to.

[lavish lantern]

I've been using gpt 2 for a while , try it out it's great

[crisp kestrel]

they informed us

[visual comet]

I use qween (3_4B) for clipping in ComfyUI workflow, but as AI I prefer Kimi, it's far better as general AI, just my two cents

[visual comet]

seriously? what are classified as "advanced" models?

[crisp kestrel]

maybe the research mode

[young hollow]

gpt 2?

[young hollow]

True

[hushed hare]

How to I get the Pro role on this server, bcs when i joined this discord server I didnt' buy yet

[short narwhal]

@visual comet Advanced models are GPT-5.2, Gemini 3 Pro, 3 Flash, Sonnet 4.5, etc., so anything you choose in the model picker besides Best afaik. Research has its own limit, which is 20 a month, with it regenerating usage around one every day or two, from what I can tell

[short narwhal]

#💬│general message

[hushed hare]

thx

[coarse girder]

BRO AM I ONLY ALLOWED ONE DEEP RESEARCH CREDIT PER DAY?

[whole fern]

I got blocked for the entire month: #⚡│ask-community message

[fast pasture]

#💬│general message

this issue here resolved with this :

https://discord.com/oauth2/authorize?client_id=1469616969845178378

I coded my own lol

[crisp kestrel]

We complained enough they listened!

[coarse girder]

explain

[crisp kestrel]

They finnaly added a model picker to the assisatnt in comet

[coarse girder]

hmmmmm i akways had that, I am more hell bent on deeo research

[fossil wyvern]

still waiting for perplexity to let us add our own mcp to the windows app

would be so fucking huge

[thorny gyro]

Hello, like many others I hit unepexctedly the limit of deep research on my pro plan.. if I would have known the limit changed at least, I would have used the feature only when really needed. Any alternative ?

[livid apex]

Hey guys, I hope you don’t have any major work depending on this trap we’re falling into.

[thorny gyro]

Do you know what is the limit now, for an enterprise plan ?

[wanton crest]

No, but give it 6 months and it will be one search per month.

[old sail]

its now white circle not blue !!

https://klipy.com/gifs/funny-meme-Va9

[sharp frost]

#bringbackblue ?

[young hollow]

Yeah Proof of Degration

[vagrant scarab]

It's a shame it's expensive, I opted for Gemini, but I use Qwen more often. I hadn't used it in a long time, it's much improved, and the varied search methods are very good, plus it's free.

[austere crater]

True

Pro = Free tier
Blue becomes gray lol

[marsh acorn]

Hello everyone. At least, i see that this "deep search limit" is not a bug on my side ^^

[tender coral]

5 uploads a day wth ,its even less then two days before 🤣

[naive sage]

It's reset weekly - 50 per week.

So they don't refresh as often as the daily ones.

It's up to 10 per day, but still within the weekly limit of 50.

I'm not sure why it would show as 5 for your account, if you are on Pro. It's supposed to be 50.

[tender coral]

oh okay, thanks for clarification

[naive sage]

I think using the Google Drive connector is one way to analyze more files.

It may not count as uploads when queried that way.

[tender coral]

well yea, im on pro but i think i just used my oploads on this week

[mental dew]

I just canceled my subscription

[tender coral]

huh, curious

[agile basin]

yup, me too.. kinda relied on the deep research thing and now that it's not available anymore (by that I mean heavily limited), perplexity does not make sense for me anymore

[mental dew]

Yall should do the same

Imo

😭

[naive sage]

Have you found any good deep research alternatives with higher limits? Curious to know what people are migrating to. I've heard people suggest Gemini or Qwen. Gemini certainly has its own pros and cons.

[distant aurora]

@young hollow ngl, scira ai looks like a good alternative to perplexity atm.

They just need to market it.

[agile basin]

That question is still in the air for me. Another option is just to learn how to live without ai again lmao. Or even better, build my own AI rack. The latter seems kinda impossible tho

[naive sage]

Given compute cost right now, yeah, it doesn't seem as viable as it did even a few months ago.

[agile basin]

Also I'm wondering about T3chat, if that is worth something

[distant aurora]

https://developer.nvidia.com/embedded/buy-jetson?sharetype=link

Bruh.

https://www.maginative.com/article/nvidia-introduces-jetson-orin-nano-super-a-249-compact-generative-ai-computer/?sharetype=link

Your own ai rack. There.