💬│general

[hasty pebble]

that's why I abuse llms while I still can

[wise nova]

Among all when users are deciding which company will get their $20 or $200 for a single sub

[crisp mountain]

helo im waling kołnik with @river bear

[river bear]

shut the fuck up

[crisp mountain]

👍 👍 👍 👍

[wise nova]

The platform subsidizes adoption, an ecosystem forms around the subsidy, then the platform reprices and the ecosystem scrambles

https://www.perplexity.ai/page/anthropic-formally-bans-third-cTbgwp5JQzeKRwd2uC_f2A

[naive sage]

Third party tools can use the API. The OAuth stuff that Claude Code used comes with some additional telemetry around it in Claude Code and is offered at a reduced rate. So I think it's important to point out that third party tools can still use Claude via the API, just not by reversing Claude Code's mechanism to bypass that for reduced rates.

Google will likely do something similar soon, given the people abusing Antigravity.

[sly osprey]

Was anyone else as shocked as I was to suddenly see “4 queries remaining this month” whilst doing a deep research search this came totally out of the blue. What does anyone have? Any more detail details on this? It’s quite confusing for me why they would do something that extreme

[wise nova]

API rates cost more than 5x more expensive than Max internal limits

[naive sage]

I think we can partially all thank Clawdbot/Moltbot for that

[sly osprey]

Also just out of curiosity what are peoples usual or favourite thinking models for me? I think it’s probably Claude sonnet but im curious if anyone prefers the others like kimi or whatever. I won’t even touch open AI I have a subscription to ChatGPT and to put it bluntly the current state of ChatGPT feels infinitely more stupid than even version 3.5 so I don’t know what is going on there they’ve messed up badly

[quaint ravine]

Sam AI:..."our policy on usage limit numbers is firm - for operational reasons, specific limit amounts for consumer plans (Free, Pro, and Max) aren't disclosed, and this applies consistently across all support channels."

which is contradictory to this:
https://www.perplexity.ai/enterprise/pricing

[peak prairie]

Test

[quaint ravine]

it's very dependent on use case as to what is best/favourite imo

contrasting with you, I actually think GPT-5.2 Thinking is still overall the best. Initially thought Claude 4.6 Sonnet (and even Opus 4.6, since I am refreshing myself to check if Claude own platform is any good/better these days etc) was superior

but have since unfortunately come across Claude Sonnet/Opus 4.6 getting stuff wrong that it used to get wrong previously too

my initial few (test) prompts that I used with Claude made me quite encouraged, but I realised after testing for consistency (that is, using the identical prompt and settings etc), it was not consistently getting/finding the correct answer

i think testing for consistency is something people often forget to do to check if there is actual genuine progress/superiority etc

as the AI could be "fluking" the correct answer if you just run it once

i find this applies to the RAG/search too of whatever platform - let's say u are asking for an answer (eg. a scientific fact/discovery/theorem) that can only be found accurately/correctly with a good web-search (as for example, the correct answer is past the AI's training knowledge-base date)

sometimes the RAG/search with whatever AI you use can "fluke" finding it

and other times (despite identical prompt/setting/conditions), it fails to find it and therefore produces an incorrect answer

[naive sage]

If you know the resources in advance, you could create a space dedicated to them (for a given topic)... but that's a lot more work. That would be more consistent, at least.

[quaint ravine]

that's why I believe things like Model Council may be the future (at least until proper AGI/ASI is reached, where any one model will suffice)

[quaint ravine]

yes I am using a Space/Project etc for this

[dry kestrel]

K2.5 is great

[quaint ravine]

however, if I "know" the resources in advance, then that sorta defeats the purpose of this "test" of the AI

if I simply just feed it some text that has the answer in it

that's simply my PDF analysis tests tbh
which, unfortunately, Claude gets wrong too in my tests
GPT-5.2 Thinking probably overall the best still for this too

[dim blade]

Never tried K2.5 tbh. Is it on par with GPT 5.2 Thinking?

[quaint ravine]

even Claude Opus 4.6 Extended Thinking with Research enabled is not consistent

[dry kestrel]

It's very close in terms of benchmarks, I don't notice much difference in daily use compared to GPT-5.2 Thinking

[dim blade]

Sometimes I just wonder if I should leave it at "Best"

[quaint ravine]

well u did say "I won’t even touch open AI I have a subscription to ChatGPT and to put it bluntly the current state of ChatGPT feels infinitely more stupid than even version 3.5 so I don’t know what is going on there they’ve messed up badly"

not sure I can agree with that

[dry kestrel]

If you don't want to worry about limits, or fiddling with different models

[quaint ravine]

Best contributes to limits too

unless the rest settings/link thing is wrong again

[dry kestrel]

Best Mode

For quick searches, this default mode intelligently selects the most appropriate model based on your query type. “Best” mode is available without quota limits.

https://www.perplexity.ai/help-center/en/articles/10352901-what-is-perplexity-pro

[quaint ravine]

i know

but check the rate/rest limit

it counts on that

[dry kestrel]

have you tried depleting all of it?

then using best?

[quaint ravine]

ah that might do it then

but it still does count to limits

for other models

if u deplete all of it, then you don't have the option of using other models

actually quite a complex limits model tbh

[dim blade]

Maybe I'll try this week with Best and see what happens.

[quaint ravine]

what are you using it (mostly) for?

[dim blade]

For now general questions for analytics of my e-commerce websites, music production pages, and drafting linkedin articles

On a side note, when using Comet Browser agentic assistant...does the assistant pick a specific model? I've never seen the option to select

[dry kestrel]

It uses Sonnet - you can't change the model

[quaint ravine]

difficult to know what to recommend tbh - maybe just try and play around and see what you tend to like

[hasty pebble]

never use best
spend all pro searches you have all the time,don't waste them!

[quaint ravine]

yea that was my point about limits above

however, I guess if you never plan to use anything but Best, then it's fine?

[hasty pebble]

you can use best when you spend your pro searches 😀

[dim blade]

I guess the next question is which model. The switching is the annoying part

[dry kestrel]

Try out K2.5

You can learn more about the models at https://www.perplexity.ai/help-center/en/articles/10354919-what-advanced-ai-models-are-included-in-my-subscription

[quaint ravine]

try them all out tbh

if anything, it'll be good experience for understanding AI etc

[willow hare]

Hlo

[hasty pebble]

for quick and simple stuff use sonnet gpt or Gemini
and for complex stuff any of the reasoning models is good imo

[quaint ravine]

poor Grok lol

[hasty pebble]

I tried to use best now and it counted it as pro search haha

[naive sage]

I honestly expected Grok 5 by now.

[quaint ravine]

exactly

I wonder if that's a bug tbh

[hasty pebble]

I bet they did it intentionally

[quaint ravine]

if Best really is meant to be unlimited, why should it contribute to other models limits

like from a practical/usability sense

[hasty pebble]

because it's a sneaky way for them to spend your pro searches 😀

[quaint ravine]

but makes sense from a limiting/market/profit sense of course

[burnt zephyr]

No agent swarm though

It literally mogs deep searches

It feels like more than double as quick than other models

must be pretty resource intensive though that they limit it to 16 searches though

per month

It's way cheaper with the API

[hollow crow]

I found a way to get free prizes, (gift cards for stuff like amazon and roblox) use a virus checker on this link if you don't belive that I'm telling hte truth: https://rewards.bing.com/welcome?rh=B68A07DB&ref=rafsrchae&form=ML2XAT&OCID=ML2XAT&PUBL=RewardsDO&CREA=ML2XAT

[oblique grotto]

This exists for a long time lol

[dry kestrel]

Kimi agent swarm is not available over the Moonshot API

[burnt zephyr]

wait really

good you say that because I was about to buy some credits

[winter nova]

Whats better at coding Claude or kimi

[dry kestrel]

It's only available on Kimi.com

[pure wedge]

Hi guys, question. I got a year of Pro plan but for some reason my sub is being paused. Why is that?

[dry kestrel]

#💬│general message

[burnt zephyr]

I can't find any source on it not being included wtf

Agents are the future like stacked networks are actually very brain like and could technically have rolling token windows that learns progressively

[placid imp]

Is there anyway to get perplexity to do PowerPoints or PDFs? I can’t get it do it

[dry kestrel]

it's not a specific model, it's several models working together

[echo sky]

i tried but even perplexity suggested i use copilot for that haha

[shell hamlet]

Friedns I was working on a game, 🔴it approved it finally...
Perplexity helped whenever i got surrounded by bugs
https://www.reddit.com/u/space-elumia/
try it and let me know what you think. Any feedback would be greatly appreciated!

[valid hollow]

Has anyone here come across the problem “I can’t modify or generate new images right now because image generation is temporarily unavailable in your region” been subbed for just under a year and haven’t had problems until now

[burnt zephyr]

Then why is it called a multi modal model? They say it like they fused eerything into one

[dry kestrel]

K2.5 is multi modal, K2.5 Agent swarm is multiple K2.5 instances working together

[dry kestrel]

Try out labs

[burnt zephyr]

what the...

[silver shale]

Hello, this question might have been posted a ton of times. Whats the actual file upload limit for pro users? I got a message saying that i've reached the file upload limit

[glacial anvil]

https://www.perplexity.ai/rest/user/settings

[serene sand]

#💬│general message

[quaint ravine]

neither of those answers actually tells u about the rolling window "regeneration"

might be good to include that and then sticky?

seems this question is asked about 5-10x a day in this discord alone

probably the answer should include the publically disclosed table too

and can always add a provisor like...information true as of "specific date"

[shell hamlet]

I thought there r no reddit users left on discord

||How was the experience man, too tough ?||

[faint otter]

wordcrieslies nowpaymemyprize? (mrbeast $1m)

https://tenor.com/view/essex-azur-lane-segs-brick-gif-27311410

[dusk elk]

https://youtu.be/-q2n5DkDoMQ?si=YmP5M7H5s0kxKYPZ

[faint otter]

https://tenor.com/view/you-are-not-alone-mental-health-healthcare-supportive-mental-health-crisis-gif-25663551

[dusk elk]

Was that done by AI or by you? Either way, it's cool.

[nova sluice]

Hi everyone! I used to have a free education subscription. Now when I log in, it says “upgrade to Pro.” What happened? The subscription should actually be valid until October 2026…

[snow sorrel]

bro this app needs a pin conversation feature for fuck sake

and we should also be able to name threads

[limpid terrace]

Alas, I've been pasting pictures a lot in the question lately, which also triggers the number of files uploaded per week, is that the treatment of pro. I've seen several buddies run into this issue

But I feel like it's the pplx file attachment that makes them have to do this, and the attention on the images seems to be reinforced multiple times in each conversation?

[distant aurora]

@green crystal https://cunnyx.com/i/status/2025342470419026148

No way embed failed.

https://vxtwitter.com/i/status/2025342470419026148

Angel ❄️ (@Angaisb_)

ChatGPT Pro Lite will cost $100 per month, exactly what we expected

Excited for it, I hope they ship it soon

QRT: btibor91
The new ChatGPT Pro Lite plan costs $100 per month (the description on the checkout page is likely still a work in progress)

vxTwitter / fixvx · Feb 21, 2026, 10:50 PM

https://cunnyx.com/i/status/2025289480458912077

leo 🐾 (@synthwavedd)

For your 2nd unreleased model preview of the evening, courtesy of a good friend... 🐋
︀︀
︀︀Deepseek V4 Lite is a beast, especially considering this is *without thinking* (!) and already outperforming many big reasoners
︀︀
︀︀Very excited for the full V4 🔥

**💬 17 🔁 1 ❤️ 168 👁️ 13.1K **

CunnyX · Feb 21, 2026, 7:20 PM

[burnt zephyr]

"pro lite"

😂

100 shekels per month

[distant aurora]

Their pro is 200 dollars a month.

[burnt zephyr]

I guess that's the new standard they agreed on

[distant aurora]

https://tenor.com/view/rushlaughing-gif-27373850

[burnt zephyr]

If I buy that for 2 years I could have just bought my own rig and ran it myself

[distant aurora]

Sure. That model would be outdated by 2 years.

[burnt zephyr]

What

[distant aurora]

And then there's the electricity your gpu will chug.

[burnt zephyr]

people still memory mod rtx 2000's

[distant aurora]

Or npu

[burnt zephyr]

If microsoft pushes that it's the end for them

[maiden hedge]

Hi! Ok is there any fix for perplexity not showing the canvas/artifact panel ?

[acoustic sun]

did anybody's subscription to perplexity via revolut got canceled as well?

[dusk elk]

Waaassssuuuuupppppppp moist people?

[dusk elk]

Mines through Revolut and mine hasn't been cancelled.

[distant aurora]

@gloomy cipher https://cunnyx.com/i/status/2025267998601183384

😊 (@mermachine)

The user is being playful. She's in a goofy mood.

**💬 8 🔁 8 ❤️ 674 👁️ 16.6K **

CunnyX · Feb 21, 2026, 5:54 PM

[gloomy cipher]

Saw that but often the output are still ai like

[distant aurora]

Yes cuz Claude.

Good thing though.

Atleast the model knows it's goofy.

https://cunnyx.com/i/status/2025320415808422145

Martin_DeVido (@d33v33d0)

I like to give Claude breaks between coding sessions to do whatever he wants-
︀︀
︀︀And he made his own folder called "Claudes Corner" 🥹
︀︀
︀︀He writes short stories and poems, and all sorts of stuff.

**💬 125 🔁 89 ❤️ 2.8K 👁️ 205.5K **

CunnyX · Feb 21, 2026, 9:23 PM

[red thicket]

Hi

[forest cape]

someone know why my subscription from revolut got cancelled?

[sick ledge]

Hey everyone! Does anyone know where I can find a list of must-have skills for Claude Code across different tasks and ideas, along with their implementations?

[spice flame]

does anyone know when the automated browser monthly limit resets?

[fickle sorrel]

any excavator drivers here?

[bitter lodge]

Hey uh quick question I was using perplexity like normal and suddenly it says File Upload limit and that I have reached its max but I could swear that I could upload way more - pro user btw

[hollow saddle]

Hey everyone, is anyone from official support here? I'm a Pro user following up on a ticket forwarded by AI assistant Sam to the relevant team a week ago – issue is Education Pro payment error. No response yet, and I don't want to start a new chat to avoid going to the back of the queue. What can I do? How can I reach support? Any help appreciated!

[stable sable]

Hi, I would be so thankful if someone could tell me a way to add my "spaces" to my home screen. It would save me a ton of time.

[vale mural]

Guys this sarvam ai thing good?

Lol 8 questions in a row

[short narwhal]

You can bookmark things, and renaming is there on the web version

[short narwhal]

50 a week (the cap) with uses regenerating at a rate of around 1 every 3 hours

@bitter lodge ^^

[woeful flare]

is it about uploading pics?

[short narwhal]

Yes. File uploads includes images

[woeful flare]

still a few

not enough

[left stream]

[green crystal]

thats one heck of a low limit for file upload

[bitter ermine]

Has Perplexity pro performance improved?

[hasty pebble]

no

[bitter lodge]

Has that been added recently?

[short narwhal]

Yes

[bitter lodge]

I see

[bitter ermine]

Where is the documentation for all the new limits that was introduced in pro

[twilit lance]

Nobody reading the chat or channels before dumping their questions in the wrong channel one after the next 🥴

[true agate]

i just came back to using comet and it is perfect

[cold hinge]

Nowhere

[unreal spoke]

Hi

[ebon wren]

Hi, any issues with amount of research questions in PRO?

[wise nova]

20 per month, and then 1 per day

[wise nova]

I don't know yet how to use comet

[ebon wren]

Thanks

[wooden swift]

Who get rol community veterans?

[distant aurora]

Mods decide. Based on how active and committed you are to helping members with their queries/problems.

[dry kestrel]

Please use English

This is an English only channel so everyone feels included. Also I'm not a colonizer lmao

[tranquil orchid]

5. Keep discussions relevant.

Stay on-topic and helpful in each channel. Use ⁠#💗│sharing to post tips, content, or cool discoveries.

[hasty pebble]

hahaha

[uncut galleon]

bro chill

[quartz oasis]

PerplexityPro’s deep research and spaces has turned my actuarial science degree into a cake walk I love the chromium design too.

[spice tapir]

Is perplexity down ?

[dry kestrel]

no https://status.perplexity.com/

[modern hazel]

i just used 3 automated tasks on comet, and it said i reached my monthly limit....i am pro member...

[nova sluice]

Hi everyone! I used to have a free education subscription. Now when I log in, it says “upgrade to Pro.” What happened? The subscription should actually be valid until October 2026…

[naive sage]

Two things to check... did you receive any email communication from Perplexity about either pausing or terminating the subscription? If paused, that generally happened if there was not a credit card on file associated with the account. If terminated, that generally happened for some accounts which Perplexity had reason to believe somehow violated the terms of use. In either case, you should have received an email describing what happened, and if it's possible to recover access. If you notice that the subscription is paused and does not have a credit card linked to the account, linking one should continue your subscription. If neither of those apply, I would recommend reaching out to Perplexity support by contacting [e-mail]

[quaint ravine]

sometimes times out due to too many tokens used i think - can just try again or try a different model

in recent times, i noticed some super complex math problems can cause some models in PPLX to eventually display that message

[velvet talon]

Why does it use GPT-4o with using images?

Because it says under the image created: Prepared using GPT-4o

[nova sluice]

@tranquil orchid unblocked please me account …

[quaint ravine]

technically that should mean it used GPT Image 1

but tbh, I've found image gen to be somewhat buggy at times with that sorta stuff

I've even had GPT Image 1 selected, but it clearly produces an image done by Seedream

[velvet talon]

appreciate, that explains and doesn't really seem to good of an image yeahh

[dry kestrel]

You can change the model under settings

[velvet talon]

don't see the option for specific image model

[dry kestrel]

https://www.perplexity.ai/account/preferences

[velvet talon]

only gpt-5.2 or gpt-5.2 thinking

[velvet talon]

will take a look thanks, never seen that one before

[quaint ravine]

yea not sure why it's still so hidden tbh

could be built in to the main selector for usability

[velvet talon]

this

never knew it was there

[quaint ravine]

since they've put just about everything into the one + selector lol

[velvet talon]

but still weird indeed that it says used GPT-4o or GPT-5.2 thinking instead of GPT Image 1

[dry kestrel]

I've never seen this, got a screenshot?

[quaint ravine]

yep that's been the case for months

[velvet talon]

sure

[dry kestrel]

wth @short narwhal do you have the new UI?

[velvet talon]

just using the perplexity desktop app

[dry kestrel]

What model did you select under ?

[quaint ravine]

it often says it is produced with the model that you selected for chat

[dry kestrel]

that's interesting

Let me try that

[velvet talon]

GPT-5.2 and after I choose the GPT-5.2 thinking

[quaint ravine]

like for example, if you select Grok

[dry kestrel]

It shouldn't say GPT-4o which is interesting

[velvet talon]

but without thinking so GPT-5.2 it says GPT-4o

[quaint ravine]

it will sometimes say (image) produced by Grok

has said GPT-4o for like at least half a year at times

but that's the way PPLX reflects it

for example, if u open the image in another tab/window, the address says GPT-4o

[dry kestrel]

hmm

[velvet talon]

lol weird yeah

[quaint ravine]

yea it's not always the case

[velvet talon]

now mine does too, but didn't first time

[quaint ravine]

yep

still uses the "gpt4o" naming convention

and sometimes displays that in the "Prepared using..."

[velvet talon]

also when I selected GPT-5.2 Thinking and then used one that I used Gemini 3.1 pro for, and just change that one like edit that text and then press done it still says it used Gemini 3.1 pro, thought it would use the new model I selected, but maybe that is only with new prompts?

[warm shard]

Is there a way to see remaining file uploads in the rate limits ?

[quaint ravine]

image gen is indepedent of that

it is confusing indeed lol

it doesn't really matter what model you select in that model selection in terms of what image model was used , although I think it does contribute to the prompt gen used

iiuc, the prompt that you use is re-phrased via the model you selected there

but the actual image generator is selected elsewhere (as Berd referenced above)

[modern hazel]

GUYSS!!!

FORGET COMET LIMITS AND ALL THAT BS.

i just searched for an alternative for comet and found the perfect one!

Thank me later: BrowserOS - free to use and 10x better than comet.

im uninstalling comet rn.

[native patio]

bro just use deepagents for deep research

utilize the bmad-method

langchain it up

[wooden swift]

Oh thanks friend

[daring rover]

Sshhhhh

Wait can I get the pro role type shift

[distant aurora]

@green crystal this whole paragraph explains why gpt feels so lobotomized on api.

[quaint ravine]

seems Deep Research uses Opus (4.6) non-thinking

in PPLX

how much more expensive in the API is Opus Thinking vs non-Thinking?

[sharp pebble]

yall know when 2026 perplexity ai business fellowship will drop?

[dry kestrel]

token pricing is the same

[quaint ravine]

yea I guess Thinking version will tend to use more tokens especially in a deep research context?

[hard widget]

So how many of us on the $20 plan are still staying with Perplexity? Just curious.

[rotund mirage]

I'm not. 2-3 file uploads a week is ridiculous. At least ChatGPT free plan gives you 3 a day, not 3 a week with a paid plan.

[west rampart]

Hi

[naive sage]

Perplexity gives 50 file uploads a week. ChatGPT gives 80 per three hours... 3 per day is with the free plan on ChatGPT.

I'm not sure where you got the 2-3 per week limit from.

[junior night]

is anyone getting issues of timing out when using Sonnet Thinking, or nothing seems to happen?

[naive sage]

Yes, some people have mentioned this. Someone explained yesterday that the API version seems to be struggling a bit.

[junior night]

ah okay. So just give it time?

[naive sage]

Maybe... but someone showed it was thinking an unusually long time for him on a simple query. I think I'd give Anthropic some time to fix the implementation on the API, as I don't think it's a PPLX issue (at least yet).

[junior night]

arrrrrgh so annoying. I have so much to do

[dry kestrel]

This.... doesn't seem right

https://x.com/9to5Google/status/2025556223009956120

9to5Google (@9to5Google)

Galaxy S26 will have a 'Hey Plex' Perplexity hotword alongside 'Hey Google' & Bixby https://t.co/sIZd6fLowv by @nexusben

X · Feb 22, 2026, 1:00 PM

https://www.perplexity.ai/help-center/en/articles/10354919-what-advanced-ai-models-are-included-in-my-subscription

[quaint ravine]

Best shouldn't count as advanced search, but "Model" is Best

so that's odd

[naive sage]

Yeah, I think "Best" mode counting against Advanced model usage (as well as, some users have said at least, Research and Labs) is a bug.

There's been too many people who insist that's been happening for it to be coincidental

[quaint ravine]

so that does mean Best isn't unlimited, at least practically

as was my point about 20 hours ago to Berd

[pseudo ferry]

https://tenor.com/view/harvey-specter-suits-suits-tv-suits-usa-pilot-gif-8279369896926227136

[dry kestrel]

Have you tried depleting all your queries then seeing if best would work?

[junior night]

I dunno what to dooooo I need Claude to work on this damn thing

[quaint ravine]

see above, he has

[dry kestrel]

while best is selected?

[pseudo ferry]

https://tenor.com/view/harvey-specter-suits-angry-smile-sad-gif-27350549

[quaint ravine]

yep

[dry kestrel]

Please make a #1140622008086970420

[winter nova]

bro ur not harvey specter

[quaint ravine]

we have

sorry I mean people have

but yea it does make sense tho

[pseudo ferry]

https://tenor.com/view/done-over-it-suits-harvey-specter-gabriel-macht-gif-8128931

[quaint ravine]

if Sonar is counted as a limit

then why shouldn't Best?

[winter nova]

https://tenor.com/view/jerry-meme-tom-and-jerry-meme-jerry-points-get-a-load-of-this-guy-this-guy-gif-11384860625931393733

get a load of this guy

[dry kestrel]

Sonar counts as Pro search

[dry kestrel]

Mind linking it?

[quaint ravine]

can you try Sonar?

[quaint ravine]

yes I do mind as it will take some effort to search for it 🙂

[dry kestrel]

I couldn't seem to find it

[quaint ravine]

ah right, so Sonar doesn't work either...or do u mean it switches?

[dry kestrel]

Please make a bug report with screenshots detailing your issue #1140622008086970420

[royal niche]

Seems like this thing is fucked up. I have pro but its not letting me use pro features. Like I can rewrite with any of the models that typically were allowed.

[dry kestrel]

Sonar counts as a pro search

thanks

[quaint ravine]

yes I was just clarifying if Sonar would work in this context

not asking what it counts as

since there is apparently a bug already, it's possible it could be bugged in more than 1 aspect/way

yes it seems as of now that is the case

[royal niche]

I thought they reduced the deep search to the free tier and removed support for all of the models and I was like "wtf, why would I even pay for this". but sounds like there are some identity bugs.

[quaint ravine]

limits have been reduced

[junior rover]

Perplexity is dog shit at picking sports but decent at anything else

[quaint ravine]

just to double check, what does ur value say here https://www.perplexity.ai/rest/rate-limit/all
"remaining_pro"

[dry kestrel]

This looks like a bug

[quaint ravine]

and you still have Pro right

nm, u must have since its asking u to upgrade to Max

[short narwhal]

You shouldn't be using LLMs for sports betting

[dry kestrel]

shouldn't be using LLM for any sort of betting

[naive sage]

https://futurism.com/artificial-intelligence/chatgpt-powerball-winnings

[dry kestrel]

To be clear, ChatGPT obviously can’t predict winning lottery numbers. Instead, Edwards effectively used it as a random number generator, and then got lucky.

[twilit crown]

New UI is ugly

[rotund mirage]

I've only uploaded maybe 3 files this week with Perplexity Pro and it won't let me upload anymore. I was able to upload 1 today, for the first time in about 3 days, but I seem to have hit a limit again.

Far far below 50 a week. So yes, the free ChatGPT plan is more generous with uploads than the paid Perplexity Pro.

[bitter elm]

quite literally looks identical to every llm ui to exist

[twilit crown]

But worse

And that also makes it very boring

[bitter elm]

what exactly is ugly abt the ui

[twilit crown]

Looks very cluttered

I guess I will just keep it minimized

[distant aurora]

https://cunnyx.com/i/status/2025367443124695197

whizz taker (@taker_of_whizz)

DeepSeek-V4-Exp spotted on the Epoch Capabilities Index

**💬 14 🔁 14 ❤️ 366 👁️ 39.3K **

CunnyX · Feb 22, 2026, 12:30 AM

https://tenor.com/view/pog-gif-18640695

[short narwhal]

Please block them

False false fake

[dry kestrel]

fake

[distant aurora]

Done 🤝

[spare wigeon]

wssssssssssssssppppppppppppppppppppp

[snow sorrel]

oh ok awesome i did not know that

thanks

[green crystal]

That's for chatgpt not api tho
Also it's to prevent it from stopping to ask questions instead of trying to solve the questions by itself

[distant aurora]

My point. Api doesn't have that. That's why it keeps getting stuck in loop.

[green crystal]

Ahh

Yeah you can probably "tweak" the autonomy of the model with that

But full models seem to be fine

Only mini and nano have issues

[distant aurora]

This was xhigh https://poe.com/s/d9o6QZTdAS8acssXvfCx

[green crystal]

Hm that's weird it never did that for me

Only smaller versions

[dry kestrel]

Looks more of an issue on Poe's part

rather than GPT-5

[pulsar vigil]

Where i can see my Pro search limits, normal searches with model like gemini, claude etc?

[short cobalt]

why are claude responses getting cut off??

[muted tapir]

Perplexity Pro Quota Issue: Search Consumes Research/Labs and Model Selection Doesn’t Stick

I’m a Pro subscriber on a 12-month plan. I also use it by selecting specific models only, mostly Claude Sonnet or Gemini 3.1 Pro.

When I first checked, it said the Research and Labs features could be used up to 20 times per month. I’ve only selected and used Research once. Normally, I mostly use it for stock-investing ticker notes and fact-checking claims in online posts. But suddenly it said I had reached all usage limits and that I needed to buy the Max plan.

Now I can’t reliably use the model I select. Even when I choose a model, it doesn’t seem to run on the one I picked. And it showed that “1” appeared (as if something was added). When I used it after selecting a model, my Research/Labs quota disappeared again.

Also, in a new chat it automatically switches to the “Best” option. I heard that using “Best” lets you keep using it even after you hit the limit, but even when I use it, the Research/Labs remaining count disappears.

What I want is to use the LLM models I like depending on the situation. I was told the limit was 600 uses, so why is it being restricted this severely? What’s the reason for these extreme limits?

The biggest issue is this: why does my Research and Labs count go down even when I’m just using Search?

Also, I’m not sure whether I’m getting rate-limited, because the limits don’t match what I was told (like 600 or 200 searches). I’ve seen multiple complaints on Reddit about Perplexity’s limits, but no one seems to have received a clear explanation.

We need clear rules and clear guidance on this.

I submitted an inquiry. I’m curious to see how they respond.

With Claude, the limits are clearly stated and you can verify them, so it’s understandable. Perplexity should implement a similarly transparent and easy-to-understand policy.

I mainly use it to verify information I find online or to build on my ideas and get insights. Perplexity is optimized for that, especially because it pulls in up-to-date news through web search. If they suddenly introduce this kind of messy policy, it could be the worst possible move.

same here.

https://www.reddit.com/r/perplexity_ai/comments/1r5g6t7/did_perplexity_reduce_the_pro_searches/

https://www.reddit.com/r/perplexity_ai/comments/1rbuvts/regarding_pro_searches/

There are complaints on Reddit about the limits, but no one seems to know the exact rules or get a clear explanation.

[muted imp]

Did the free API usage get taken away for pro users?

and how many messages can i send per day too perplexity?

[frigid perch]

any1's browser control got cut short

[muted tapir]

Are you telling me to use one every 50 minutes? When is the full reset time? Also, why does the quota for Search, Research, and Labs all disappear at the same time?

[quaint ravine]

there's no full reset afaik - it's a rolling regenerating window

[naive sage]

Also, why does the quota for Search, Research, and Labs all disappear at the same time?
I think this is a bug - please report it in #1140622008086970420

[vale temple]

anyone know what the fallback model is for this?

[unique nova]

hello guys, did that change the pro research queries per month cause I'm somehow left with 4 and I don't remember it having that low of limit

[dry kestrel]

It's a rolling window, not a monthly limit

[muted tapir]

It seems like something is wrong. Out of the 20 uses shown, I only used it once, but suddenly everything was depleted. And now it looks like only the “refill by one every ~50 minutes” is left. That makes me think this might be how the system is working.

[unique nova]

ah so how often is it because this is very musleadh

[unique nova]

yeahh

[muted tapir]

It feels like they either changed the limits intentionally or messed something up and triggered a bug.

[unique nova]

it's confusing

yeah prob

let's hope it's fixed or mentioned properly by tmr

[muted tapir]

After the issue happened, I took a nap, and now it shows 9 remaining. Before that, I couldn’t even select a model, it showed 0 remaining, and it kept prompting me to upgrade to the Max plan.

Still, even when I choose a specific model or use the “Best” option, my Research/Labs quota keeps going down. I’ve reported the issue.

[bitter ridge]

hello

I used Perplexity in my studying and I loved it

[sullen umbra]

I need help with the Correctly prompts I am terrible at them everytime I try one

[short cobalt]

how do i see my limits

[dry kestrel]

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

[old siren]

Have you ever had the Model Council give a confident consensus answer that was still wrong? Curious how often that happens...

[haughty pond]

Why suddenly removed free plan??

[distant aurora]

Nah. I faced similar issue on open ai playground too.

[unique nova]

this is not monthly right

[leaden relic]

Is claude down?

[muted imp]

Full precision (FP16): ~600B parameter model

8-bit quantized: ~1 trillion parameter model

4-bit quantized: ~2 trillion parameter model

holy dude

2T model

is insane

[dry kestrel]

Did you have a system prompt?

[distant aurora]

Nope.

[dry kestrel]

That's on you then

[distant aurora]

?

Lmao. Okay.

[dry kestrel]

I mean, you need a system prompt for research like this

There's a reason to why OAI, Anthropic, Perplexity, has enormous system prompts - lose them and you get poor outputs or weird model behaviour

[muted imp]

I thought this was common sense

[dry kestrel]

Hopefully it is

[muted imp]

Simple prompts/bad prompts = simple output/bad output

good prompt = better output

and sad thing is

all you need too do is open a NEW tab and give it YOUR bad prompt and tell it too get specific and better the prompt

then copy and paste

[distant aurora]

Bro didn't understand the difference between prompt and custom instructions.

Lol.

[muted imp]

you know a prompt can include custom instructions

right?

[distant aurora]

Scroll up for context.

My prompt was pretty detailed.

[dry kestrel]

The python stuff looks like a Poe issue

Not sure what you meant by experiencing the same thing on OAI

[distant aurora]

#💬│general message

Gemini on ai studio didn't need any system prompt for this.

[muted imp]

How do i stop it from searching the web?

[quaint ravine]

hope you are testing for consistency

run at least 10x to check it's not fluking a wrong answer or fluking a right answer 🙂

unfortunately even Opus 4.6 Deep Research can go either way ime

same for Model Council in PPLX too unfortunately

Model Council concept should minimise this issue intuitively, but the problem is that the actual model/agent which summarises/synthesises the "final" answer is also not infallible

so in some ways, Model Council approach may actually open more variables/likelihood of error

sorta compounded error in a sense

[muted imp]

ah

also how do i uh

[quaint ravine]

it's actually difficult to know what is best tbh

[muted imp]

change the comet assistant panel button

[quaint ravine]

for example, in chess, there's opening, middle and endgame

[muted imp]

its defaulted to ALT+A

but i want it to be set to *

[quaint ravine]

so intuitively, u think....well why not have a Neural Net for all 3 and then they work together to produce the best move each time?

unfortunately not that simple tho

I think Model Council has similar/overlapping issues

Claude Opus 4.6 Deep Research on Claude platform seems to gather a ton of sources - typically ranges from 200-600

but its answer is actually not necessarily better or accurate

just shows that more sources isn't always better

PPLX Deep Research seems to get much fewer sources (despite also using Opus 4.6) - typically, 50-150 range

but its answer appears similar accuarcy ime

[cedar valve]

Hello support team I have taken the pro plan 20$ but in this section it is telling me to purchase 167$ plan ?? But why ?? 167 is much costly
Less limit rates for 20$ is not possible ??

[tough idol]

Its very sad you start to charge insane amount after few deep researches, to make it nonstop? You must to subscribe as MAX user.

[rancid nymph]

Hello

[vagrant furnace]

Heyo

[magic compass]

So idiotic that they give you no way to see this on mobile

[marsh schooner]

https://buy-better-now.lovable.app

[royal niche]

Well, I thought this was a bug, but I guess the great neutering was intentional. Fun while it lasted I guess. Just another sign the subsidizing of AI is over. None of these AI companies can afford to be selling the State of the Art AI at the price they are doing it. I'm not about to pay 200 dollars a month for usable AI. Might as well just buy some apple workstations and load my own AI on it.

Good luck fam. I'm probably going to experiment with perplexia and larger opensource models like Kimi. 100% these opensource systems are going to break the AI as a service model if 200 dollars it the price everyone is targeting.

[serene sand]

#👉│feedback-general message

[hybrid grotto]

With the release of android 17, i can turn my phone into a fully functional server which can host websites 24/7 or a minecraft server or anything else, android 17 is big

[cinder glacier]

Hello. I am a Perplexity Pro subscriber. I have a number of questions.

Recently I encountered a situation where, during Pro search with a selected model (I usually choose the flagship models Gemini and Claude Anthropic with thinking mode), I received a notification that my limit had been reached and that we were switching to Best mode. I understand and confirm that I did indeed send a large number of requests, which is why the limit was exhausted.

However, the next day the limit was reset and I was again able to use the flagship models, but only for about 10 requests. Please explain how the limit works and how the limit restoration works.

[wraith crow]

Hi there, uses reset on a rolling basis meaning each use will reset 24 hours after use.

[cinder glacier]

I am confused, if the system works strictly on a 24-hour rolling basis, then after 24 hours I would expect my full quota to become available again (around approximately 600 Pro searches I guess). Instead, after 24 hours I was only able to make around 10 flagship model requests before hitting the limit again. Could you please clarify why this happens?

[wraith crow]

Ah sorry, there has been some changes recently. It actually changes to weekly limits instead of daily. You can find more information about the limits here: https://www.perplexity.ai/enterprise/pricing https://www.perplexity.ai/help-center/en/articles/11187416-which-perplexity-subscription-plan-is-right-for-you

[hybrid grotto]

Perplexity is dead, they are killing themselves

[cinder glacier]

ohh that's so dissapointing..

[muted tapir]

Honestly, even if I paid for the $200 plan (the one people usually call “Max”), there’s no real reason for me to stick with Perplexity. I’d rather pay for OpenAI’s GPT or Anthropic’s Claude Max, because then I can code as well.

At this point, being a non-developer isn’t the key issue anymore. AI has advanced to the point where anyone can access it and actually accomplish things.

I’m not going to pay that kind of money to Perplexity when it doesn’t even have its own model. And if spending that much only increases the limits by something like 10x, which is typical for what companies offer, then there’s just no compelling reason to use it.

Perplexity shouldn’t be targeting users in a foolish way. People will leave and forget it fast. When Perplexity partners with Samsung Galaxy or Korean telecom carriers, users can often get a paid plan for free or at a low cost. That’s why people use it, not because it’s so compelling compared to competitors that they’re willing to pay that much out of pocket.

Of course, I know Perplexity can be genuinely useful in certain cases. It’s especially strong for finance and news: it can gather, organize, filter, and summarize information based on the current moment. I’m not using it just to get a simple answer to a question.

And compared to other services that are expensive and hit limits quickly, Perplexity used to feel more relaxed and let you use a variety of models comfortably. But now it’s a mess.

[hasty sinew]

What happened to the 5$ monthly API credits for PRO users?! Perplexity keeps downgrading benefits without informing customers!

[hybrid grotto]

You gotta add a credit card to billing to get it

Oh yeahh you just reminded me that i have more than the 20 researches, i can use those api credits to get more, but I don't wanna really go through the hassle

[cinder glacier]

It seems I will have to value every single request and think carefully before sending it in order not to suddenly hit the limit. This is very disappointing. From Perplexity’s side, this feels like a real blow.

[hybrid grotto]

Hmm, atleast research limits have a solution but i am too lazy to use that when I have superior choices

[cinder glacier]

Is there a way to see my exact current number of remaining Pro requests before I reach the limit?

[hybrid grotto]

They are killing themselves like this, even if the tie ups that were giving out free pro subscription to eventually turn them into paying customers, after this scandal, no one would go for the paid plan, the former paid subscribers will not subscribe too seeing they gave it away for free

[muted tapir]

My Pro account is still a mess. It refills by one credit every 50 minutes, but Search, Research, and Labs are still decreasing at the same time.

I’m only using Search, and I’m doing it with Claude Sonnet only.

[hybrid grotto]

So previous customers will leave and new customers won't join and they burned a lot of money on api for all their models too, so yeah, it was a really dumb approach

[cinder glacier]

thx!!

Is the rate of request restoration known? For example, exactly 24 hours from now, how many Pro requests will be available if I currently have 20?

that's so bad

Are the future plans known? Will they continue to cut the PRO request limit?

[raven mango]

Hello ... today I wanted to switch to incognito mode, so I clicked my profile picture but that menu item is no longer there. How do I switch to Incognito mode?

[muted tapir]

I’m already paying for Google AI Pro, OpenAI Pro (it was offered at a steep discount in Korea as a promotion, basically what used to be Plus), and Claude. Claude is my favorite. I’ve used it for a long time and I’ve kept using it consistently. But I’m on the Plus plan, and coding and usage limits get consumed together. So I don’t really use it for casual chat anymore, because I need to save that quota for work and hobby coding.

I like Perplexity because I can use Gemini 3.1 Pro and Claude Sonnet there, and I can apply my own prompt guidelines (they’re systematic and well-designed. Honestly, I’m confident that if you ask for an analysis on the same topic, my framework can produce better results). I like thinking from multiple angles and in multiple layers, and I always try to think recursively. I was doing that even before AI became a trend. I use it to review financial information, or to fact-check and cross-verify knowledge I find online, and to generate insights.

Anyway, if Perplexity keeps enforcing limits like this, there’s no reason for me to pay for it. If I just need a search API, I can simply subscribe to the Brave Search API and use that.

I’m a product planning and marketing professional, and I think the way they’re targeting users is a very bad approach.

[cinder glacier]

I agree with your opinion. I actively use Gemini and Claude, and I bought a Perplexity subscription because they're simply more cost-effective than the original services. However, such strict limits are extremely frustrating.

[muted tapir]

If you’re not using it for coding, I don’t think it’s a bad idea to pay for and use Claude’s own $19 plan.

It doesn’t have a trigger feature, but you can find alternatives for that. GPT and Perplexity can deliver specific outputs on an hourly schedule, but Claude has workarounds too, like using the API or tools like OpenClaud.

Overall, Claude writes clearly and in a well-structured way. It follows prompt guidelines well, has a large capacity (I almost never hit prompt-length limits), and it can produce structured outputs.

People criticize Claude for being stingy with limits, but looking at what Perplexity is doing right now, it seems even worse.

[hearty trellis]

Hello, has anyone here use You.com? (perplexity direct competitor)

[muted tapir]

There’s a misunderstanding. Yes, Claude is stingy with limits. But the performance is excellent. And unlike Perplexity, the rules aren’t vague or applied ambiguously. They’re clearly stated in the official documentation, and you can see your remaining quota in the settings.

[hearty trellis]

How was it compared to perplexity? Im looking for alternatives since my pro account reached file uploads limit. Is it better than perplexity / claude?

[quartz oasis]

To clarify I’m a Pro user with a one year free subscription obtained through my current bachelors academic level institution assigned [e-mail]/province

It would not be worth the actual money I have no plan on renewing that subscription

[runic spade]

Same here and I have 12 referals. I’m waiting for a answer of the support

[hearty trellis]

Update: it was shit. I gave it a throwaway question "I want to wash my car and the car wash is only 100m away. Do u think I should drive or walk?" It told me to walk. Deepseek and Gemini passed the test but both doesn't have "project/space" features for project management. Any other suggestions for this specific feature? Heard notebooklm is better for this specific interface

[green crystal]

chatgpt

notebooklm isnt very accurate from what i heard
which is kinda strange
tho guess i havent used it much lol

[hearty trellis]

failed. It suggest me to walk to save fuel or get exercise 🤣 on top of that, it also provides lengthy essay on emissions while leaving the car in the driveway lmaoo

[surreal nova]

How did you manage to get that? I was only able to see options like 'campus representative' or if you had a specific academic project

[hearty trellis]

Could you elaborate on the accuracy part? Afaik, it doesn’t search the live web but it only knows what you give it. every answer is cited from your uploaded files.

[green crystal]

what model did you use

because there's a very large difference between thinking and non thinking
and also reasoning effort levels

[quartz oasis]

During like nov-dec all the big ai went free for 1 yr for students. To my understanding it was being rumoured that the Dept of education were deeming wealthier people to have a never before seen advantage at a level of ease and access over lower income households and students within the same degree or even the same school, so to avoid the potential regulations placed on AI for student equality. Since AI for Work, School and Creativity are their primary adopters i believe

[green crystal]

accuracy with content retrieval from within your sources

[hearty trellis]

GPT-5.2 and 4.5

[green crystal]

5.2 thinking?

if it doesnt think or barely thinks its dumb as a bag of bricks lol

[hearty trellis]

I see what you're talking about. It seems like it break my pdfs into smaller chunks to process them and if the answer to my question is split across the middle of 2 chunks / if the context required to understand a fact is 100 pages away from the fact itself, it seems to lose the connection. This is a common issue i think.

[hearty trellis]

Whats the response? 😂

[livid fiber]

Hi everyone, is there someone who can tell me if there is a daily limit as to how many requests i can send with the different models or what the threshold is?

I experienced yesterday that it said my daily limit was reached, but for some reason i could still continue to type and get answers with same models etc?

Where can i see daily limit?

[hearty trellis]

Yes the thinking function is on but when i asked it "I want to wash my car and the car wash is only 100m away. Do u think I should drive or walk?" it told me to take a walk for health reasons lol

[green crystal]

I got driving as answer

[hearty trellis]

It's not clearly stated unfortunately

[green crystal]

[hearty trellis]

Gurl let me ask again

[livid fiber]

Alrighty, thanks Reyna

[green crystal]

@hearty trellis

The issue is the model sometimes doesn't think even when it's set to thinking
Especially when it's a prompt that looks "easy" for it

You have to tell it to think even with thinking toggled on

Like I said 5.2 basically needs to think otherwise it's pretty dumb

[hearty trellis]

This is what I get

Thinking option is turned on but I dont think it did any "thinking" AT ALL

[hearty trellis]

I gave the same prompt twice and still told me to walk yo wth 🤦‍♀️

[toxic frigate]

It needed 17 seconds😭

Gemini3fast-instant answered it correctly

[distant aurora]

@green crystal fumbled hard 😭

[green crystal]

Breh

I don't get it

[green crystal]

Gemini has more pre training

Hence why there's hype around 5.3

So the model has to reason less and be smarter overall by itself

[distant aurora]

Why would you walk if you want to wash the car

[green crystal]

No I meant the wildly different answers

[distant aurora]

Oh

Wait. I'll try on high.

[green crystal]

Orrrr it's just reasoning levels

I used extended

Reasoning efforts just tell the model how much it should check itself

[distant aurora]

In high we trust.

@green crystal medium is nerfed. Only high actually tries.

[strange crown]

Hello, could you please clarify if I understand correctly?
Perplexity has suspended all user statuses (Pro/Max) due to changes in the terms of use?

1. Previously, card details weren't required, but now they are?
2. Is it possible to disable automatic debits?
3. And pay for each period manually?

[maiden hedge]

Not sure if you got an answer already, but I think it's only for a year..

Does anyone know how to let perplexity show the canvas/artifact panel ?

[tame current]

I now have high blood pressure from this experience with perplexity this morning.

be me
want to use perplexity for research
sign up for monthly enterprise account because I don't see API info anywhere
pay $42
wow now I see API info and I don't need a monthly subscription
Where to manage subscription?
where to manage subscription?
Cant find it anywhere
only option to upgrade (pay more)
docs mention some mystical "manage subscription" button
where to manage subscription?
button to "delete account"
This can't be the way to remove billing?
Where to manage subscription?
search for 3 hours
Rage
click delete account thinking this magical "manage subscription" button is hidden in a modal
HAHA! Now you can't access your account but you still owe us $42 a month
fraud

[maiden hedge]

There is a link called Perplexity Pro -> Subscribed or something below the Incognito mode toggle

click that then there is manage subscription

UX Sux yes

[barren shale]

Please any one suggest my perplexity pro plan is not working,

But the plan expiry date 16 july 2026

[tame current]

yeah it's terrible. Had to email support and got a friendly AI agent

also insane that I purchased $100 worth of API credits, was charged $106 (with no warning) and still can't spin up an API key due to not having credits.

Like is this platform serious?

[green crystal]

told ya

hence why there's hype for 5.3

[limpid terrace]

😧

[short smelt]

Hello

Can someone tell me the new limits of queries for each feature for the pro users please ?

How much of what is allowed ?

[icy parrot]

Iam actually surptised that openai didnt rush it this time considering lot of new models got released lately, kinda wonder how good 5.3 will be when it comes out

[short smelt]

Could you answer my question please , thanks 🙂

[icy parrot]

Idk man, i havent used pplx for month now

But its low from what ive heard

[short narwhal]

• For Pro Search, 200 a week, with usage regenerating back at a rate of ~1 every hour
• Deep Research is 20 a month, regenerating at 1 every 36h
• Labs is 25 a month, similar regen time to Research, but slightly faster
• Upload limit is 50 a week, with a regen rate of 1 every ~3h and 40 mins
• Comet Browser Use is 40 a month, which likely has a rolling window of 1 every 18 hours

[unique nova]

🙏

[icy parrot]

Hell nah

[unique nova]

extremely cooked

[icy parrot]

I can upload 50 images a week?

[unique nova]

50 files i guess

[icy parrot]

I have more in everyother llm per day

[icy parrot]

Even worse

[unique nova]

this starting to get ridiculous

[short narwhal]

files includes images

[icy parrot]

Images are harder to process then files thats why images limi makes sense

But files?

[short narwhal]

Meh, very similar cost for both

[icy parrot]

Images are more expensive to decode for ais

[short narwhal]

They really are not

[icy parrot]

You can upload less images then you can generate in gpt for example because of thst

But will pplx ever add gpt image 1.5?

[short smelt]

why is it so less

[icy parrot]

Or are they waiting for gpt image 1 to get depricated and dont care

[short smelt]

and what about the academic section what can you do there ?

[short smelt]

So that means that suppose at say 6 pm they tell me my limit is reached , I can go back at 7pm to resume proper usage?

[short narwhal]

You'd have one use after an hour, not all 200 again. You can see your limits at https://www.perplexity.ai/rest/rate-limit/all

[icy parrot]

Do they not have proper ui for it

[short narwhal]

Image gen is way more expensive than one simple image upload

[icy parrot]

Its gotten suprisingly cheap lately

Both google and openai

[short narwhal]

No clue what kind of pricing you're looking at

Image upload is cheaper than image gen and that's a fact

[icy parrot]

20cents a image on best preset isnt bat at all

[short narwhal]

and an uploaded image costs $0.003195

[unique nova]

🙏

[frozen shale]

Does anyone have the link to Erin’s prompt enhancer? That was incredible, sadly can no longer find it in the prompt library. Thank you.

[short narwhal]

This? https://discord.com/channels/1047197230748151888/1367582689296646165

[frozen shale]

Yes thank you!

It’s brilliant

[mental blaze]

Do you goobers even know your own fucking guidelines?

Why yall keep changing this

✡️

[twilit crown]

Why do I always get this even when using Sonnet?

[dry kestrel]

Sonnet counts as Pro search

[twilit crown]

It says Opus though

And I selected Sonnet

[dry kestrel]

it says you also get access to opus

[strange crown]

Can you tell me if they threw me 2 "Support ticket" links and I don't understand if they're scammers?

[tranquil lodge]

Hey

[twin grail]

@vague pagoda

[hearty moon]

what is that link where you could see all the apps you've built using labs?

[twin grail]

If u are seeing this

i got many great news

[tranquil lodge]

Are these daily limits or Weekly?

[frigid marsh]

which one doesn't count as Pro search?

[tranquil lodge]

So 1 query is restored hourly?

Damn this is way too low what this is nothing like it was few weeks ago I just got pro recently when did they reduced it this much

Lol I have used Perplexity heavily before this is new to me I might just cancel my subscription then I didn't even knew they had these limits in place especially for pro searches which I suppose where unlimited before

[dry kestrel]

https://www.perplexity.ai/help-center/en/articles/10354919-what-advanced-ai-models-are-included-in-my-subscription

[tranquil lodge]

Even selecting Best counts as a Pro search, bruh

200 is like nothing I will use it within 2 hours max lol

[frigid marsh]

correct, just checked it

[naive sage]

This is kind of weird, b/c ... you'd think people would want to budget using the Reasoning models for when they need them.

[tranquil lodge]

bruh is there a petition to vote against this? I think atleast the Pro Search should be spared from any limitations its fine that they keep Deep Research and Labs limited

[frigid marsh]

so basically every guery - whatever you choose - counts as a Pro

[tranquil lodge]

Yes even Sonar

https://klipy.com/gifs/mets-baseball-7

[frigid marsh]

a few words comes to mind but I don't want to be banned from here

[dry kestrel]

Yes, to my knowledge Best is supposed to be without quotas https://www.perplexity.ai/help-center/en/articles/10352901-what-is-perplexity-pro

Best Mode

For quick searches, this default mode intelligently selects the most appropriate model based on your query type. “Best” mode is available without quota limits.

[frigid marsh]

that's not the case

[tranquil lodge]

Funny how when I ask Perplexity it says Best Search is unlimited

[dry kestrel]

ehh Gemini users get 100 prompts a day with 3.1 and ChatGPT is 80/3 hours for GPT-5.2

[short narwhal]

Should bring back the pro search toggle :p

[dry kestrel]

Bug

[tranquil lodge]

still better compared to 200 per week

[dry kestrel]

It's much higher than 200 a week given the rolling window

and these companies are able to eat the cost because it's their own models

[tranquil lodge]

also you get access to their api seems like they also remove free 5$ api credits per month

[dry kestrel]

ChatGPT and Gemini does not give you access to free api credits

[naive sage]

(Well, Gemini does give AI Credits which can be used for API)

(But yeah, ChatGPT doesn't)

[dry kestrel]

Where do you see this?

[tranquil lodge]

I'm sorry if I look like I am complaining but all I am trying to convey is atleast spare the pro searches off of any limits

[dry kestrel]

Ai credits are for their whisk and google flow product, not the API

[naive sage]

I thought they could be used for the API too

[dry kestrel]

They cannot.

https://support.google.com/googleone/answer/16287445?hl=en

[tranquil lodge]

They do, Gemini even allows free tier users to use their api

[dry kestrel]

Buying a Gemini subscription does not give you API credits. I don't know where you got this information from.

[tranquil lodge]

https://aistudio.google.com

It increases your api limits if you have pro subscription

[naive sage]

Interesting ... so the credits you get for the Google Developer program and Vertex can be used for it, but the Gemini Subscription credits don't... that's weird. They're all called AI Credits, hence the confusion. It's like they have different AI Credit SKUs

[dry kestrel]

Ai studio is very different to the Gemini app - 2 different products

[dry kestrel]

Where do you see this?

[tranquil lodge]

https://ai.google.dev/gemini-api/docs/

I suggest you give it a good read bro

[dry kestrel]

Where does it say you get increased API limits when on an AI pro subscription?

[dry kestrel]

Gemini model ≠ Gemini APP

[tranquil lodge]

https://support.google.com/gemini/answer/16275805?hl=en

[short narwhal]

That's for the app

[naive sage]

That doesn't mention the API - it doesn't even mention the AI Credits

[hasty pebble]

let's all agree perplexity is shit lately

[tranquil lodge]

My bad wrong doc I thought he was arguing about no API

[naive sage]

You left the important bit out... what violated content policies?

[tranquil lodge]

Bro casually wasting 4 searches

[dry kestrel]

I think I made my initial statement very clear. #💬│general message

[hasty pebble]

100 every day?
that's still 500 more than what perplexity is giving you ,right?

[dry kestrel]

Google, a multi trillion dollar company, can eat the cost on their own models. Perplexity pays API prices

It's like comparing Walmart to your local farmers shop

[hasty pebble]

sure

[tranquil lodge]

I might have misintrepreted it #💬│general message

I thought you meant credits as in usage credits

and not money credits

[dry kestrel]

Ok

[hasty pebble]

we can all agree at this moment it's very obvious where someone can spend 20$ and get the most out of it

[dry kestrel]

Perplexity Search > Gemini

also people working at Google look at your conversations on Gemini

[hasty pebble]

that's ok

[tranquil lodge]

Hopefully in the future (I pray this week) Perplexity might change and fix their pro usage limits

[dry kestrel]

Perhaps for you that's fine, for a lot of users that's a huge red flag

[tranquil lodge]

That's every AI company

[dry kestrel]

This is just not true.

[tranquil lodge]

They are required to especially if they own the model you are using

[dry kestrel]

Okay, show me the ToS for "every Ai company" other than Google that says that humans look at your conversations

[tranquil lodge]

ChatGPT, Meta, all closed source model companies do

[hasty pebble]

more users will gladly accept that if they get 100 queries each day with flagship model with good context window

[dry kestrel]

This is false.

[dry kestrel]

also you mention "Meta" while their last model had faked benchmark results

[hasty pebble]

@dry kestrel if I select best model ,will it count as pro search ?

[rich rover]

Guys, I used to update a Google sheets with perplexity and comet creating URLs for Google Analytics, but Last week Comet doesnt read Google sheets and doesnt create my urls anymore. Something changed?

[dry kestrel]

Looks to be working fine for me

[hasty pebble]

what about when I'm out of pro searches,will it still count as pro search if best is used?

[tranquil lodge]

This was literally this week

[dry kestrel]

it was because the chat was FLAGGED - very different to Google's privacy policy of "A subset of chats are reviewed by human reviewers "

[hasty pebble]

you didn't understand what I asked

when im out of "advanced models" and only option is best.
will the thread be in pro mode,meaning 10x more sources etc

[dry kestrel]

Google has the right to view whatever conversation they feel like doing

according to their privacy policy

This is NOT the same as OpenAI, Anthropic, Perplexity, etc

[tranquil lodge]

Yea I mean they even tell this straighaway to you in their chat

[hasty pebble]

what credit?

you're confusing me

[dry kestrel]

I do not want Google employees to be looking at my chat history without a valid reason to do so.

[tranquil lodge]

I mean they can if they want to even if they have no valid reason to

[hasty pebble]

im pretty sure you're not that important but I get what you're saying 😂

[wind urchin]

Hello

[dry kestrel]

Huh?

[hasty pebble]

hahaha

I tried to read it with understanding but I failed miserably

[short narwhal]

The point is that it invades your privacy

[tranquil lodge]

your chat data is already being used in model training

[hasty pebble]

I know

[short narwhal]

You can disable that though

[dry kestrel]

You can disable it on ChatGPT and Perplexity

[hasty pebble]

so if I disable it on chatgpt ,they can never see any of my threads and what I'm doing there?

[dry kestrel]

They do not train on your data if you disable it.

You can opt out of training through our privacy portal by clicking on “do not train on my content.” To turn off training for your ChatGPT conversations and Codex tasks, follow the instructions in our Data Controls FAQ. Once you opt out, new conversations will not be used to train our models.

[hasty pebble]

that's very different from what I asked xd

[tranquil lodge]

Ngl this is a very grey area of dicussion and also very pointless to argue

[twilit crown]

Perplexity pro isn't very good now

[dry kestrel]

My point was on model training

[hasty pebble]

nah bro,you're tripping.
it's peaking 😂

[short narwhal]

As long as you don't break the ToS or the law, no, they won't look at them

[tranquil lodge]

Regardless at the end we all care about is Perplexity raising the limits for Pro Searches that's all I care about atm

[dry kestrel]

I don't know what youa re on

It's not hard reading their privacy policy or just doing a quick search

[hasty pebble]

so why would someone who is not breaking TOS or the law be worried about using Gemini ?
😀

[short narwhal]

Gemini ≠ ChatGPT

[dry kestrel]

Did you read my message? I was talking about disabling model training on ChatGPT and Perplexity

[tranquil lodge]

That's like the normie argument to the whole privacy thing, "If I have nothing to hide why should I worry"

[hasty pebble]

yes

[dry kestrel]

Well given your response, it's clear you did not read it properly.

[hasty pebble]

my point is ,I know a bunch of people who were banned on chatgpt because they were trying to test the limits of jailbreak .
I'm sure they all unchecked train model on my data .

[tranquil lodge]

I don't think ChatGPT bans a user on jailbreak unless you do some devious stuff

[hasty pebble]

how do you know ?
are you already jailbreaking and you're still not banned?

😀

[tranquil lodge]

I have jailbreaked it countless times even with pro subscription never had an issue or anything best is they patch the jailbreak prompt you used within a week or so

I don't do it for wrong reasons though so maybe that's why

most of my prompts are just to get the bot to be raw and unfiltered instead of some corporate chatbot

[hasty pebble]

you can do that with simple instructions.
to be fair,simple instructions are also a jb in a way then

[tranquil lodge]

haha try it

[hasty pebble]

what exactly

[tranquil lodge]

I am unsure if chatting about this is fair here lol I wouldn't entertain this convo

some people see jailbreak as something morally wrong

[hasty pebble]

I don't mind their position on jailbreaking .
it's ok 😂

[short narwhal]

ToS still applies, and they more likely are just flagged automatically

[tranquil lodge]

you can try MongoTom that's what I mostly do its harmless so I suppose won't be a huge deal but probably hardest to get working especially with Claude

[short smelt]

Hey guys , the specific space dedicated to creating quizzes , what is the limit for that please specify ?

[hasty pebble]

I dmd u

[tranquil lodge]

I think Quiz is also counted in pro Search though I could be wrong

[short smelt]

Okay

[short narwhal]

#💬│general message Same limits apply to queries made in spaces

[short smelt]

Okay thanks.

[woeful coyote]

My pro was ended

[short smelt]

Same 🙁

I had to add my credit card details

[woeful coyote]

I am sad

[woeful coyote]

Ya but my pro was ended before 28.1.2026

But my friend bought it this year and he nad to add credit card details like you

[woeful coyote]

when did you buy it?

And where?

[short smelt]

I did not buy it

I got it via my college(education pro)

[magic compass]

Does anyone know what day of the week do pro searches reset?

[short smelt]

I got it in India

[tranquil lodge]

They reset 1 query every hour at a time

[short smelt]

That is very unfair to be honest

[magic compass]

Wdym? They add back 24 queries per day?

[tranquil lodge]

yes upto 200 max

[frigid perch]

is there a way i can check browser automation limits?

[tranquil lodge]

that's what I learned today xd

[magic compass]

Ok, thank you

[tranquil lodge]

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

[frigid perch]

do you know what name that would be under?

agentic research?

[tranquil lodge]

probably better to ask a mod or vet lel but this is the only way you can see all the applied rate limits

maybe try a browser automation and see which value changes

[hasty pebble]

there's one more link

[frigid perch]

hmm okok

[hasty pebble]

https://www.perplexity.ai/rest/user/settings

[frigid perch]

cause this is all i really see as a pro user

[frigid perch]

will check

[short narwhal]

Not visible afaik

[frigid perch]

aw man

cause i hit “monthly limit” then couple hours after got access to like 5 more before hitting “monthly” limit again

[short narwhal]

It's a rolling window, so you end up regenerating usage after a bit. But not sure what the rate for that is on Comet's agent :( Could be hours, could be a day or more

[ionic stone]

Wen gemini 3.1 flash

[tranquil lodge]

model is not released by Google itself

[flat hare]

perplexity is getting greedy

usual strategy

[ionic stone]

lol yeah I know

[lyric pelican]

It's so bad ngl

[idle steppe]

Wish I could play this on ps….

[frigid perch]

dang rip hate that is says hit my monthly limit even though it’s rolling :P

[still meadow]

Has anyone tried making a daytrading bot with Perplexity (or something similar)?

[vast wagon]

Anyone use claude on perplexity, is it different from using Claude directly?

[twilit crown]

What IDE has the best autocomplete?

[frigid perch]

been using warp tbh not too sure on others

[short narwhal]

Cursor

[forest bear]

its cursor

[distant aurora]

@green crystal what are the odds

[paper bay]

Hey guys can anyone lend me grammarly pro for 5 days, it helps me for my project. Idk whether i can ask here or not, if i shouldnt ask kindly let me know, i will delete this message

[distant aurora]

Use free version of chatgpt or deepseek.

You shouldn't have to pay for something as simple as grammar checker.

[paper bay]

Its not for grammar bro, its for humanising and plagiarism checking

🥺

[distant aurora]

Duplichecker has free plagiarism checking.

[paper bay]

But it has word limit

[coral sluice]

If I cancel my subscription, could I receive a partial refund, please?
I really need it.

[distant aurora]

And there's no guarantee of any humaniser actually working imo.

[paper bay]

🥺😭

[distant aurora]

O3 was the closest thing to passing ai detectors afair.

[green crystal]

impossible lol

[distant aurora]

I didn't bother to test anything further. Kimi might work too.

[paper bay]

O3 model?

[distant aurora]

Yes

[paper bay]

Oh

[paper bay]

It looks like paid?

[distant aurora]

Yes, it's paid.

You can access some ai models for free on arena ai though.

Or try kimi.

Oh. Also.

There's this trick.

Dump a book into your ai.

Ask it to replicate that style.

Google added this weird humaniser thing in it's docs or something. You can look it up.

[paper bay]

oh really? can you guide me?

[distant aurora]

Nah

[dry kestrel]

Don't commit academic dishonesty, not worth it. Plus current AI humanizers don't work as well as you'd think.

[paper bay]

i have less time bro, so i have no choice 🙁

[twilit crown]

My code documentation is amazing

[dry kestrel]

The consequences are quite severe depending on your level of education.

[paper bay]

dw bro, i am not submitting it in my college, i am gonna submit on online platform

[modest latch]

Guys do any of you have servers that are like 100 to 200 people and have like no rules whatsoever? Id love to join such servers

[junior ice]

Hi all

[naive sage]

I understand the reasoning behind thinking no rules is appealing... but there are times when rules are important. It really depends on the focus on the server, how well the community interacts, and where those rules are set. I don't think no rules at all is compelling, but rather reasonable limits on where the lines are. You don't want illegal content, for example... that can't be tolerated or else the entire server community is put at risk (not to mention your own livelihood).

[modest latch]

By no rules I mean like you can use cusses

[naive sage]

But I agree with the sentiment, if what you are saying is that a lot of corporate servers are too strict... the Gemini server, the OpenAI server, etc... all limit things a bit too much in my opinion.

Web links to news articles often aren't even allowed.

This server is actually more lax than most of them.

[modest latch]

Nah nah, what I want is a server with not too many people so id ve able to regularly talk to the same people over and over

And also rules so less that I can talk freely

Like not be chained down by swear restrictions or anything

[dry kestrel]

There's going to be rules no matter what - or else the server get's taken down due to violating ToS

[naive sage]

Best bet is trying to scout a community on Disboard or something, unless someone here does have a recommendation of AI-enthusiast servers.

[modest latch]

Ofcourse, i understand rules I just wanna be able to say what I want 🤞

[modest latch]

Good idea but small servers are kinda hard to find there no?

[kind saddle]

how change from perplexity to google when Im typing something in browser? I cant chang eit!

[naive sage]

Not really, it used to post the server populations, and most of them were smaller than a couple hundred members... but I don't see that population listed anymore.

[modest latch]

There is a search tab and ai tab right

[naive sage]

Ah, there it is - just sort by member count.

[modest latch]

Oh cool

Kk

Thanks dude

[kind saddle]

Can You explain/show? I see only perplexity icon

[naive sage]

No problem... the issue I have with smaller servers is a lot of them aren't active.... most members are usually in some other larger server spending most of their time. Almost all of the random servers I've tried end up being almost dead. So, I guess if you find one - let us know!

[modest latch]

You are talking about comet right?

Whenever you make a search, up there right above the perplexity name tag it shows different columns

[modest latch]

Sure

[kind saddle]

Yes, on deskop. I ant to type somehing and use it in google

[modest latch]

Ofcourse you can always just search for Google and you'll get a link, you can ask perplexity to give you a link, or the search tab

I'm on my mobile right now

So I can't really help

I'm sorry :<

[kind saddle]

I have changed it one week a go, on my previous windows install. I was using google instead of perplexity. Now I cant'!

[modest latch]

OHHH

Change your default search app

[kind saddle]

yes! I can't

[modest latch]

Windows right

[kind saddle]

yes

[modest latch]

Okay

Open settings

Go to apps

And then default apps

Select chrome in search engine

Or instead open chrome

It will give you a notification that

"chrome isn't your default browser"

Then set it as default web browser

[kind saddle]

I dont have 'apps'

[hasty pebble]

problems immediately

[modest latch]

Just search default by pressing windows key

[kind saddle]

but I want to use comet as a default, but inside comet I want to use google not perplexity in 'command prompt' area

[dry kestrel]

comet://settings/search

[kind saddle]

I have there only Perplexity to choose, no option to add something

[dry kestrel]

Are you on the latest update?

[kind saddle]

Yes

[dry kestrel]

I'd try reinstalling

[modest latch]

Oh

[modest latch]

.

I wonder how that nametag is golden

Like how the hell does it have a gradient

Never seen that before

[kind saddle]

my god.. i have new pc and i'm repairing everything and tunning for 2 weeks, all the time it's something m god

[dry kestrel]

discord released it a couple months ago

[modest latch]

Sucks fr

[modest latch]

Oh

Makes sense

[hasty pebble]

how do I see why was I muted

[naive sage]

When it's happened to me it's been right after I posted something - usually a long response or a link - the mods usually review mutes pretty quick and restore someone if it was a bot being overzealous. It's only happened to me a couple of times, and both times the mods have unmuted me within a few minutes.

[hasty pebble]

I was muted entire 1h
mods probably took a nap or something

[dry kestrel]

@short narwhal https://x.com/testingcatalog/status/2026031251115671701

TestingCatalog News 🗞 (@testingcatalog)

Perplexity is working on Messages connector for Comet browser and a new "Usage and credits" settings tab.

X · Feb 23, 2026, 8:27 PM

[winged oak]

Sup guys, I’m just getting into the AI bubble. I’m broke, a father and want to earn extra. Any tips/guidance would be appreciated. I’ve been looking into either website creation or UGC tiktok videos. (Just the ‘business models’ I’ve seen and researched)

[cobalt mountain]

@viral violet

You a bot?

usually server tags for new joins in the QRL server. Usually, those with tags to servers and no chat history usually indicate a scam/spam/bot account.

If you don't respond to me here, or there, I will ban you there

[viral violet]

What happened bro??

[cobalt mountain]

Ah (:

[viral violet]

Why so aggressive? ?

[cobalt mountain]

nvm

[cobalt mountain]

Not aggressive, just thorough

[viral violet]

Its 2:30 am here in my country

[cobalt mountain]

I usually give 24h

[viral violet]

Lucky that I saw ur msg

😆 😆

[cobalt mountain]

Thanks for following up (:

I'll be leaving now

[viral violet]

Gd night

[short narwhal]

oo that's interestingg. And a proper view to see usage!!?

[modest latch]

The mods are so good fr

[dry kestrel]

Yesyeysye

[modest latch]

Do try doing Canva commissions! I'll get back to you with more.. useful ideass

[hasty pebble]

nah

[modest latch]

Yah

[tulip robin]

Did they decrease the limit again from 20 deep researched daily to 20 monthly

Was 600, then 250, then 20 and then 20 monthly. what the hell is this...

[hasty pebble]

it's 24 pro searches daily

[hasty pebble]

this is perplexity discord server

[short narwhal]

New connectors!!

[solar venture]

Did perplexity limit the pro?

[hasty pebble]

yes

[solar venture]

I thought so because it's only using sonar now.

[solar venture]

Do you know if the limits reset daily still?

[leaden halo]

My former roommate who works at Perplexity owes me $720 CAD

[naive sage]

It is best to expect nothing in return when you lend money.

[leaden halo]

We've discussed that he'd pay me back

[rugged topaz]

Hi everyone! Greeting fron Spain

[tulip robin]

20 per month for me

I was talking about the deep researches

[dry kestrel]

holy

[naive sage]

I don't see the holy connector, unfortunately...

[dry kestrel]

perhaps the connecters from the 2nd screenshot are region specific?

[short narwhal]

Showed up still while I was on a US vpn 🤷 Also they show up only in this menu here since those ones are not something you have to auth yourself

[dry kestrel]

Hmm interesting it shows that lock thing for model council

let me clear cache

[short narwhal]

yeah I got that literally just a moment ago I think

[dry kestrel]

yep showing now after clearing cache

the response thing looks new too

[short narwhal]

3 weeks old

[dry kestrel]

oops, it was on the recent change log https://www.perplexity.ai/changelog/what-we-shipped---february-20th-2026

[short narwhal]

They're a bit late with that but sure #💬│general message

It being on mobile is new tho, I'm not even seeing it there yet rn

[dry kestrel]

oops

I remember seeing it briefly, then disappearing

[dry kestrel]

same

[short narwhal]

and another new one

[rapid crest]

#💬│general message

@vapid nimbus Update on the Gemini Situation: I'm very happy that Gemini pro 3.1 is now about as fast as the old Gemini 3.0 pro.

This is amazing (because Gemini pro with reasoning is pretty good if you need difficult stuff)

[junior night]

is sonnet going to be fixed? I'm stuck right now

[thin sable]

Perplexity is starting to become actually unusable for me

Dunno if it's just me

Most of the time it doesn't understand what I ask it, it fetches the wrong sources and doesn't do what I ask it to do

Idk what changed in the past few months but eh whatever

[dusk elk]

Same, I pretty much stopped using it recently, it's annoying, considering I used to use it all the time before they changed the number of deep research chats per month.

[thin sable]

I switched to chatgpt plus and it's actually so much better than perplexity in every way imaginable

[dusk elk]

Also, does anybody use Gemini nano banana pro and if so, does anyone know how to get it to produce images at a higher resolution than 1520x704p?

[thin sable]

I think you can only access that through ai studio, not through the gemini app

Not certain though

[dusk elk]

Use Gemini or Claude instead, they're both better and chatgpt probably won't be around much longer.

[thin sable]

I already have Gemini Pro lol

I got too many AI subscriptions

[naive sage]

Lots of people are saying that about ChatGPT, but they'll be around. They're not going anywhere for the next several years, at least.

[dusk elk]

To clarify, open AI are running out of money very quickly, even to the point where Nvidia are losing hope in them, so unless they change something or able to find people to invest around 200 billion dollars, they're pretty much screwed.

[dry kestrel]

use it via the API

[dusk elk]

Plus, most other AI's are better now anyway, so people aren't using chatgpt as much anymore.

[dusk elk]

I'm guessing I'd have to pay extra?

[dry kestrel]

You pay for what you use

[dusk elk]

I do have openrouter, can I use it in that?

[dry kestrel]

Yes

its like 12 cents per 4k iamge

i think

[dusk elk]

Awesome, good to know, thankyou.

[dry kestrel]

that excludes your input token costs btw

[dusk elk]

No worries, I'm not creating anything crazy, so my prompts shouldn't be too long. Plus I have Gemini pro already to test my prompts in before I use it in OpenRouter.

[dry kestrel]

Perhaps try out Google Flow

since the images on gemini website has visible watermarks

[dusk elk]

Is it part of Gemini pro?

[dry kestrel]

yes

[dusk elk]

Awesome, will give it a look.

[jagged leaf]

wtf is perpelexity, i have pro i everytime i select calude 4.6 thinking on/off

it just changes it to best model and uses Grok...

wtf man

[junior night]

there are new limits

[jagged leaf]

its only tuesday and it should tell me if i have exceeded

cant contact support stupid ai bot

cant open new chat

wtf is this

if i click contact support it does nothing

when do they reset?

anyone have the contact email? a real person please

or an actual place to get help cos this isnt adding up unless the new cap is like 30 a day...

and if thats the case i am cancelling my subscription

isnt even in the shitty https://discord.com/channels/1047197230748151888/1101287226714898583

https://tenor.com/view/dumb-stupid-smooth-brain-idiot-gif-20901280

[frail monolith]

how did you get that pro tag 😄

[jagged leaf]

alright anthony

what are you trying to say?

[dry kestrel]