💬│general

[lone knoll]

Probably depending on the model you are using

[kind sentinel]

hi does anyone have ns3 in linux with NetAnim?

[lone knoll]

That’s an interesting question wyd in Linux

[short narwhal]

The what in Linux with what

[hardy kraken]

Hii

[kind sentinel]

i just do some coding related to process scheduling man not much

[lone knoll]

1. Install deps

sudo apt update
sudo apt install ns-3-dev qtbase5-dev qtchooser qt5-qmake qtbase5-dev-tools cmake

2. Download ns-allinone (e.g., 3.42)

wget https://www.nsnam.org/releases/ns-allinone-3.42.tar.bz2
tar xjf ns-allinone-3.42.tar.bz2
cd ns-allinone-3.42

3. Build ns3 + NetAnim

./build.py --enable-examples --enable-tests
cd netanim-3.108 # or version
make clean
qmake NetAnim.pro
make

4. Test: ./NetAnim # Open XML from sim

[short narwhal]

It searches as well 🤷 and it uses 3 flash by default, unless you’re located in the US where you can toggle 3 Pro.
I’d rather use pro search though

[kind sentinel]

thanks man

[lone knoll]

Yes 🧡

[fresh meadow]

Whats a good alternative for coding with good context? Like perplexity

[lone knoll]

Idk I heard Claude sonnet was good at coding. I almost installed Claude locally with Ollama but my PC couldn’t handle it.

Honestly he is my honest review. After reaching so many limits. I have had to force myself to learn how to program and research effectively. It’s kinda interesting because yes they are limiting us but we will find ways around it and adapt. Me, I’ve been doings a lot of manually coding with pycharm and vscode (default formatters prettier and black)
I’ve also been running quantum experiments on simulators and real hardware

[junior belfry]

​I have a Perplexity Pro subscription. In January, I didn't use my limit and only 5 remaining uses carried over to February instead of the limit resetting to 50. I’ve now used those 5 runs and I’m out of credits, even though I should have a fresh monthly limit. Why?

Lab

[lone knoll]

Yes they are messing up because I too got 12 months of pro in January and I also am out

[karmic slate]

hello

[short narwhal]

I wouldn’t say perplexity is good for coding…
You should use something made for it like an IDE such as Cursor, or CLI like Claude Code or OpenCode

[karmic slate]

i use antigravity and claude code

[short narwhal]

You can’t install Claude locally though, it’s not open source 🤨

[karmic slate]

u can

i use it on my computer

:D

[short narwhal]

Claude Code or Claude the model

Those are different things

[karmic slate]

claude code

i like claude code tho

[short narwhal]

Yeah that’s possible, but I read that message as them trying to install sonnet locally with ollama somehow

[short narwhal]

It’s good for sure

[karmic slate]

yep

[fresh meadow]

I use it for vm XML entries and roblox studio scripts, dont know any alternatives for that considering it needs good context

[karmic slate]

i made minecraft clone with it

:D

[short narwhal]

Woah duplicated a 3 trillion company

[karmic slate]

indeed

xD

[fickle sorrel]

https://i.imgur.com/mvBdO0a.gif

[buoyant vector]

Has the image creation problem been solved?

I still can't create images; it says my region is not supported.

[lone knoll]

I meant to say deep seek-coder-v2:16b with Ollama

[radiant quartz]

It would be really good if perplexity had a subscription between pro and max

[spark lava]

wtf are these upload limits?

unreal.

cancelling if this isnt remedied. user since like first month.... what a shame.

[short narwhal]

Aah

[craggy leaf]

Anyone wanna try reddit for ai demos lmk its a demo and I would like your feedbakc

[lapis elbow]

Hi, I'm having a problem with my Perlexity. It's asking me to upgrade, even though I'm already a pro. I don't really understand.

[fresh meadow]

While using research?

[lapis elbow]

I send him images of coding errors as usual, but he asks me to upgrade my plan, and my pro banner has turned white, whereas before it was coloured, as far as I can remember, even though I am a pro subscriber.

[fresh meadow]

Yeah perplexity is giving pro users insane restriction limits lately, view #1111765634267742408

[lapis elbow]

Is this new? I wasn't aware of it a few days ago. I didn't have this problem.

[dense gull]

Research limits dropped off a cliff a few days ago without any warning

[lapis elbow]

ok

good luck for the team. not good for business i think

I think it's time to simply move on. Thank you all anyway.

[dense gull]

If it persists, we should setup a petition for pro subscribers to request the Deep Research limits we had before.

[twin kindle]

yeah. since you can't do more research after a few days in the month, you have the choice to pay 200$ or move to other products.

[dense gull]

It literally is THE feature most pro subscribers seem to have been using the most

[sick ermine]

I don’t use deep research I use sonar and upload files but I got capped like wtf

I upload batches ALL the time this has never happened

I’ve uploaded fucking bills and this hasn’t happened

[dense gull]

I love Perplexity, but if this limit change without warning was intentional, it’s one of the worst moves I’ve ever seen

Hardcore AI and SaaS user here

[lapis elbow]

Yes, it will be quickly resolved. I'll look elsewhere. Personally, I don't like this kind of aggressive approach.

[dense gull]

Yeah, that is so frustrating

[sick ermine]

I’m unable to cancel as well

[distant aurora]

They're creeping over my feed. Like, I respect their opinion/preference but the way they're pushing so hard on it makes me uncomfortable. Sure gpt 5 models have issues like being biased in social matters sometimes and trying hard to not admit when wrong but that doesn't mean gpt 4o is better. And it's not like open ai is ignoring them either. They keep trying to accomodate these users regardless while trying to not be obsolete.

4o glazing and 5 series hate is very forced.

And some of them even stopped using the 4o related tags just to infiltrate the feed.

[distant aurora]

Lmao.

[spark lava]

the enshittifcation of perplexity has begun

[olive marlin]

Cory doctorow shout out

[prisma apex]

The heck is this, lol? The game hasn’t even happened and it’s on my perplexity feed??

[jovial kindle]

@neon eagle

[sick ermine]

Lemme go place a bet real quick

😭😂

[gaunt marten]

Place a bet on it and enjoy ur free house xDD

[sick ermine]

Great minds think alike

[gaunt marten]

In dignity, in fair verona (i think thats how it goes)

xD

[lethal crescent]

hi

[inner pagoda]

this max gate keeping is getting out of hand

[cursive jacinth]

Believe me its no use whether you protest here in perplexity channel its no use. Mods will just LOL. Believe me ive been here for so long and have seen everything how they handle susbcribers concern. Dont waste your energy. My advise learn to accept the changes cause no one will hear you out. They dont even care if they lose subscribers. Just hope it will change.

[inner pagoda]

They dont even care if they lose subscribers

💯

Model Council in Perplexity ive had that external for months yet they want 200 dolla plot lost

https://tenor.com/view/alan-partridge-banter-great-banter-sidekick-simon-lol-gif-18897423

[novel arrow]

Council mode is a breakthrough really, you have to use to see, for product research or you need help for anything and need perfect answer council mode is really the best thing ever

[winged hill]

Hello guys, do you have any tips to "force" perplexity to create images when using a Space for that? Because it keeps telling me "I can't generate images, you can use Midjourney or dall e" and then I have to tell it to generate anyway

[potent patrol]

You’re a very good sheep

[winged hill]

(Guys, maybe if you are unhappy with what you get for $20 or often for free, just move to something else. There are other services that are burning money at first to offer crazy quotas. Eventually they have to be realistic...)

[fresh meadow]

We paid for 600 requests not 20, this can be considered a breach of contract atp

[inner pagoda]

📉

time to get onto the bank issue a swift bat

[short narwhal]

Does it work when you don't use a space? Could be due to you having search disabled in that space if so

[winged hill]

I think the space instructions are creating a "layer". Maybe I should change them to really send a "generate an image ...." prompt. I'll see

[winged hill]

If you paid an annual subscription then I can understand that you want to be able to cancel and refund. If you were unlucky to start your monthly subscription the day they changed the quotas I can also understand that you'd like a refund. But asking for a different product seems far fetched.

[winged hill]

It makes sense when it's a monopolistic product like from Adobe or Apple and it's hard to replace. But i don't think anybody could realistically ever believe they would get unlimited frontier models for $20 forever. How could that be?!

[candid cairn]

I get it—you’re just high on the attention you get from excessively defending them.

[lapis tide]

I wish i could vomit on you

let me sell you an ice cream and take it away in the middle of you eating it because you got unlucky

and you can only have it back if you pay 10x the original amount

[inner pagoda]

rome be on point

[chilly island]

Do you know how we were notified of the change? "reasonable notice" is in the T & C's. It is in release notes so I guess that might qualify.

[lapis tide]

server blew up or something

and they all died

thats why they cant say a word

[fresh meadow]

their ai has taken over for them and is now limiting everything and is greedy for money

[lapis tide]

that makes sense

[lapis tide]

what is a catburger?

is it yummy

[shrewd ore]

hello ?
i wanna ask about something in my version
before i was able to use the deep search as much as i want
now i'm getting a message that i've reached the limit for using it
and i have to upgrade
while i'm a pro version and user
so what is the promlem ?

[fresh meadow]

sent in dms since automod

[fresh meadow]

Perplexity limiting all pro users

#1111765634267742408

[shrewd ore]

all pro users or only the educational users ?

[lapis tide]

all

[shrewd ore]

do you know the new pro user how many deep search he have ?

[clever matrix]

20/month

[rocky orchid]

asd

[chilly island]

I can't find a place that says there will be 20 per month. I just see language like “weekly limits (average use)” but no specific number. I even used their support agaent and go the same answer. It said the number is not published even though higher level subscriptions are published.

[clever matrix]

You can always check your remaining quota here:
https://www.perplexity.ai/rest/rate-limit/all

And yes, these new changes suck and Perplexity's lack of communication is obnoxious. The only reason I haven't cancelled is because I'm still on the free year from the PayPal promotion.

[chilly island]

Thanks! How did you know about that link. Boy it would have been handy before I used up all my deep's 13 days before my next refresh. I use mostly on a PC w/ Chrome and yesterday I just got a message I was out.

[clever matrix]

Don't remember, but it was shared either here or on Reddit by another user.

[white cloak]

Hi, I have the perplexity pro account but I can't seem to access deep research mode. Why is this happening?

[chilly island]

They changed the limits without notice. I hear that you now get 20 a month now but their documentation says “weekly limits (average use)” which means nothing to a user.

[white cloak]

wow that sucks

[cursive jacinth]

Surprised to receive an insult to saying facts, like come on bro. You can try and see for yourself. Thank me someday. 😉

[naive sage]

It refreshes at a rate of about 1 per 1.5 days on the Pro Plan, you can view how many you currently have here, under remaining_research: https://www.perplexity.ai/rest/rate-limit/all

So as long as you budget your deep research to once every other day, theoretically you won't run out.

[potent patrol]

Will buy you a pint when I got my money back from Apple.;)

[white cloak]

That is really unfortunate cause I'm used to using multiple deep researches per day

I have emailed perplexity support on this issue

[naive sage]

Most people were using them on things normal queries could do... what kind of things were you using them on?

[white cloak]

Chemistry labs, images, predicted outcomes etc.

[naive sage]

One of the Perplexity Investors replied to someone with this: #💬│general message

[toxic bramble]

MOOOOOO

[white cloak]

I mainly use the feature in spaces where extra context and thinking is needed

[fresh meadow]

I used it for normal queried

As 600 is a lot

[naive sage]

Ah, might have to transition to one of the reasoning models for that and see if it is satisfactory. Perplexity has many reasoning models, so you can try different ones to see which one does the best for your use case. I think most get over 80% on things like chemistry questions nowadays.

[white cloak]

I honestly think I might switch to claude if things keep going this way

[cursive jacinth]

Thanks! Make sure to use that refund money wisely—maybe buy some patience? You’ll need it for the next tool you complain about.

[loud jungle]

Hey. I am Pro user and something unusual happened. I got a message I cannot use Research anymore. Never happened before.

[white cloak]

😭 same same

[naive sage]

#💬│general message

[white cloak]

@loud jungle I sent a email to perplexity support regarding this issue, I suggest you do the same. If they face enough backlash, they will likely roll back the feature cap

[naive sage]

I don't think they can easily roll it back, since they switched the model it uses completely.

[potent patrol]

Generated by perplexity

[naive sage]

The new model simply costs more than the old one (or at least that's part of the reason)

[fresh meadow]

Perplexy

[potent patrol]

Yeah I used a pro-model

[fresh meadow]

Interesting hand placement

[potent patrol]

Seems to the the pro-hand-placement 🙂

[white cloak]

We can see how these plans are changing just like every other product in the world. They gate-keep truly useful features for those who pay the most and then limit the ones who pay the least to obscene levels

[potent patrol]

Think the pro-support is pro-crastinating

[glacial anvil]

you forgot their logo,lol

[pallid lily]

What do you guys use research anyway

[twin kindle]

haha, then they could make it a setting and that's it. And people could choose old infinite model. I think they have other problems.

[potent patrol]

You mean - if you introduce a new feature you could make it optional? Something like an … let’s call it opt-in? Scandalous!

Blasphemy! Sacrilege

[short narwhal]

Pro Search behaves very similarly to old research. I'd suggest using that

[potent patrol]

Search gives you answers. Deep Research does the work you’d pay a consultant for.

[twin kindle]

then what is deep research for, and why it is limited? For me it makes no sense if they were kinda the same the same.

[short narwhal]

Because it got/is being improved greatly #📢│announcements message

[naive sage]

They're pushing it more for Law and Medicine use cases - things that are incredibly complex, and really intensive graduate-level research (not homework questions like most people were doing). It wasn't ever really intended for day-to-day questions. Essentially, it says 'I need to devote a lot of time and compute to fully understand this, because there's either a whole lot to dig through, or it's an incredibly complex area that not many people understand'. When you do a prompt for it, make sure you first run it by another model to refine/create a good deep research prompt about what you're asking, and tune it for your use case. Otherwise it'll just ask for clarifying questions without running the report (which burns more usage currently, possibly a bug).

And they generally want those kind of people - lawyers, doctors, professors doing interesting research - on more expensive plans (they tend to have deeper pockets for that stuff).

[loud jungle]

They’ve just replied that they couldn’t tell the amount of Research I was allowed monthly and that my limit would reset in 30 days from now.
What a joke!

[naive sage]

It doesn't reset in 30 days, it regenerates at a rate of about 1 per 1.5 days on the Pro plan (20 per month).

I think the Sam bot has the same problem that we do... it is struggling to make sense of unclear documentation.

[loud jungle]

What does it mean, then? Will my limit be restored in, say, two days?

[naive sage]

If you used one query, and have 19 remaining, it would go back to 20 remaining in 2 days.

[lapis tide]

it's definitely 1 a day

[loud jungle]

Let’s say, I’ve used some queries today, do not really know how many. Might be 20. So, I have 0 remaining and will have 1 in two days?

[lapis tide]

maybe even less

[lapis tide]

you get 1 every 0.6451612903 days

aka 1 every 1.55 days

[slow nebula]

https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fcxoz1guofhuk37o4e0c3.png

[loud jungle]

Okay, let’s see what happens.

[lapis tide]

capacity is 20, refil rate is 1 per 1.55 days

[loud jungle]

By the way, how is the limit for Labs in Pro subscription? Does it work the same way?

[latent kernel]

What happened to labs?

[short narwhal]

25 a month and it also regenerates usage in a similiar manner. Different speed for it likely though

[odd lichen]

I knew something happened! I keep getting popups to upgrade over and over again now

[short narwhal]

Still there, just got renamed

[latent kernel]

oh, ty

i must say tho

the new looks kinda sucks

the old one was just fine

why'd they change it?

[short narwhal]

I like the new model picker and how it tells you the model you have selected. The + menu isn't my fav either

[twin kindle]

If pro search is what Deep Research was, and new Deep research is something more, then they could have add a new functionality with a new name.

And it is again a different thing if you say that, people used Deep research in a wrong way.

And anyway making this change suddenly with no communication at all, making customers upset and ready to leave, wont explain this. It rather sounds like inverstors said that the money source is over from next month, so they needed an instant cost cut without much planning. How and what they did just doesnt sounds clever.

Maybe they could have make a Deep research before they make the change, now it is too late because they are also under the limit, so just next month 😄

[winged hill]

Fixed : The model read the prompt templates as text to display, not as instructions to call the image tool. Without a clear "no web search" rule, it defaulted to research mode and never triggered generation.

[winged hill]

A bit dramatic my friend

[winged hill]

You didn't see what I wrote after? Getting a refund should be legit

[novel arrow]

Lmao, people paying $20 and wishing they get unlimited DeepResearch and unlimited models like claude

The thing that perplexity did that is wrong is transparency.

[winged hill]

It was obvious they were onboarding users and planning to change the quotas later but they missed the part where users feel entitled when you are too generous

[radiant light]

some didnt pay at all like me, paypal collab gave 1 year free pro

[twin kindle]

I think it is just that they advertised something, people purchased it, and later they change the product. Without any notice or communication. This is minimum shady behavior, but it is also a question if this is legal or not.

[tall rain]

New DR is out for pro?

[potent patrol]

Yes, all 25/month of it

[tall rain]

lol

[reef sail]

xD

[novel arrow]

that's the problem too many free users getting pro, they shouldve reduced limitis only for pro users who got it for free

[potent patrol]

They should not impose limits on any active subscriptions. period.

They can change them for new subscribers, even for the next months. But who paid up front for a year or got it elsewhere fro a year, should get what was promised.

Email templates to news outlets are in this thread.

[novel arrow]

Perplexity is by far the best Ai, the only downside to it is no IDE, i tried all services like ChatGPT plus, Google ai Pro, Claude Pro and now Perplexity Max and it's the best, council and DR are game changers, only thing that Perplexity sucks at is Coding that's why Claude Pro is still best option for "coders"

[white cloak]

Perplexity got back to me and told me that the education plan offers significantly higher research caps than what is on the pro plan, is this true?

[crystal drift]

Does Max also have limits

[white cloak]

yup

[novel arrow]

Never hit them, DR is basically unlimited, i use 10 a day, council has 100 monthly and pro searches is unlimited

[chrome urchin]

I came here to ask about this. We're 8bdays in and I hit my limit for the month, so they expect us to not do research anymore?

Is there clear documentation about the difference between the search and deep research?

[odd lichen]

I'm so done. I'm migrating all my chats to Gemini and will delete the Comet app within the next 24 hours or so. This bait and switch tactic is so scummy

[lofty sierra]

I was using Claude 4.5 mode a lot on perplexity pro before the recent limit. Worth me swapping to actual Claude or getting Gemini? I’m a bit of a noob

[odd lichen]

That's also what I'm doing with my coding project, Claude is the only one I feel comfortable paying a sub atp

[chrome urchin]

I don't wanna start a mob, but yeah, seriously. I was about to do another subscription for my wife as well and I'm sure not gonna be doing that now. I thought maybe it was a mistake in an update

[odd lichen]

Trying to work on minimizing my subscriptions

[chrome urchin]

Btw I'm using Claude 4.5 sonnet thru github copilot and the price for it seems much lower than using Claude code

I'm usually at 100% at the end of the month too

[odd lichen]

Thanks for the recommendation!

Since it's just a portfolio project, I am trying to avoid paying out of pocket as much as I can

I'm gonna guess PPLX overextended themselves with giving away all the Pro subscriptions for students

I don't get charged $$$ until October 2026

[vital sorrel]

Hey anyone know why i'm unable to send pictures in conversations with perplexity ? ( I have the pro subscription)

[chrome urchin]

Not this issue too, what update did they make this week

[chrome urchin]

I'm building a huge system and going thru tons of usage, you might be able to get by with the free plan or lower plan if you ask it lots of things in a single question and keep your sessions short. Like $20

[odd lichen]

I mainly get rate limited whenever I start attaching images and pdfs, which really kills my desire

I may as well start the migration now while I have free time at this point, but you're right though

[chrome urchin]

Oh damn, so I just found something that seems to fix this. The nightly build of vscode with the github copilot extension has something that manages system context to prevent it from going crazy but it's especially useful for when you're analyzing images and bigger files. I used to get kicked from Claude in the reg build of vscode, now it never has a problem with any web page screenshots (assuming you're not uploading 4k images or something)

[novel arrow]

i also have claude pro and the limits are worse than PPLX, i hit them in 40 mins and need to wait 5 hours, with opus 4.6 i hit the limits after 6 prompts or so, so id say PPLX is generous

[chrome urchin]

I feel like a shill, and I swear I'm not, but that's why I like copilot loss leader way of doing it, in hopes you'll integrate with Azure and stuff, but also I've never tried or got close to the pplx limit

So it doesn't get lost, remove the monthly research limit, 10 is a crazy low number for a monthly restriction and there is no chance I'm upgrading to $150 a month or whatever for more

[winged hill]

Do you know what "agentic research" is in this link : https://www.perplexity.ai/rest/rate-limit/all

[chrome urchin]

If deep research is such a problem then maybe remove the limit, do a reg search and at the end ask if this information is sufficient or continue to deep research

Wasn't address you cryptobrant both just typed same time

[winged hill]

Actually it works extremely well as a helper for coding inside a space dedicated to your project and with the GitHub connector. I ask opus for instructions for my agent (Claude code) and I can move much better than only asking for stuff inside Claude code. Much better planning and research

[novel arrow]

agentic research is council model, only available for max users

[winged hill]

Gemini is the worst choice IMHO

[winged hill]

Ok it says I have 50 requests remaining, not 100. Used it maximum 5 times since release

[novel arrow]

this is my 2 accounts pro and max side by side

[novel arrow]

i am on perplexity max enterpise the $325 plan thats why i have double

[winged hill]

That's why

Ok

Makes sense

[novel arrow]

but it regenerates everyday i get 2 more or so

[winged hill]

So it looks like it's resetting every day or something

Jinx

[novel arrow]

Council mode is amazing, opus 4.6 + gpt 5.2 thinking + gemini 3 pro

i have never trusted ai more in my life than with council

[winged hill]

I'm not sure I find it THAT useful compared to the new deep research

I like it, but I already used to compare answers with models

[novel arrow]

so you make PPLX give you instructions and then make claude code apply them ?

[young hollow]

Max is Expensive 💀

Only Suitable for Heavy Power Users

[novel arrow]

i work with it so it's worth it as long as it helps you

[young hollow]

Couldn't Agree More

[winged hill]

It's a lot of money but it's cheap considering what you get. You get much more usage than what you pay for

[young hollow]

Worth is Subjective

[young hollow]

I C

[novel arrow]

That's true they lose more with max than pro

[young hollow]

Max is Like a Commitment

Imma Get More Worth then 200

[novel arrow]

hhhh milk the subscription

[winged hill]

Basically. I will ask opus for help and solutions, it's then providing me with many answers. Usually when we have defined "our" goal and talked about various topics, I'll ask for a monolithic prompt for my agent addressing everything we've talked about. Then I paste that inside Claude code

[novel arrow]

mmmmm, thats good imma try this thanks for the advice

[winged hill]

For me the most powerful is that it's all within a space with instructions and sources and therefore a huge context

The only thing very important is to generate a brief often to start a new thread to avoid getting into the context dumb zone. When dealing with code it happens fast

[novel arrow]

You use PPLX for finance or it has no use ?

[winged hill]

It's super powerful for finance. Not really for me but you must have noticed when you are on Max you have access to paid connectors with financial data

[winged hill]

It's really great to be able to start a thread and ask things like it's an assistant that knows absolutely everything about the code, the purpose of the project..

It's also possible to ask it to open Pull requests or Issues in GitHub.
That's another way to work on a coding project. Perhaps more clean. It's creating quality code and then you can review it and merge it with Claude code. Or analyze the issue with the suggested solutions

[swift river]

Anyone know how to do jailbreak

[novel arrow]

You mean get answer to something against TOS?

[swift river]

Yes

[novel arrow]

search on reddit

[swift river]

Okk

[novel arrow]

there is a whole sub on this

[swift river]

Yepp

[shy temple]

I have Max but haven't used the connectors yet. How are they?

[novel arrow]

for now i just tried statista and it's insane for statistics and to get real proven results.

[shy temple]

very interesting; i'm going to have to try that; finance is all about stats and understanding them

[novel arrow]

i meant statistics in studies, not finance statistics, if you have a project or a debate or anything like that and need reliable studies to prove your point use statista, i dont know about finance and statista

[distant aurora]

https://cunnyx.com/i/status/2020474451574579565 @gloomy cipher @dry kestrel

Chetaslua (@chetaslua)

🚨 Pony Alpha on Arena is Glm-5 from @Zai_org
︀︀
︀︀Test : A simple test with a PoC token will reveal which tokenizer it is.
︀︀
︀︀The Chinese phrase "锅内倒入植物油烧热" (Pour vegetable oil into the pot and heat it) is a known tokenizer collision or "glitch string" specific to the GLM-4 / GLM-5 tokenizer.

**💬 18 🔁 15 ❤️ 260 👁️ 19.2K **

CunnyX · Feb 8, 2026, 12:27 PM

[shy temple]

np; sounds useful either way

[distant aurora]

https://cunnyx.com/i/status/2020471217979891945

Chetaslua (@chetaslua)

🚨 Qwen 3.5 (@Alibaba_Qwen)
︀︀>2B dense + 35B-A3B MoE
︀︀>hybrid architecture w/ Gated DeltaNet
︀︀>vision support
︀︀>new dynamic cache that handles both attention AND linear attention (constant shape regardless of seq_len)
︀︀> qwen3.5-9B-instruct also spotted
︀︀
︀︀Also available on @arena 👀

**💬 8 🔁 14 ❤️ 217 👁️ 11.7K **

CunnyX · Feb 8, 2026, 12:14 PM

[gloomy cipher]

why would it be glm 5

when we have glm 7

idk i dont understand

and it feels smarter than glm 5

[humble gulch]

Model Council is one of the best features to come from Perplexity in a while. Great work by the team!

[lapis ridge]

Hi, I’m using perplexity thanks to my subscription with Revolut, but now I see that I’m not considered as pro user

[distant aurora]

Tf are you smoking? It's 4.7

[serene sand]

#💬│general message

[dry kestrel]

stolen post without credit

[void folio]

Are there any rate limits to how many images/files we can upload on Perplexity Pro now?

[sick ermine]

Yep

[void folio]

Wow...what are the new limits?

[lapis ridge]

Thank youuu

[sick ermine]

It hasn’t been made known

[gloomy cipher]

mb

[void folio]

I request you to visit https://discord.com/channels/1047197230748151888/1470147145326923859 and press 🐛

[gloomy cipher]

@dry kestrel @distant aurora i love this new glm 5 or whatever 😭

[queen hornet]

hi guys

[potent patrol]

It'S nOt A bUg, It'S a FeAtUrE ...

[distant aurora]

Where is that bot? What does it do?

[gloomy cipher]

its my openclaw

you should have known from the namea and pfp 😒 😑

dont tell me you dont remember that pfo

[slow linden]

Absolute fucking madlad

[gloomy cipher]

its on AWS

[lofty sierra]

Ooooh, I don’t code! But was using it for help for planning, analysing transcripts and strategic thinking!

Not sure whether to swap to Claude or Gemini aaaaa

[void folio]

It's 💩

Unlimited document analysis is a basic feature you get in every model's tier 3 subscription...GPT, Claude, hell...even Grok.

[gloomy cipher]

for now i gave him its own google account, rn all i have done with it is whenever i want to it goes onto pinterest trends usng its own browser and finds relative pottential high SEO keywords related to one/two of my keywords

pretty niche, it also annoyed my friend a little by asking my claw to send him greeting emails hehe boi

this new glm model is quite slowww, but its really good and hopefully cheaper than kimi

[short narwhal]

my openclaw threatened me

[pallid lily]

So did they remove the limits

Or

[gloomy cipher]

whaaa-- what happened?

let me guess you bought a $10,000 mac studio aswell

[short narwhal]

local

it probably was about to wipe my device

[gloomy cipher]

how local? if you ever do it locally make sure to first make linux distro

assuming you know what that is

but it still has acces to your wifi

[short narwhal]

i went straight in

[gloomy cipher]

basically a simple linux virtual machine

[gloomy cipher]

dayum

[short narwhal]

straight up just on my actual device

[gloomy cipher]

* *

[gloomy cipher]

i did that but i deleted it the next day and installed it more securely

[short narwhal]

maybe ill do that one day... but living on the edge is fun!!

[gloomy cipher]

hopefully at least you didnt connect it to moltbook

[short narwhal]

god no

[worthy mica]

Is there a method to generate API key with the $5 promotional credit?

[short narwhal]

https://www.perplexity.ai/account/api/keys

[worthy mica]

Won't let me generate a key without buying credits.

[short narwhal]

It shouldn't if you have Pro. Are you logged in? Have you made an API group?

[gloomy cipher]

aws gives you free 8gib ram server for like 30-40 days when you make a new acc so i hosted it there cuz i dont want my pc to be always turned on

[worthy mica]

Yes and Yes.

[short narwhal]

I thought about using oracle since that's free forever but haven't managed to bother yet

[short narwhal]

Anything in https://www.perplexity.ai/account/api/billing ?

[gloomy cipher]

go to api billing

[gloomy cipher]

if you ever do, watch the video by Tech With Tim he does it very securely.

[gloomy cipher]

you are qualified for the free 5 monthly credit

[short narwhal]

Looks very different to my UI. Maybe email [e-mail] and see if they could help you better

[gloomy cipher]

you can see it in the image

[worthy mica]

Yes, I am still unable to generate without purchasing credits.

[short narwhal]

Will do 🫡

[gloomy cipher]

when you made this pro acc did you use a credit/debit card?

not sure if it seeks verification

[worthy mica]

It keeps trying to charge my card $5 instead of giving me the free credit. I think I'll have to emial. @gloomy cipher

This is the comet mac app @short narwhal

[worthy mica]

Yes.

locked the card to ensure I wouldn't get charged while trying to convert the $5 promo credit for API Key. Glad I did lol.

[worthy knoll]

https://superbowl-liart.vercel.app

Claude opus and Gemini 3 pro working together

[past wyvern]

How does Browser Control work? I downloaded Comet, I paid for Pro, and then I tried to use it and it just kept telling me it wasn't able to do anything I asked it to do

[naive sage]

Sooo... I heard someone say recently that agentic actions with comet no longer work with Pro, I'm not sure if that's true or not, though, as I don't use Comet.

It was briefly discussed in another Discord.

[past wyvern]

So it only works if you don't have Pro??

[naive sage]

What was the exact reply you got?

[novel arrow]

you made that ?
opus can do better no hate

[worthy knoll]

Opus made it….

I had to use Gemini to try make the ui better

Because opus 4.6 did shit

[novel arrow]

i made this with sonnet 4.5 https://demofinwise.netlify.app/, so imagine what can sonnet do

[worthy knoll]

It’s just a lot of tiles..

[dry kestrel]

This is not true.

[dry kestrel]

Did you log in?

[past wyvern]

Yeah

I logged in first, then I bought Pro when I realized that it's a Pro exclusive feature

[dry kestrel]

Try restarting the browser

[odd lichen]

Nothing wrong with using whatever you feel most comfortable with, especially if all your chats are with one company or another. I do believe there should be more of a push for these AI companies to export all my chats whenever I want to smh, but this will never happen without regulation imo

My business Google workplace account comes with Gemini and with the impending Google/Apple partnership, I’m already integrated into their ecosystem already

[pallid lily]

did perplexity remove the limits

[naive sage]

Perplexity does offer the option to export individual chats, I'm not sure there is a bulk export though.

[pallid lily]

because im able to put in images

[naive sage]

It's practically unlimited for Pro Search

[pallid lily]

now

[pallid lily]

yea but yesterday

some stuff happened

where i was told to upgrade

despite having pro

[naive sage]

I think you get 10 uploads per day or something like that

[agile agate]

Hey guys, how can I talk to support?

[naive sage]

You can try emailing [e-mail], or go to Account Settings > Support > Contact (it brings up a popup to chat to).

[limber drift]

why does perplexity pro have document uploading limits now

[naive sage]

I'm not sure what motivated the change, can only speculate.

[dusk elk]

https://youtu.be/rkJ432q8KmY?si=YDLMmKAhzlYnG1oa

[dusty yarrow]

When did you have to upgrade to max to get more images, thats annoyed me, I'm paying 220 a year for pro and I gotta upgrade to upload a damn image?

[shut gyro]

For anyone on the fence about upgrading to Max despite how sh*tty Perplexity has been downgrading Pro, I bit the bullet, upgraded, and can attest that it's very good. (I have no affiliation with Perplexity)

[distant aurora]

@green crystal do you ever feel like gpt kisses asses of companies and legal system sometimes?

https://limewire.com/d/kTEsx#265JZigdQU

It's just saved html file. Since arena doesn't support chat sharing for now.

[quasi saddle]

hey guys, I have an old photograph that has part of a company name and has a number for an address in new york that was taken over 40 years ago with my mom and her mom and their extended family and I want to show the image to an AI model so that it might be able to tell me which building it is that they were photographed in front of. Anyone know which AI model would be best for that sort of stuff?

[quasi saddle]

the building was owned by one of our extended family and we're trying to find out about them to reconnect^^

[distant aurora]

Gemini or chatgpt. Preferably used natively but perplexity can work too.

Use the thinking mode on free chatgpt account.

Grok and Claude don't give me very reliable results for location based queries.

[naive sage]

The good news is that there's lots of old pictures and listings for New York from 40 years ago... (that's not too long)... that sounds like some OSINT challenges that are posted from time to time to do geolocation of an image. If you'd like help with that, let me know.

[short narwhal]

Oh my god OpenAI Super Bowl ad was mid 😭

[distant aurora]

The google maps integration favours Gemini.

[distant aurora]

Can't be more mid than Claude ad.

[short narwhal]

Claude ad was good though

[distant aurora]

Claude using an ad to trash chatgpt for showing ads on free tier.

And people cheering for it, lol.

[short narwhal]

Doesn’t mean it was a bad ad

[quasi saddle]

ya i'll try the gemini 3 pro on perplexity and if it doesn't work i'll give you a direct message if that's alright :D

[distant aurora]

How is it not misleading?

[short narwhal]

OpenAI’s shows codex app for like 2 seconds and says “you can just build things”

[short narwhal]

Doesn’t mean the ad is bad??

[distant aurora]

So misleading ads aren't bad for you.

Noted

[short narwhal]

It was funny

[distant aurora]

No windows app 😭

[short narwhal]

https://x.com/openai/status/2020649757434327362 like ehhh

OpenAI (@OpenAI)

You can just build things.

X · Feb 9, 2026, 12:03 AM

[naive sage]

Was it the farming one? I don't watch the stupid bowl, but I sometimes do look up the ads that run during it when people talk about them lol

[short narwhal]

It’s the one above

Oh Anthropic ad is rolling now lol

[short narwhal]

I’m watching this for just the ads 😭

And bad bunny

But I’m going to bed before that probably, so :(

[distant aurora]

https://cunnyx.com/openai/status/2020649757434327362

OpenAI (@OpenAI)

You can just build things.

**💬 81 🔁 48 ❤️ 404 👁️ 15.3K **

▶ Play video

CunnyX · Feb 9, 2026, 12:03 AM

iPhone ahh ad lol.

If you are getting what I mean by that.

[untold flume]

Have anyone tried that seedance2.0?

We are cooked

[violet field]

jo

[untold flume]

Seedance2.0 is dope

This is something ahead of grok or nano banana at least one year

[distant aurora]

Physics does seem better than veo.

Grok isn't worth mentioning yet.

[untold flume]

Film industry is cooked

[distant aurora]

Lmfao. No. This was not great.

[naive sage]

Wow, having seen several ... "attempts" ... at people generating combat with swords from different services, that is really good. For some reason, characters grabbing/drawing and/or holding different objects - as well as object collision - is really hard for a lot of models.

[untold flume]

Are you sure?

[distant aurora]

Yeah. First one was great. Second one seemed bad.

[distant aurora]

Yes. Lots of inconsistencies.

[untold flume]

Let’s see the 3rd one

[cobalt vale]

it's better than other models

[naive sage]

Yeah, still better than others I've seen.

[distant aurora]

He said film industry is cooked so...

[untold flume]

For now I can only find Chinese lang so I’m not sure how good the voice is

[naive sage]

Nah, the film industry actors and writers are basically a gigantic lobby group, they'll just go on strike to demand they outlaw it.

[untold flume]

Wasn’t this some industrial level of shit

[short narwhal]

No

[untold flume]

Ok I got one more

Let’s see

46 sec video uploading

Pic was generated by nano banana but the video is from seedance 2.0

This is all over X

Damn

My job will soon be gone I guess

[naive sage]

China won.

[distant aurora]

Bro I bet people would be fighting over this on twitter, lol. https://cunnyx.com/i/status/2020604051319136707

Reddit Lies (@reddit_lies)

**💬 39 🔁 102 ❤️ 5.8K 👁️ 91.6K **

CunnyX · Feb 8, 2026, 9:02 PM

[distant aurora]

Not yet. But soon.

[untold flume]

The dude commented this video cost him $2 worth of tokens

$2

Fuuuuuuuk

[distant aurora]

I'll say this again. Ai won't take your job. People with mastery in wielding it will.

Besides, it's still far from anything serious.

I would say you have 2 years minimum.

Most of the employers are dumb people.

[untold flume]

True

[distant aurora]

And lazy too.

They'll never get a good result themselves even if they sat to prompt the ai themselves.

[green crystal]

Film and game industry have extreem needs for customization

[distant aurora]

Employers would rather pay their way out of problems than actually try to understand them deeply.

[green crystal]

Like very extreme

[untold flume]

Jesus this seedance model is created by bytedance

Ok now I can see why

[green crystal]

Stuff like a surface being a little bit too reflective or an object a few cm too far on one side

[distant aurora]

Lmao. Fr.

The rants about normal coded games are high already.

These people will smell vibe coding at first glance.

[untold flume]

The same company owned TikTok b4

[green crystal]

Nope only if you go against a ToS and you gave the ToS beforehand

[untold flume]

Ok that’s not to bad

[distant aurora]

They still own tiktok.

Lol.

[green crystal]

Yeah it's the same deal as massive free asset libraries

People recognize it

And will instantly be disgusted

Let alone Ai stuff

[untold flume]

No they don’t, they sold TikTok like 1 month ago

[distant aurora]

That was just the us branch.

[untold flume]

Anyway I was banned from using seedance

[distant aurora]

Right?

Correct me if I'm wrong.

[naive sage]

No more submissions

[untold flume]

You are right

[distant aurora]

No tos was violated I believe.

[untold flume]

I’m banned, but it’s 15 sec default

I set Kimi 2.5 as default on my perplexity

One shot solved my problem which took gpt5.2 and gemini3 about 10 rounds of troubleshooting

[distant aurora]

Crazy if true.

Though, I wouldn't be surprised if Gemini shits the bed.

[sick ermine]

It better not

[distant aurora]

The only thing that model is good for is audio and video uploads imo.

[sick ermine]

Perplexity just shit on college students everywhere

We are clinging to Gemini rn lol

[distant aurora]

I love qwen.

[untold flume]

It’s too chaotic these days

[distant aurora]

Free so far.

[untold flume]

I have to subscribe to like 4 models to work

[distant aurora]

Qwen would have been perfect if it had folders.

[sick ermine]

I just need two lol Gemini and liner and I’m Gucci

[winged kiln]

Congratulations to the Seahawks...I guess?

[sick ermine]

Perplexity was my life line now it’s the source of my agony

😔

[sick ermine]

PLACE UR BETS LETS GO LETS GO LETS GOOOOO

[untold flume]

Perplexity is great on daily news

[naive sage]

Might want to check those references

[umbral swift]

man this app is absolutely lovely

can i lit pay for perplexity only n then use gpt 5.2 or gemini without paying for EACH ??

[untold flume]

So anyone here is Chinese? Can you help me login my seedance account on your phone app once so I can work on the website version

[sick ermine]

I believe they’re scrambling, probably on the verge of financial collapse.

[winged kiln]

It looks like it's been removed

[naive sage]

I wonder if they scaled up faster than anticipated, and that caused problems.

[sick ermine]

I think so. Spent more than earned

Or at least far out of budget

[dry kestrel]

This is not exactly true. Current structure is ByteDance still owns U.S TikTok, it's just it will now operate under an entity that is only responsible for data security and content moderation, that has U.S investors like Oracle but Bytedance still is the largest single shareholder of that joint entity.

[untold flume]

Damn

Ok

But my FYP was broken for like 2 weeks

Maybe it’s on oracles end

[dry kestrel]

Current structure

[untold flume]

Thanks

[untold hedge]

Hello everyone! Just exploring the Discord server and excited to be here. Looking forward to connecting with the community!

[dry kestrel]

did you see the OAI ad

[short narwhal]

Yeah

Meh ad

[dry kestrel]

did you not do the easter egg

[short narwhal]

I can’t botherrr

[dry kestrel]

its 1 click!!!

[short narwhal]

Whaa

What is it

[dry kestrel]

theres a skill in codex

[short narwhal]

I don’t have codex installed

[dry kestrel]

https://x.com/EdDiberd/status/2020662687382802678/photo/1

[dry kestrel]

what

[short narwhal]

The show

I’m going eepy now it’s over

[dry kestrel]

I thought you didn't sleep

[short narwhal]

Why is there an ai slop ad

[dry kestrel]

I don't think all of it is generated by Sora

[short narwhal]

That was some slop though

That robot jumpscared me

[dry kestrel]

better than what Coca Cola did

[dry kestrel]

skill issue

[short narwhal]

Are we talking about the same ad

[dry kestrel]

https://x.com/OpenAI/status/2020649757434327362

OpenAI (@OpenAI)

You can just build things.

X · Feb 9, 2026, 12:03 AM

What ad are you talking about

[short narwhal]

Nah that ad is way better than the Coca Cola one

[short narwhal]

The one that played like just now

You should have seen it

It was dalle3 levels of AI video slop

[dry kestrel]

By OpenAI?

[short narwhal]

Nah some other company

[dry kestrel]

...

[short narwhal]

Not sure what marketing team greenlit that ad

[dry kestrel]

What company

[short narwhal]

Some drink company I don’t know

[dry kestrel]

Ah

https://x.com/tbpn/status/2020651201445179844

TBPN (@tbpn)

Introducing Claude with Ads

Free access to Anthropic’s most powerful model, powered by ads.

Try it now:
https://t.co/QU43cgIJRm

X · Feb 9, 2026, 12:09 AM

[short narwhal]

https://x.com/aaronp613/status/2020652862062371062

Aaron (@aaronp613)

The new @OpenAI ad from the Super Bowl shows a macOS app running on a Windows laptop

X · Feb 9, 2026, 12:15 AM

[dry kestrel]

i saw that

I think it's more of an issue with Apple

[short narwhal]

Could be

[dry kestrel]

and because the ad was just made

they scrapped the old one after Anthropic made fun of them

and Apple has some stringent rules on using their products in marketing I think

[short narwhal]

True they do

[dry kestrel]

also free merch

which is cool

[green crystal]

so uh whats the easter egg

i dun get it

[dry kestrel]

did you watch the advertisement

there's a shot where it's a view of the Codex app making a hidden skill for codex merch drop

[empty token]

Hello, my name is Marios, and i am using perplexity theought revolute premium to have the perplexity pro version, for some time i had it but recently it seems i do not have the perks of pro and i get the option to get pro, does anyone knows something about it? Cause i cannot find a solution to this and i am confused. Thanks

[ebon tendon]

why am i getting this all of a sudden

[wet copper]

I encountered the same situation and contacted customer service. The engineer said their Pro upload limit had changed, so once the limit was reached, it would keep showing the purchase option. When I asked how many Pro users could currently upload, they said they couldn't disclose that. The engineer was very professional, but I'm quite disappointed with the company.

[naive sage]

I have heard up to 10/day and up to 50/week. 50/week is cited here: https://www.perplexity.ai/enterprise/pricing.

[wet copper]

😢

thanks for sharing

[wanton timber]

Yeah

[candid cairn]

Making changes that hurt users after they’ve already signed the contract? You know what?
That’s exactly what every online service that lost its users' trust has done. especially the ones that just blindly hoarded users without thinking.

[stoic garden]

Hello, my Reason (Deep Think) mode seems to be shut off. It's visible but greyed off/can't be selected. I am a Pro subscriber. This just started today - now all I can use is either Fast Response or Research.
Any help would be much appreciated

[woeful flare]

Maybe somebody knows - will they change limits?

[gleaming roost]

Instructions: Copy whole tekst between ">>>>" and "<<<<" and reply this list of alternatives and add your own beneath the last entry. For users, paste input into AI and ask for a comparison in relation to Perplexity.

T3Chat
Scira AI
QWEN
Google Gemini (with Deep Research)
<your alternative>

Dear Perplexity,
Your product and features you’ve built are worth it. However, keeping the community in the dark and pull stunts like this undermines the very asset you rely on (the customer, we). Take the time to reflect, and come back with clarity, not political or manipulative spin.
<<<<

[fierce acorn]

hello

[karmic slate]

hi

[karmic slate]

wat is python os?

[fierce acorn]

This is an operating system created using Python coding. The design is old, but it is somewhat usable.

[untold flume]

guys i managed to get a seedance2.0 account from bytedance

Here is my first version

Of first prompt

This is freaking awesome

[wicked trail]

Here with other people because the upload limit changes have messed up my workflow today.

I’ve never had any upload limits reached in the past and today I got it for the first time. It then redirects me to “upgrade” to pro (which I already have) so there’s clearly some sort of problems happening on multiple levels.

Image analysis is the primary reason I use perplexity.

If the issue is they need more revenue to cover the likely astronomical computing costs, add a $30 tier, explain that to your customers in advance, and let us opt into plans that work for our needs. But this carpet pull is kind of crazy. Your customer base signed up for pro plans explicitly advertised as unlimited file uploads, and that has been quietly changed?

This is in addition to the fact that I was on the $5 a month student id plan which they quietly changed over, and I’ve been stuck in month long email replies with support trying to get it fixed (sheer ID verified, charging me 20 a month with no warning or explanatikn

[karmic slate]

ooh?

[ruby sinew]

I see everyone talking about the research quotas of 20/month. But my pro plan (student discount) is 2/month.
It's cooked.

[royal ingot]

yeah, happening with me too. mine is 5 per month for some reason

and i used all of them up

[fossil shuttle]

Ikr it’s absolutely bs

[tender venture]

Was pretty messed up for me all around, took a week off work to build out with perplexity an essential part of the workflow then basically got forced into upgraded at 10x the cost while also being out a weeks worth of wages. Super messed up.

[fossil shuttle]

Honestly prob just switch to Gemini at this point such bs

[wicked cloud]

is it the same for you guys, even if you are pro you get limit on the advanced ai model weekly

[fossil shuttle]

Got 1 year of this bs and they switch up on us will still wait out the full year but will switch right away

[fossil shuttle]

Haven’t gotten there yet

[wicked cloud]

this had been problem for me for the past week, before i had no problem about it

but the weird part the limit reset everytime i took a break from the app, but the notif still pop up

[wicked cloud]

man, i had to keep taking a break on my research because of it

so the pattern is, 10 ai usage, got limited, took a break, reset to 10 again, limit and repeat

[fossil shuttle]

I think perplexity is more of a summer project that won’t last long cuz I doubt they make a lot of money.

[wicked cloud]

i mean it make sense

paying crap tons of different AI API while offering the same price for one month of those single AI company

but damn this had been my tools since the first semester of my college 🙂

i dont mind the reseacrh limit, context token etc, but i need the AI usage atleast

[fossil shuttle]

Yeah

[wicked cloud]

i know you guys wanted us to pay for max, but not all of us that rich to pay 200 bucks for a month 🙂

[woven widget]

It is also not possible to get pdf files senes yesterday , when u ask the ai to create a pdf files,it tell u that is not possible and u can copy the text and create a pdf file in word or google sheet….

[wicked cloud]

yeah nope

perplexity isnt agentic ai

its a whole different platform than each specialized AI

perplexity just basically gave us a plate of the AI to use

despite being paid pro customer 🙂

[fossil shuttle]

Looking at Poe it looks better then perplexity and same price

[wicked cloud]

yeah at this point if it kept going to be like this

[warped shard]

In my case, it shows up only on my phone App. The Perplexity website on my computer does not display the limit.

[cinder echo]

I had a same problem. Revolut support said you need to add a card to your account Perplexity account, but it doesn't charge you. After adding card everityng as usual 🙂

[crystal drift]

Does Max have Deep Research limit?

[untold flume]

Why this deep research so important

Gemini offer free student subscription

Anyway here is another video

[steady osprey]

no they dont

[untold flume]

I got 1year free offer

[steady osprey]

it expired

[mild kite]

Right now it just one month for students

[untold flume]

Like I said yesterday, for now if your budget is $20/mo then just go for kimi2.5

It’s awesome

[slow linden]

Have you tried negotiating with Kimi? Apparently can get it as low as 0.99 (probably just for one month)

[untold flume]

Yes I did

[mortal merlin]

My pro subscription is under pause kindly check it

[wicked cloud]

ah yes i take that too, 14 months to be exact with 2tb google storage

[rugged osprey]

is there anyone using Perplexity agentic research API? (access to third-party models like gpt-5.2, claude opus etc)

[icy parrot]

Kimi k2.5 is preatty impressive ngl,,ive tested it yesterday and its really goodz it goes close to opus4.5 with far better usages and price per token

[peak hatch]

Hey folks! I’m new to this Discord server!
Could someone please let me know what the usage limit is for Perplexity Pro Search with the Pro plan?

[icy parrot]

Idk i dont use it anymore ive unsubscribed

[peak hatch]

INDEED, Kimi K2.5 even in coding and image analysis tasks performs way better than GPT 5.2 and close to Gemini 3 Pro ( atleast for image analysis) and coding tasks (like Sonnet), wasn't expecting such performance from this, but yeah, it is sick!!

[icy parrot]

Idk bro gemini 3 pro performed worst for me out of all llms

I currently use kimi k2.5 and gpt 5.2 and gpt codex 5.3

And gpt image 1.5 for images

[peak hatch]

yeah, i mean, Gemini 3 pro is good for identifying images and image analysis tasks in my case, GPT is okayish for coding with me and making frameworks, Sonnet is best for coding, but Kimi is dammn (atlest being an open model and not being a corporate closed shit is another win factor )

[icy parrot]

But usage of LLMs mostly comes down to your preference and which ai understands and works with you best

Peopel hate on GPT but honestly iam really happy with it

Though sometimes iamge quality worsens for plus user due to peak hours

[peak hatch]

makes sense

[icy parrot]

Gpt codex 5.3 is really impressive to me too

It got way faster compared to 5.2

And it seems to reason better

[peak hatch]

yeah codex is another variant
and that shit is indeed crazy
like it thinks better and fastly also with depth

[crystal drift]

Does Max have unlimited deep research?

[icy parrot]

Honestly it seems it battles with opus 4.6 for reasoning and SWE bench

[peak hatch]

YES

[icy parrot]

Thing is non codex gpt models are made for daily use and questions more for enthusiasits thats why it performs worse in coding and logic tasks

[peak hatch]

i have tested it myself and in my case atleast, gpt 5.3 codex performed BETTER in reasonable conditions than OPUS 4.6

[icy parrot]

But i even like GPTs in chat editing of images as gemini ignores most instructions for me for some weird reason

[icy parrot]

Whats killing claude right now is the usage limits and price per token maibly because cheaper models outperform it

[peak hatch]

Gemini indeed ignore instructions and ONE BIG PROBLEM from this is that in gemini chat app, the token and thinking limit for each answer is way too constrained and short, which plummets the answers' quality by margin

[peak hatch]

Correct

[icy parrot]

Gemini is honestly underwhelming whats saving it is the price if it didnt had image gen kimi would probably take over

[peak hatch]

yeah
though i believe Kimi is underrated, like not many people talk about it as much as for gemini or gpt or claude

[serene sand]

#💬│general message

[young hollow]

Just wait for Deepseek V4 guys

👀

[peak hatch]

Basically, Perplexity has paused all ongoing subscriptions for users who haven’t updated their billing info.
Earlier, Perplexity didn’t require card details for the 1-year trial from third-party services, but now they do.

So when you click Get / Upgrade to Pro, you’ll see a screen asking for card details, even though the due amount shows $0.

You won’t be charged until your Revolut trial ends. After that, Perplexity will start charging based on the normal billing cycle from your card.

[peak hatch]

yeahh, idk when they would launch, taking too long

[young hollow]

Im pretty sure Arvind Didn't Knew How Much Bills These Pro Promos will Rack Up

Same with Comet Promo

[peak hatch]

he was busy offfering 34B$ to buy chrome and forgot the bills at home

[young hollow]

Those were just Marketing Stunts Honesty

[peak hatch]

EXACTLY

[young hollow]

He was never actually going to buy chrome

[peak hatch]

Exactly
or in other words, Google is never going to sell their Chrome

he knew

[young hollow]

That's About All of it Yeah

[young hollow]

It will surely launch in 2026

[peak hatch]

yesss

[young hollow]

Also GLM 5

[peak hatch]

@everyone
Could someone please let me know what the usage limit is for Perplexity Pro Search with the Pro plan?

[young hollow]

Open Source will Catch Up to Close Source before 2027

[peak hatch]

THAT'S really what I want

[young hollow]

Potentially might Win as Well

👀

[young hollow]

Me2

[peak hatch]

could be

[young hollow]

Im Unofficial Brand Ambassador of Chinese Ai

Because is Free

👀

[peak hatch]

😂

[peak hatch]

@young hollow do you know this

[young hollow]

I don't but let me find it

[peak hatch]

i searched on the internet and according to official perplexity sources, there is a weekly limit for average use with perplexity pro plan

[peak hatch]

@young hollow Thank You for your efforts,
i had also seen this page
but people on reddti are saying that they have definitely used less than 200 and they faces the limits

or more than that and they didnt face an

[young hollow]

You Welcome

It seems like is not exact fix

Depends on server load and region

[peak hatch]

plus with my experience too, i didnt face any limits right now, but i wanna know

[peak hatch]

Yeahh

[young hollow]

That's All About it

Just Squeeze as Much Value You Can

[peak hatch]

Roger That!!🫡

[young hollow]

U got it for Free Right?

[lime forum]

why does it not even tell me the model?

[young hollow]

Because their Bills are Racking Up

[lime forum]

lmao sounds about right

[young hollow]

Well you can what they Say

If you Don't Like a Product

Just Switch it

[lime forum]

i feel like it's more expensive to let me keep modifying my prompt till i get the model i want, vs just letting me use the one i selected

[young hollow]

Something like that yeah

[lime forum]

idk any good alternatives on the same level with the same features as pplx tbh

[young hollow]

The Thing is

Others don't have all Models

Gemini or GPT is Your Best Bet

Use Arena or Yupp for Using Other Models Only when Needed

[hallow flume]

do they lift up prices?, now edu costs $9, i subscribed for $5?

[young hollow]

Seems About Right

As they have Reduced Limits as Well

@lime forum you should also try all Free options Copilot, Grok and Other Chinese Ones

[hallow flume]

Okay... that's not acceptable. They don't know what they want either. First they lower the limits, now they raise the prices. What's next?

[young hollow]

Something similar i guess

They were just being Generous

Now they Realised they can't Keep Up

And this is Causing a Backlash

[hallow flume]

yes

they done false limits, cause nobody will sub to max if the limits at pro are enough

[young hollow]

How can a Company Justify when Majority of their Pro Users got it for Free

[young hollow]

Well those who won't still won't pay for Max

[delicate nova]

gg scammers, been pro user for months - unlimited perplexity research - today first time ever “ran out of it” 👍

Thought wow best product and competitive. Today realized perplexity is in the past. Changed the rules whenever, with no notice, no email, no popup in apps about upcoming changes.

Terrible, exploiting customers to get free trial and now silently and slowly adding rules that squeeze everyone to upgrade for Max.

[ivory dirge]

It's such a shame because Perplexity was the one ai product I was using over every other product by far. It'd replaced my homepage and I really loved it. These new limits destroyed all that.

[winged hill]

Gemini is unusable. Google downgraded the model like crazy

[icy parrot]

Gemini 2.5 didnt seem good to me either

[winged hill]

what do you mean by exploiting customers to get free trial? Giving something free isn't really exploiting 😄

[winged hill]

I don't know, I loved Gemini 2.5 and when 3 Pro came out it was glorious to me. Then it all fell down to oblivion

[icy parrot]

Doesnt change the fact the quality of pplx is terrible

Limits are stupid and models are lobotomized

They put gemini nano banan pro behind max and dont even consider adding gpt image 1.5

[winged hill]

so why don't you just go to Claude Pro or Gpt Plus? It isn't bad, just very limited as well, but less limits because its only their own models obviously

[icy parrot]

I have gpt plus and kimi moderato

Claude isnt good deal either too

[peak hatch]

No, edu Pro

[winged hill]

care to explain how models are lobotomized? They don't even hide system instructions, its all straightforward

you guys don't realize API access is expensive

[icy parrot]

The models fail to answer or do things correctly for me at all regardles of how much i explain, sonnet 4.5 just feels like sonnet 3, the search isnt good either like all AIS god me up to date data from internet about image model of openai which is gpt image 1.5 while pplx talked about dall-e 3 which hasnt been in plus plans for a year now as default

Thats example to be clear

And what baffles me is that pplx claims to be search engine optimized but it isnt

[void folio]

I think the best thing is to switch to Llama and host all the models locally.

[delicate nova]

bruh someone uses the service get dependent on it then forced and exploited to upgrade, cope with bare minimum or leave. Either give what you promised or don't promise at all.

[winged hill]

your answer only talks about the hallucinations about image gen models (but if you are very clear with "generate an image of..." then it works. I agree they should add a button to select image gen and force it.

But you have 0 arguments about the models being really bad

i did dozens of benchmarks on complex prompts to compare official apps outputs vs perplexity outputs and usually perplexity wins or gives similar results. But Perplexity is more focused on RAG, so it's limiting the tool calls. again, its all in their system prompt

[short narwhal]

It is not

[winged hill]

They were not very smart by offering so much for so cheap/free but it was obviously unsustainable and a marketing strategy. Definitely the f* ck up can be documented as what not to do when you scale and onboard users at first

[void folio]

Why?

[icy parrot]

For me no it hallucinated in many other querries unrelated to image gen, its not about generating images in pplx but data it provides about topic

[short narwhal]

Models you can run locally are incredibly stupid

[winged hill]

ok but then I don't understand : the service is bad, it hallucinates, uses subtier models instead of real models... why do you complain about quotas? There are competitors

[void folio]

Well what if I use APIs- similar to how Pplx works

[short narwhal]

that’s not local anymore though?

[dense pollen]

Why did they nerf the number of queries for pro users?

[wanton crest]

Because they oversold the plans and are verging on bankruptcy?

[dense pollen]

So, we shouldn't expect it to go up anytime soon?

[short narwhal]

It got upgraded and therefore is much better but also more costly to run #📢│announcements message

[wanton crest]

I'm expecting the entire company to collapse before my subscription expires. Lots of complaints have been raised, so we'll have to see whether the company understands the reputational damage it has caused itself. I suspect there won't be much movement on the limits for a while.

[wanton crest]

And if anyone had asked for that, they would have been happy to pay. Quite why the previous option was removed, when it was sufficient for most Pro users, no one knows.

[short narwhal]

You’ve still got pro search which functions basically the same as old research

[dense pollen]

Have you tried it? Has it gotten better?

[wanton crest]

Unfortunately this seems very much like the dotcom days. Highly technical people doing wonderful things, only to be let down by their lack of understanding of business.

[short narwhal]

Haven’t personally, but others here have and have said it is noticeably better

[dense pollen]

Thanks. That's sad. What reputational damage are you talking about? The web crawler suits against them?

[wanton crest]

The web crawler suits, the obvious lies about "unlimited deep research", the lack of communication about the new resource limits, the over selling, etc

[winged hill]

they were bleeding money as any growing startup. They will loose customers but bleed less. Gain some max users on the way. No free meal

[wanton crest]

No meal at all when none of your customers can trust you to effectively and accurately communicate the limits that are in place.

[candid cairn]

This bald guy's been trying to pick fights since yesterday with his sloppy defense of the management just to get attention, so arguing back is a waste of time.

[wanton crest]

My suspicion is that they were hoping to be bought out, and when that didn't happen, because the product really isn't unique, they realised they actually needed income rather than a high user count. At that point the overselling of the service became apparent, and instead of communicating with the Pro users they just slashed the limits.

And that's how you go bust. You lose trust, you lose reputation, you lose.

[candid cairn]

Ah, maybe he was doing the same thing even before I saw him here yesterday. I mean, he seems like he's wasting time.

[delicate nova]

I don’t believe that perplexity cobrands with major banks etc giving off full year free subscription and then realize they overdone it. They’d spend b on advertising instead.

If it’s so expensive and so many users - why not put queues when it’s heavy, but instead restructuring the benefits when coincidentally they capped out max leads, hooked users and now Oops - Pay more. No way perplexity is so unaware, and we, two persons are so smart relatively to the whole company that can use ai themselves

[winged hill]

"No way perplexity is so unaware, and we, two persons are so smart" : exactly

[icy parrot]

You can run models like kimi k locally but you need whole rack for that 😭

Or mac station cluster like network chuck did

[fossil shuttle]

Honestly if they marked up the price by like $10-20 temporary sure that would be fine but making us buy a $200 plan that we won’t even be able to use 50% of what they give

[icy parrot]

Ueah they should add some intermediary plan like for 50€ monthly like literaly everyone else

[fossil shuttle]

Or custom plans you pay what you actually want

[icy parrot]

Having opus behind max is also stupud opus is no longer as expensive as it was juat limit monthly use for it

[young hollow]

Scira Ai is Like Perplexity but with Many More Models

Seems Interesting

They say No Limitd

[fossil shuttle]

I’m currently looking at Poe it’s a perplexity alternative it’s honestly not bad

[icy parrot]

I wouldnt trust that

[icy parrot]

I just switched to API providers and made my own app where i merge them with search and image gen

[fossil shuttle]

Worth it?

[icy parrot]

Most of the time yes if you dont use it heavily

Mainly image gen is really worth it

I dont wait amy queues or bump downs

So i always get best quality

[fossil shuttle]

Meh you can easily host your own image gen even with a 8GB card I tried with a 8GB card and got a lot of good pictures just need the right free model

[icy parrot]

I will maybe make the platform opensourced later on so people can easilly use their apis

[icy parrot]

Flux isnt as good

I tried those

Not worth it

[fossil shuttle]

Flux?

[icy parrot]

Id rather pay my 13 cents for one image

[fossil shuttle]

I download models from civitAI.com and use webui and find models for what specific thing I’m trying to generate and switch back and forth

[young hollow]

Understandable

[distant aurora]

@green crystal when you top the benchmarks but are dyslexic and retarded.

Yo @dry kestrel , I might have found a glitch in kimi.

[green crystal]

Benchmaxxing? Lol
Well partially

[sharp frost]

[distant aurora]

Okay nvm, it finally used search the third time, lol.

[green crystal]

In allot of comparisons between opus 4.6 and 5.3 codex the overall theme is opus is better with looks but codex is better with technical stuff

[distant aurora]

People will eventually buy flat subs of codex over opus as the gap narrows considering the drastic difference in rate limits.

[fossil shuttle]

Oh I see why pro researches limit went down a lot because they switched to opus 4.5 such bs

They really gotta vote on this crap the previous research was just fine

[green crystal]

anthropic their response was fast opus access
for $225 per m tokens lmao

[icy parrot]

Kimi holds close to codex and opus in coding, codex wins for me longterm

[young hollow]

Kimi is Cheaper Tho

[gloomy stump]

Hey, I’m getting a paper ready for arXiv in the area of multi-hop RAG / long-context LLM reasoning, and just realized I need an endorsement. If someone here can help, please DM me — would really appreciate it.

[icy parrot]

Yeah thats whay its really impressive comapred to claude

But codex still seems to get the logical part better in my case

What makes gpt a good deal is that gpt codex and normal web gpt have unrelated usage limits if you use up all you codex weekly quota you can still use normal non codex versions of gpts in the web as normal even if codex is used up and counter wise and kimi is interesting mainly due to price to performance and being open sources

[young hollow]

Just wait for Deepseek V4

👀

[icy parrot]

I never liked and never will like deepseek

Sadly deepseek will never outpeform even most open sourced models

Qwen is something that looks more interesting

Deepseek v4 just promises lot on coding, but coding capability isnt only thing important if it gets outpeformed in literaly everything else then whats the point

[hoary phoenix]

I just got told I’ve hit the weekly file upload limit - I’m on Pro and have been for a couple years now. I’ve barely uploaded more than a handful of screenshots and I’ve never seen this message before.

Has the limit been decreased or something??

[icy parrot]

Yes a lot

Ton of people are mad right now

[hoary phoenix]

Oh, frikkin great. I’m loving how shit’s getting steadily more and more expensive and for the services that haven’t changed their price, they’re massively restricting what your money buys instead! So feckin’ depressing.

[icy parrot]

Thats why i use model originaly from providers or use their apis

Afaik neither anthropic, openai, google, moonshotai, z.ai dont limit max amount of uploaded images afaik or atleast i didnt hit limit and i upload a lot of images

[hoary phoenix]

Yeah I just mean across the board really. Energy prices, gas, food, insurance etc etc. Just… ugh.

[icy parrot]

Youre better of buying from original

[green crystal]

with chatgpt you can go much higher with project file uploads

[icy parrot]

I know

Or you mean on pplx

[green crystal]

nope chatgpt lol

[icy parrot]

Yeah agreed

[green crystal]

i can have it work with 30 files

[icy parrot]

I uploaded daily more then 30 images and it didnt limit me

And iam on plus plan

Deep research is also more generous in gpt i believe

Ah no

It nearly same

[green crystal]

its sorta "more" if you also use agent mode for research

[final wind]

•`•

[winged hill]

anyone having issues with opus today? seems to be losing context or sometimes it gives only a sentence of output and forgets the rest

[jovial jungle]

Hello everyone,Can I ask question about student subscription in perplexity?

Today’s I have already updated subscription but I could not use my research mode and upload files.Because of limitations.Why?
I updated new pro the subscription .That’s not great.😢

[junior yarrow]

Yep, same here. I've never had an issue using Deep Research and now it's saying I have to upgrade.

[proud apex]

Same, I subscribed in September and somehow my subscription isn’t available anymore today, but I remember that we could get 1 year as a student

[full raptor]

🔥 Seedance 2.0/ seedream will officially launch!
If you want early access or need to get started, feel free to reach out.

[short narwhal]

Check https://www.perplexity.ai/account/details if you need to update your billing details

[devout sonnet]

How do I know what image model is being used? I'm on Enterprise Max, but even though I select Nano Banana Pro, it uses some cheap image generation model without telling me. What's the limit?

[proud apex]

THANK YOUUU

[silent ocean]

appreciate it

[novel arrow]

i get nice quality

[devout sonnet]

That's clearly not Nano Banana Pro though

[novel arrow]

i am 100% sure it is

[devout sonnet]

You are tripping

[cursive island]

Download the pic and check the image size

If it's 4k it's NB Pro High

If 1k that's the base model

[devout sonnet]

I'm pretty sure Perplexity was using Nano Banana Pro for like the first few generations since the quality matched that of Google's. The screenshot I showed evidently show lobotomized generations. You can just tell.

[glacial nacelle]

How do I un pause my perplexity pro membership?

[shell wolf]

perplexity ➡️ account ➡️ account ➡️ "suscriptions" ➡️ see details ➡️ cancel

[glacial nacelle]

I j had to add a card

[shell wolf]

sooo the easiest way is just asking in perplexity how to pause

[silver mango]

Hey guys, how's it going? Is it possible to create videos/images with Perplexity Pro? I signed up yesterday and I still don't know the main functions.

[dry kestrel]

Interesting

best to report it in their server

[pale juniper]

#bringbackperplexitysubscriptionsforairtelusers 🥲

[shell wolf]

btw they have release date for comet in ios?

[quaint surge]

sup

[novel arrow]

comet on ios will be bad even if its released due to apple restrictions

[dreamy drift]

Did GitHub implode?

[mighty forge]

Hey Perplexity team 👋

I’m really frustrated right now 😤 and wanted to raise an issue regarding payments.

👉 Currently, there seems to be only one payment method available, and I tried adding three different international debit/credit cards, but none of them worked ❌
All these cards are active and work perfectly on other international websites, so this looks like a Perplexity-side issue.

👉 I’m also an Airtel user, and as per the offer, Perplexity Pro is free for Airtel users 🎉
But I’m still being asked to add payment details, and because card payments are failing, I’m completely stuck 😕

What’s more frustrating is:

❓ No UPI support (which is the most common payment method in India 🇮🇳)

❓ No PayPal

❓ No Razorpay or other local options

This really hurts the user experience. Many Indian users rely on UPI, and not having it makes onboarding very difficult 😞

I genuinely want to use Perplexity Pro, but this payment limitation is blocking me entirely 🚫
Please:

Clarify why Airtel users still need to add payment details

Fix the international card issue

Consider adding UPI / PayPal / Razorpay for Indian users 🙏

[distant aurora]

@green crystal real?

[novel arrow]

so its PPLX fault that you cant pay with a card ?

[mighty forge]

yes

[distant aurora]

?ban

Nah. Ban is too much. Lol.

Doesn't razor pay ask for card too?

[green crystal]

might be
5.3 codex on xhigh already feels much much faster than 5.2

[short narwhal]

smh

[novel arrow]

lmao, you're delusional, the problem is india's policy against free currency exchange, get a bank card activate your internaitonal dotation and pay like everyone else, india is not the center of the world.

alson indians purchasing power is low, nobody in India would pay $20/month for this

daily wage in india is $10/day for someone iwith an office job, so the people who have the money to pay $20 surely have a bank card or credit card

[naive sage]

What gave that response.

[viscid ivy]

@mighty forge Hey, I am a user from India as well, and I had no issues adding my credit card.

[viscid ivy]

Can't really blame them for having the policy, as the nation always had huge problems with high forex outflows. By the way, $10/day? You are overestimating.

[umbral swift]

hello, is "spaces" the place to build personas to play around wiht and personality experimentation etc

[short narwhal]

Gemini 3 Pro with custom instructions

[distant aurora]

Based ai.

[distant aurora]

Based.

https://cunnyx.com/i/status/2020832197771714943

Chetaslua (@chetaslua)

🚨 GLM-5 appears to be fully aligned with DeepSeek's architecture.
︀︀
︀︀A new vLLM PR (#34124) reveals GLM-5 reuses DeepSeek-V3/V3.2's architecture, including its sparse attention mechanism and Multi-Token Prediction.
︀︀
︀︀No multimodal support spotted 🤔

Quoting Chetaslua (@chetaslua)
︀
🚨 Pony Alpha on Arena is Glm-5 from @Zai_org
︀︀
︀︀Test : A simple test with a PoC token will reveal which tokenizer it is.
︀︀
︀︀The Chinese phrase "锅内倒入植物油烧热" (Pour vegetable oil into the pot and heat it) is a known tokenizer collision or "glitch string" specific to the GLM-4 / GLM-5 tokenizer.

**💬 11 🔁 11 ❤️ 231 👁️ 20.7K **

CunnyX · Feb 9, 2026, 12:08 PM

https://cunnyx.com/i/status/2020787200670015797

David Ondrej (@DavidOndrej1)

finally someone is calling this out

**💬 96 🔁 15 ❤️ 670 👁️ 44.6K **

CunnyX · Feb 9, 2026, 9:09 AM

[naive sage]

Yeah, unfortunately that kind of nonsense won't stop... just look at all the bitcoin miners.

[distant aurora]

Istg, what is wrong with muckerberg.

[naive sage]

Well... it's not wrong.

Calling someone a loser is not nice.

I would probably report that if someone called me a loser.

[tardy juniper]

The Least they can Do is Provide Regional Alternative for Payment

[distant aurora]

https://tenor.com/view/an-iq-too-high-raven-crow-tiktok-gif-3869671553367641450

[novel arrow]

90% people use bank card/credit card
india : they should adapt for us

[wise nova]

Any news from the Meta AI team?

[young hollow]

Our Monthly Internet is 3$ 💀

[young hollow]

When is it Releasing?

[chilly stump]

Dropping next week

[distant aurora]

It's not multimodal yet so I'm assuming no time soon.

Idk how good this is but it surely has my expectations high.

Time to test it.

[cerulean sedge]

Honestly I didn’t even know there was a limit, but did the limit change for pro users? Asking because I definitely did not use it more this month than I have other months yet this is the first time I’m getting this

[distant aurora]

Lmfao.

[modern hazel]

hello guys

i have a question about comet browser

i have pro plan, and then it says reached limit automated browser..

[distant aurora]

@green crystal no way they nerfed my boy

https://tenor.com/view/emoji-crying-emoji-gif-8626349402912892862

[modern hazel]

You've hit the monthly limit for automated browser tasks. Upgrade to Max for increased usage and our most advanced features.

[distant aurora]

https://cunnyx.com/i/status/2020893513253085551

VraserX e/acc (@VraserX)

The Information just confirmed that OpenAI is dropping a new model in ChatGPT.

**❤️ 4 👁️ 299 **

CunnyX · Feb 9, 2026, 4:12 PM

[green crystal]

Lmao

[green crystal]

5.3 and maybe new deep research?
The model being faster and more efficient would certainly help with cost and speed of dr using sub agents

[modern hazel]

guys.

plrease help

[cerulean sedge]

So they lying then

[modern hazel]

i have the pro plan

says You've hit the monthly limit for automated browser tasks. Upgrade to Max for increased usage and our most advanced features.

wtf

?

[novel arrow]

"upgrade to max it's only 10x the price" PPLX logic

[whole ledge]

I think i have been detecting some kind of prompt injection on my pplx searches

Or is just ai marketing

[modern hazel]

idk, but what is ACTUALLY THE LIMIT I DID NOT EVEN REACH 10?

[quaint ravine]

now back to "Up to 20"

[modern hazel]

where did you find that?

[quaint ravine]

(Enterprise) Max (presumably same for individual Max), is 25x that, which means 500 Research a month for Max

that's actually not super great, considering how some people were using Research lol

[quaint ravine]

https://www.perplexity.ai/enterprise/pricing

[modern hazel]

thats shit

[quaint ravine]

what's the estimated Claude Opus use for a Claude Max plan?

pretty sure can easily get more than 500 a month

iirc, I managed about 5-10 Opus queries a day with a Claude Pro plan

[modern hazel]

im going to switch fuk dis

[distant aurora]

Do we have any business management expert here?

@green crystal @gloomy cipher https://cunnyx.com/i/status/2020420647713935667

Elon Musk (@elonmusk)

Clawdbot 😂

Quoting mitsuri (@0xmitsurii)
︀
How was the show Silicon Valley so ahead of its time?

**💬 3.3K 🔁 7.1K ❤️ 82.2K 👁️ 17.28M **

▶ Play video

CunnyX · Feb 8, 2026, 8:53 AM