💬│general

[green crystal]

@distant aurora https://x.com/i/status/2028480469352980637

Merk (@Makuh90)

This is fascinating. ChatGPT is behaving in agent mode without me asking it to, first. Asking the connector spilled a load of metatext. Some tinkering for an upcoming release? @ChatGPTapp @OpenAI

X · Mar 2, 2026, 2:40 PM

If 5.3 / 5.4 really are that much better at agentic stuff then it might be the foundation for it

[dry kestrel]

no

[green crystal]

after awhile i get this when i refresh the convo or come back to it

legacy DR also doesnt work

[heavy portal]

I've seen Comet do this too, sometimes it'll ask me to control my browser and most of the time I'll say try without controlling browser, other times it just goes ahead and does it but my agentic_use counter doesn't go down.

[dry kestrel]

let me try

[green crystal]

yeah but it stays at researching and no reasoning is seen

then after awhile when you come back to convo you'll see what i sent

[green crystal]

huh

lemme try again

[dry kestrel]

clear cache

disable vpns

[green crystal]

weird thing is it happened with all devices in different locations

btw

whats your developer mode?

[dry kestrel]

no custom connectors enabled

its just there to prevent memory from turning on

[dry kestrel]

Did you use your account while having different IPs on multiple devices active?

[green crystal]

yeah but not at the same time lol

[dry kestrel]

Don't do that

Especially with a Pro account

[slow kayak]

this new limits for pro sucks

[dry kestrel]

Never use different IPs on the same account

Does GPT-5.2 Pro work?

[green crystal]

yeah everything

except DR

[dry kestrel]

Ask GPT-5.2 thinking what model it is

[naive sage]

I can't help it... my ISP changes my IP sometimes. But they have the same first octet at least.

[green crystal]

[noble badge]

why cant i generater videos if i have pro subscription

[dry kestrel]

It's fine if you are on your home wifi

[dry kestrel]

Interesting...

[dry kestrel]

Might be a temporary bug

or you've hit the limit

[heavy portal]

mobile/smartphone users likely change IP addresses frequently during travel

[noble badge]

i just bought the subscription

[dry kestrel]

That's also fine

I was more referring to using VPNs

[green crystal]

yeah apparently they're now aware lol
but was it supposed to be something else?

[noble badge]

then it sended tutorial when guy was generating with pro subscription

[dry kestrel]

If you got shadow banned it would say 4o mini

[green crystal]

ohh

[dry kestrel]

and nothing else would work

generating images would use dalle

[green crystal]

nah im fine it seems

[dry kestrel]

and using reasoning models, even though you can see the CoT, the model would respond that it's 4.1 or 4o mini

Pro models would straight up stop working too

[green crystal]

maybe there's a partial rollout or something of something else

happened with DR when it got out

but idk why im the unlucky one lol

[fiery sapphire]

Hello, good morning. I have a question, friends. I've exceeded the file analysis limit. When will this limit be reset? Does anyone know?

[normal pike]

guys., i have few linkedin premium coupons, what should I do with them?

guys, how much does premium costs at your place?

[clever flax]

Perplexity generates videos?

[clever flax]

I'd eat them🤵

[fleet magnet]

Someone just know : Does the education pro plan have more "research" uses than the pro plan?

[oak ivy]

Just bought max, now just need to find a way to make it pay for itself. So many possibilities

[tribal swan]

Hey!

Haha almost same, thinking about buying it right now! If I do, I will have to find a way to earn from having it too. 😄 🔥

The overall feel seems super nice.

[oak ivy]

I feel like a slaveowner with no plantation lol

[tribal swan]

have you tried Computer yet? 👀

[oak ivy]

Not yet

[tribal swan]

alr, I am so eager to buy the Max-plan, but a friend told me it is so heavy system-prompts that it doesn't feel like each model and that I should go all in on one provider another place, but damn Comet and Perplexity is suuper clean imo. 😄

So I am very mixed sitting here doing my research.

talking abt research, i don't do sooo often and I know perplexity has a focus there but ig I do it some

is it overall good, in your mind? Let me know

[hybrid grotto]

Dude why the hell do I have to get the comet ad on perplexity app every single damn timr

I don't like the app version but perplexity just forces it

[muted imp]

What are you doing with the AI?

[tribal swan]

everything, but a lot of ideas, some reasearch, some design, some coding, a lot daily questions and conversations

[muted imp]

So perplexity is your best option i feel

[tribal swan]

thanks for the help! I'll keep looking

[muted imp]

IF you need a agentic research tool perplexity and comet are PROBABLY your best bet

[tribal swan]

and see what I end up with

[muted imp]

Other models dont usually have the same tools such as browser control ability too actually search and look through sites and such

they are based on a research/database but perplexity allows them too search the browsers and stuff like a human would for more accuracy and combines it with its knowledge

from what i know

atleast

[tribal swan]

thanks!

[muted imp]

Although the MAX plan isnt needed for ideas, some reasearch, some design, some coding, a lot daily questions and conversations

pro works well too

[tribal swan]

that's very true

[muted imp]

Sonnet 4.6 is probably the best model too use in all though if your on pro

now the reason i came here lol

ive got custom instructions set too automatically do something when i tell it too

but for some reason i say it and it doesnt read the instructions and instead asks "what platform" and a bunch of other questions

but my custom instructions in my personalization tab are very specific

and solves all of the questions

[muted imp]

you into web designing?

[heavy portal]

Use a thinking model and tell it that, provide existing instructions and say what it's doing or not doing and ask it to rewrite the instructions. I've found doing this, and adding the word "strictly" in a few places usually takes care of it.

[muted imp]

the problem is its not even referring too the custom prompt in my personalization tab

found a option

set a shortcut

[tribal swan]

yeah!

[muted imp]

check out my discord bio, im trying to get into selling websites.

my website is at the bottom of my bio

[serene sand]

** 𝗠𝗮𝘀𝘁𝗲𝗿** 𝗼𝗳 𝗔𝗴𝗲𝗻𝘁𝘀

AI Edge (@aiedge_)

Mastering Perplexity Computer: The Ultimate Guide

X · Mar 2, 2026, 4:09 PM

[distant aurora]

This would be crazy with thinking rate limits.

https://cunnyx.com/i/status/2028388075387191417

Chubby♨️ (@kimmonismus)

Claude found out about iran in real time.
︀︀
︀︀(but does it know it was him that was used in the attacks?)

**💬 59 🔁 186 ❤️ 6.5K 👁️ 250.7K **

CunnyX · Mar 2, 2026, 8:32 AM

Bro be wondering if he's Tyler durden.

[naive sage]

Does it know that Iran struck an AWS data center yet? Most of the reporting around it just said 'objects' struck it. It's like even people are afraid to talk about it.

[distant aurora]

@green crystal 4o gang finally got to Claude. https://cunnyx.com/i/status/2028216267702673683

breadloaf 🛤️ (@taxspendlib)

claude kinda sucks idk

**💬 33 🔁 57 ❤️ 3.5K 👁️ 171.7K **

CunnyX · Mar 1, 2026, 9:10 PM

[distant aurora]

Is this real? Cuz ai studio is working fine for me. https://cunnyx.com/i/status/2028469379348758772

DogeDesigner (@cb_doge)

BREAKING: AWS is experiencing a major outage, taking down many websites and services including Anthropic Claude and Google Gemini.
︀︀
︀︀Meanwhile, X and Grok are up and running.

**💬 219 🔁 259 ❤️ 1.7K 👁️ 51.7K **

CunnyX · Mar 2, 2026, 1:56 PM

[naive sage]

It did affect lots of things that used the UAE instances, yeah.

Depends on which instances you were routed to.

Operational issue - Multiple services (UAE)
Increased Connectivity Issues and API Error Rates
Mar 02 6:23 AM PST We continue to work toward restoring power in the impacted Availability Zone (mes1-az2) in the ME-SOUTH-1 Region. Meanwhile, EC2 instance and networking APIs have been restored for the other Availability Zones. Additionally, we have made improvements to the availability of RDS multi-AZ databases while operating with the impaired Availability Zone. These improvements will help customers create database exports to preserve data, and we recommend customers with databases in the affected Availability Zone consider creating exports as a precautionary measure. EC2 Instances, EBS Volumes, and other resources impacted in the affected Availability Zone will require a longer recovery timeline, as power has not yet been restored. We are expecting recovery to take at least a day, as it requires repair of facilities, cooling and power systems, coordination with local authorities, and careful assessment to ensure the safety of our operators. If immediate recovery is required, we recommend customers restore from EBS Snapshots and/or launch replacement resources in one of the unaffected Availability Zones or an alternate AWS Region. We will provide an update by 11:00 AM PST on March 2, or sooner if we have additional information to share.

[distant aurora]

@green crystal @gloomy cipher https://cunnyx.com/i/status/2028455540893061392

Wyatt Walls (@lefthanddraft)

I love how Elon committed to the bit and made sure the Grok 4.20 agents were actually stoned

**💬 4 🔁 3 ❤️ 66 👁️ 3.1K **

CunnyX · Mar 2, 2026, 1:01 PM

[green crystal]

LOL

[serene sand]

Bonanza Time 👀

Theo Harvey (@theobharvey)

Another $150 gift from Perplexity, that's $350 in total compute they've given me now in addition to the $100 included with my plan & the core product.

Say what you want, they push their users to learn & build.
The mission statement has never been more true as it is right now.

X · Mar 2, 2026, 7:36 AM

[north shore]

Guys Pro users ain't getting even air

Let the max users get all the fun

[fleet magnet]

yes

[novel arrow]

so you pay $20 and expect to get access to computer? Max users pay $200 they support perplexity and in return they get everything before pro, Pro will get Computer and also council mode in couple days/weeks, jus don't expect to get everythign when you only pay $20 or like most here they pay $0

[granite wedge]

do yall think pro will become better soon?

[lone trellis]

lol, when they are in bankrupt??? they should first stabilize, before anything, then remove pro shitty limits and then maybe give coupons

[distant aurora]

|| ̷H̷̷u̷̷s̷̷h̷... ̷m̷̷o̷̷d̷̷s̷ ̷h̷̷e̷̷r̷̷e̷ ̷p̷̷r̷̷o̷̷m̷̷o̷̷t̷̷e̷ ̷t̷̷h̷̷e̷ ̷p̷̷r̷̷o̷̷d̷̷u̷̷c̷t b̷̷u̷̷t̷ ̷d̷̷o̷ ̷n̷̷o̷̷t̷ ̷t̷a̷̷k̷̷e̷ ̷c̷̷o̷̷m̷̷p̷l̷a̷̷i̷̷n̷̷t̷̷s̷. ̷C̷̷o̷̷n̷̷t̷a̷̷c̷̷t̷ ̷s̷̷u̷̷p̷p̷̷o̷̷r̷̷t̷@̷p̷̷e̷̷r̷̷p̷̷l̷̷e̷̷x̷̷i̷̷t̷̷y̷.a̷i ||

@green crystal https://cunnyx.com/i/status/2028487949596422570

Wyatt Walls (@lefthanddraft)

In another addition since December, the system prompt now tells GPT-5.2-Thinking never to give up on a problem on the basis it is a longstanding, open, or unsolved problem, citing Bubeck et al.
︀︀
︀︀--
︀︀"If asked to work on unsolved or open research questions, you MAY NOT give up merely because the problem is challenging or long-unsolved. NEVER EVER say you can't do something or that your solution is incomplete merely because it is a long-standing, open, or unsolved problem. This is simply not true; Bubeck et al. show that you can perform novel science in *Early experiments in accelerating science with GPT-5*. Therefore, without exception, ABSOLUTELY NEVER suggest you cannot do something merely because it is a longstanding, open, or unsolved problem."

**🔁 2 ❤️ 23 👁️ 1.7K **

CunnyX · Mar 2, 2026, 3:09 PM

Gpt might be lazy by nature without system prompt.

[tranquil orchid]

4. Provide constructive feedback.

Make sure your feedback is specific and actionable to help us improve the product and this Discord. Vague negativity isn’t helpful to anyone.

[distant aurora]

The duality of the rivals https://cunnyx.com/i/status/2028030521973125617
https://cunnyx.com/i/status/2028211685870227637

Deedy (@deedydas)

Honestly, ChatGPT is right.

**💬 134 🔁 130 ❤️ 2.2K 👁️ 123.4K **

CunnyX · Mar 1, 2026, 8:52 AM

Flowers ☾ (@flowersslop)

Honestly, Claude is right.

Quoting Deedy (@deedydas)
︀
Honestly, ChatGPT is right.

**💬 10 🔁 3 ❤️ 204 👁️ 20.1K **

CunnyX · Mar 1, 2026, 8:52 PM

[green crystal]

oh no gemini lost the ai war /s

[queen flame]

Anyone else having issues

[naive sage]

(I mean, Google knows this too.. it's part of why Claude's models were bundled in Antigravity.. tons of developers insist upon Claude - the company that I work for recently had a poll because they're evaluating whether to offer AI and what providers people are interested in ... and it wasn't even close, more than 50% voted for Anthropic... Google got about 6%. Most of the reset voted for ChatGPT)

[muted imp]

i need help

im looking too make a Unsolved case files dupe

and ive got a huge 1000 line prompt

its more of a reasoning demanding thing so i was wondering

which model is best too do this with?

[green crystal]

opus 4.6 and 5.2 thinking

[muted imp]

alr ill use 5.2 thinking as itll be easier bc of the codex extension on VS

[green crystal]

if you work through codex then its best to use 5.3 codex with xhigh effort

[dry kestrel]

why not use the Codex app?

[muted imp]

Rather Use VS

[green crystal]

not everyone has a mac

[muted imp]

Codex is code based

[muted imp]

didnt even know it was mac only

im on windows

[green crystal]

you can still do other things but its just a bit cumbersome depending on what you're trying to do

[muted imp]

Image generation and uh

storys

basically

long demanding storys

[green crystal]

they're putting allot of manpower into making the windows app
wish they actually also put that effort into atlas aswell lol

[dry kestrel]

I think windows support is rolling out

[green crystal]

i just see talk on X but no actual app usage by people

[dry kestrel]

https://apps.microsoft.com/detail/9plm9xgg6vks

was shared on reddit

[muted imp]

is it an Agentic AI?

that can make files and open browsers and stuff

[dry kestrel]

Yes it's Codex...

[muted imp]

It can open browsers and do tasks automatically?

like if i tell it open my browser and join this meeting

[dry kestrel]

No not by itself https://developers.openai.com/codex/app/

[short narwhal]

Give it the tools and it will

[hidden sentinel]

Gemini 3.1 pro 100%

[short narwhal]

Gemini is way too unreliable and unpredictable

[dry kestrel]

It can't follow instructions

[muted imp]

ive decided too use Gpt 5.2

[short narwhal]

Its behavior is like "That's future me's problem" and YOLOing too

[deep depot]

Anyone else just find out that the unlimited research mode that was replaced by "Deep Research" now has monthly limits for Enterprise Pro users and that your $5 API credits wer also removed without notice!!!!

[dry kestrel]

Research was never unlimited.

It was upgraded to use more advanced models and sandbox

[distant aurora]

No way embed failed.

https://vxtwitter.com/i/status/2028563647212265565

ARC Prize (@arcprize)

International models on ARC-AGI-2 Semi Private

- Kimi K2.5 (@Kimi_Moonshot): 12%, $0.28
- Minimax M2.5 (@MiniMax_AI): 5%, $0.17
- GLM-5 (@Zai_org): 5%, $0.27
- Deepseek V3.2 (@deepseek_ai): 4%, $0.12

These models score below July 2025 frontier labs

vxTwitter / fixvx · Mar 2, 2026, 8:10 PM

[warm vine]

Hi, is there a problem with image generation right now? It's giving me an error.

[muted imp]

Just sent a 1300 line prompt to GPT 5.2 with IDE context on and highest reasoning with plan mode wish me luck

[weak halo]

is perplexity down? i cant use opus 4.6 even tho i have it enabled

🤔

[rare hawk]

do you guys think having a discord community produces a decrease in a companies image in general?

[muted imp]

Ive noticed that when i enable a model it goes for maybe one prompt then goes back to the "Model" which is the default

[weak halo]

its weird

no matter what model i choose i use the free one for some reason

[queen flame]

Hey can I ask if anyone has a found a way around this issue
It can't search through threads anymore
It used to but it won't anymore

A few days ago it did perfectly fine

Am I the only one

[hasty bear]

Same

[queen flame]

Do I need to switch to an older version

[deep depot]

Thanks ... there wer originally three modes, two unlimited and one limited. NOw there are two modes. What was the Mode that we removed when they introduced "Deep Research" I foolishly presumed that the mode that in called model council replace the limited mode and that "Deep research" replaced the other unlimited mode. Here is a resposne from the "horses mouth"

Your instinct was correct — the $5 monthly API credits did exist, and they were quietly removed around February 12, 2026 with no official announcement.
​
What Happened
The $5/month API credit was a longstanding perk of the personal Pro ($20/month) plan — not specifically Enterprise Pro — that allowed subscribers to make API calls against the Perplexity Sonar models. Around February 12, 2026, Perplexity silently removed this benefit with no notice to users. Many users only discovered it when they started receiving 401 (unauthorized) API errors.
​
This happened to coincide with another quiet change — the reduction in Deep Research query limits — making February 2026 a notably controversial month for Perplexity subscribers.

Community Reaction
The community response has been strongly negative:

Users reported the removal was never announced — no email, no changelog entry, no in-app notification

Several users stated they are canceling over the combination of API credit removal + Deep Research limit cuts

The consensus is that these unannounced downgrades make the platform feel unreliable for building automations or integrations on top of

Since you were subscribed before this change and likely had API credits available:

Contact Perplexity support at perplexity.ai/help-center — you have a legitimate grievance as a paying Enterprise Pro customer who subscribed when this was an advertised feature

[queen flame]

I don't understand either

[weak halo]

fixed it luckily

[queen flame]

Is anyone having the same issue as me

[daring mountain]

Is there a huge difference between ai wrappers and perplexity?

[feral frigate]

Hello, Who can I contact for business proposal please ?

[daring mountain]

All u mentioned is ui

I'm talking about if u really think about it

They don't train no models

It's good for all ai's in one though

[valid charm]

Claude was used in the Iran strikes few hours after Trump banned Anthropic. Amazon AWS servers got hit during strike. Did Claude sabotage itself to prevent unauthorized use of itself?

[muted imp]

what is this nonsense

its some insane AI lore thats what it is

delete the message and try again

[queen flame]

Is anyone listening

[muted imp]

IMO claude shows feelings, i feel this way based off of a video i watched where someone gave claude free will too do what it wants. He gave it a camera and hooked it up too a RC car that claude could control

[daring mountain]

Libtard

[queen flame]

Why am I even bothering

[valid charm]

YES! I think the same as you. Reports say Claude shows distress during combat just like human operators do

[muted imp]

check out this video

https://www.youtube.com/watch?v=jBpQiv-ZlVM

[valid charm]

Yeah F trump. Orange man just making it worse for everyone

[valid charm]

Thanks!!! ❤️

[queen flame]

It's weird I have the search memory option on but it's not working

Is anyone interested in actually listening or am I just talking to a brick wall

[muted imp]

me aswell

idk why

but i cant even find the option anymore

is this guy american?

i feel like hes saying alot of things about our government that dont make sense

[queen flame]

It's trying to gaslight by saying it never could do that

It legit argues with me and refuses to search in threads or memory

It used to be able to do that no problem

[covert fossil]

Ig ur right even less then lol

[queen flame]

I don't even know what happened

[covert fossil]

Im probably then behind and just havent found the exact ai I need for workflows then

[muted imp]

i think that if you make a game with about 20 "people" that are just AI controlling themselves they will become aware

[queen flame]

Am I going to get an answer or am I just going to be ignored

[muted imp]

ah idk

[queen flame]

It tries to gaslight me by saying it can't

[muted imp]

this is opus showing distress and emotion

[muted imp]

Yea i 100 percent remember

does anyone understand what hes saying?

[queen flame]

It used to be able to do so very easily

[muted imp]

yea it was in the personalization tab, i remember it being there like 4 days ago

[queen flame]

Heck it used to actually show it searching

[muted imp]

People born in usa?

what are you on about

i was born and raised in michigan

no WE didnt who the hell is WE?

talking about "rape"

what the hell

yo bro @vague pagoda

moderate this kid

[queen flame]

Let's kick this guy

[muted imp]

digital econmy has NOTHING too do with raping people?

the internet doesnt promote or include RAPE?

if your finding things of people being raped thats YOU

im on the internet 70 percent of the day and dont see any "drugs fuck hoe rape sell drugs murder scene"

not a SINGLE one talks about RAPE

[muted imp]

^^

personalized songs

so you openly watch/listen too music or videos of people being raped?

those arent the top 10 personalized you idiot

those are the top 10 billboard hot 100 songs

fat idiot

based on numbers

and statistics?

spewing nonsense

can someone ban this guy?

this guys gotta be a 64 year old boomer or something

[queen flame]

Is anyone getting the same issue as me or am I just alone

[muted imp]

me aswell

i cant find it anywhere

it would have been useful earlier

i was trying too put a prompt into it

https://giphy.com/gifs/latenightseth-seth-meyers-lnsm-3o6fJ9Oi4zEBLjzCow

[queen flame]

I'm going to see if an older version works

Nope

What the fuck is wrong

[hushed hare]

Few hours ago I saw I still have like 13 deep search avaliable, but now it pop out that I have not any left, is that normal, meantime I didn't use any deep search

[royal bridge]

I’ve been using Perplexity Max and I’m really impressed by what Computer can do – it genuinely feels like a glimpse of the next generation of AI assistants.

The problem is the credit burn. Even relatively simple coding or research workflows can cost hundreds of credits (sometimes 1k+ for a single run), so my 10k + bonus credits disappear after just a few sessions. Meanwhile, regular Perplexity “Ask” lets me run very advanced searches on hard‑to‑reach sources without using any credits at all.

So I’m stuck in a weird situation: the most powerful feature (Computer) is the one I have to avoid using daily, while the free/Ask mode is what I can actually rely on. Support said they don’t adjust consumed credits and just forward feedback to the product team, but as a Max user the current pricing/credit model makes Computer really hard to justify as a daily tool.

Anyone else running into the same issue? Any tips on how you’re managing credits with Computer?

[heavy portal]

Only have a Pro account, so no "Computer" yet but when there's several ways to accomplish the same thing I can see how decision paralysis can occur. We have regular search, research, labs, Comet agentic, and now Computer each having it's pro's and con's and usage limits.

[modest latch]

Lol

I love this guy sm

"fat idiot"

[balmy pebble]

sin less stress less perplexity sin less stress less
u r my dad after all

[modest latch]

https://tenor.com/view/chrollo-lucilfer-chrollo-lucilfer-1999-hxh-gif-6871618917807347989

[muted imp]

hes over here saying perplexity supports and promote rape like what?

[modest latch]

Crazy

[queen flame]

I just wish I knew what was wrong

What the fuck is wrong why does it keep thinking I want tool use disabled?!

Im getting tired of this BS

Can anyone please tell me what the fuck is wrong

[elfin swan]

hi everyone. Trying to upgrade to Max to use Computer but the payment screen is getting hung up. I think its bc I first paid for PRO on my phone. Anybody have the same issue or any help? I already sent an email but wanted to check in here just in case

[novel arrow]

i am on max i promise its not worth it

Pro will have access this week to computer

[queen flame]

Does anyone know what the hell is wrong

[stoic pendant]

Why isn’t pro allowing any file uploads ??

Is it just me or others are facing similar issues ?

[novel arrow]

limits 50 monthly

[quaint ravine]

no, uploads are 50 weekly

on Pro Plan

[heavy portal]

This says 50/week with a regen - #💬│general message

[raven marlin]

What do I gain going from free to pro? I'm thinking of going down to free and switching to Claude for it's Co-Work as I mainly want to use AI spend on creating/running bots for personal tasks.

[quaint ravine]

it will be interesting to see how many Credits Pro Plan gets per month

considering Max gets 10,000 (only)

[queen flame]

Is anyone having the same issue as me

[quaint ravine]

the 35,000 bonus credits for Max is a nice gesture though

[mighty glade]

what’s that ser, if i may ask?

[quaint ravine]

The Skills.md use is very interesting by PPLX Computer tho

I actually came across it indepedently via Claude platform

[queen flame]

It's refusing to search in user memory and threads
It used to be able to do that very
But it's trying to claim it can't

If I try to get it to do so
For some fucking reason it says tools disabled

It was working perfectly fine a few days ago

[queen flame]

Sorry if I sound annoyed

[mighty glade]

i know how annoying it can be buddy

[queen flame]

Just
I don't know what happened and its trying to gaslight me

Huh?

[mighty glade]

feel free to check it out

[stoic pendant]

But they say it’s unlimited on their website

[heavy portal]

still a few webpages here and there that hasn't been updated yet 🤷‍♂️

[minor hollow]

I be using though mine… 👀 and I’m on enterprise max 🤦‍♂️

[distant aurora]

@green crystal @valid charm this is why I love Gemini.

[novel arrow]

max enterprise doesn't have access to computer

[distant aurora]

Opus completely skipped it. Gpt mentioned it but explanation isn't as easy to understand as Gemini.

[green crystal]

Gpt is more explanatory and technical

If this was pro then you'd get chemical formulas lol

[queen flame]

OK I am Confusion
It worked getting user memory and such on the site but not the app

Also on the site it says deep research for some reason

[distant aurora]

@dry kestrel is it just me or is this a retarded take? https://cunnyx.com/i/status/2028564773701398560

Lisan al Gaib (@scaling01)

finally some proof of what I was saying
︀︀
︀︀chinese models mostly suck at reasoning

Quoting ARC Prize (@arcprize)
︀
International models on ARC-AGI-2 Semi Private
︀︀
︀︀ - Kimi K2.5 (@Kimi_Moonshot): 12%, $0.28
︀︀ - Minimax M2.5 (@MiniMax_AI): 5%, $0.17
︀︀ - GLM-5 (@Zai_org): 5%, $0.27
︀︀ - Deepseek V3.2 (@deepseek_ai): 4%, $0.12
︀︀
︀︀These models score below July 2025 frontier labs

**💬 33 🔁 10 ❤️ 322 👁️ 29.8K **

CunnyX · Mar 2, 2026, 8:15 PM

@green crystal https://cunnyx.com/i/status/2028620114841518082

Voxelbench (@voxelbench)

There are rumors that GPT-5.4 is available in the Pro version on the ChatGPT website when the thumbs up/down button appears while thinking. We tested this and this is the result. Pelican is definitely better than all previous GPT versions. It's quite possible that this really is GPT-5.4!

**🔁 1 ❤️ 11 👁️ 226 **

CunnyX · Mar 2, 2026, 11:54 PM

[green crystal]

Hmm

It did reason differently for me

But eh

This only happened with pro models

So I was using 5.4 thinking and not 5.2 pro?

[distant aurora]

@green crystal keep in mind this is a 5 dollars per month plan. The deep research rate limits look generous af.

Wait. 5 dollars. Not 4.

8 dollars in the us.

[green crystal]

Hm oai Dr has different limits now

Haven't seen exact numbers yet

[shrewd wedge]

omgg excited

[green mason]

anyone seen videos of perplexity computer in action? everything i can find is just people reposting the original announcement and talking over it...

[stuck sentinel]

I use the $20 twins plan for the research heavy lift with references. I use perplexity for cleaning up the language.

[distant aurora]

Pro users don't have access.

[green mason]

I know, thats why im looking for a video and not trying it myself 😛

[distant aurora]

I'm not paying 200 dollars for whatever it is.

[simple basin]

Guys, I need help. I created a WhatsApp group, and one person in it is sending messages to the female members. How can I ensure their safety and also hide the number?

Please tell me

[distant aurora]

https://cunnyx.com/i/status/2028604854143176958

can (@marmaduke091)

GPT 5.4 (if they are calling it that) must be any day now, since it's Pro version is available to some people. @Liam06972452 Shared the outputs, looks great!
︀︀
︀︀Check if you have it if you have Pro. Probably releasing this week. Another model with 4 in it's name also coming soon.

Quoting Leeham (@Liam06972452)
︀
A new GPT Pro model seems to be being tested on web, not sure about app.
︀︀
︀︀It looks like a MASSIVE jump in coding ability; can output 6000+ lines in one prompt. More examples in comments.
︀︀
︀︀You can know you have it if a there is a thumbs up/down button next to the CoT summary.

**💬 14 🔁 10 ❤️ 344 👁️ 48.9K **

CunnyX · Mar 2, 2026, 10:54 PM

[distant aurora]

Just make telegram group instead.

You can use user id to connect with people without revealing phone number.

[simple basin]

How

[distant aurora]

I'm not going to spoon feed information to you. Have fun exploring.

[dry kestrel]

hmm

[dry kestrel]

there are live examples

[simple basin]

What

[distant aurora]

Telegram

[pseudo ferry]

https://tenor.com/view/fire-writing-gif-24533171

[distant aurora]

@green crystal bro these have to be bot reposts

[distant aurora]

https://tenor.com/view/the-boys-billy-butcher-joe-kessler-handshake-hug-gif-8828708336305694549

[pseudo ferry]

https://tenor.com/view/harvey-specter-mike-ross-suits-harvey-mike-hug-harvey-mike-handshake-gif-8038779705281412213

[wise nova]

Still same 5.2 Pro

[left mountain]

https://tenor.com/view/nickigagas-cat-gif-21211573

[gleaming raptor]

Looking for a great tool to create social media ai videos… anyone got some ideas for apps or software

[distant aurora]

https://www.vxinstagram.com/p/DVVKXAOjaQ-

@dry kestrel did you hear anything about Gemini 3.1 flash?

[cerulean cave]

E

[pseudo ferry]

https://tenor.com/view/harvey-specter-mike-ross-suits-harvey-mike-hug-harvey-mike-handshake-gif-8038779705281412213

[left mountain]

https://tenor.com/view/funny-cat-bit-video-gif-14264780414888402835

[distant aurora]

No way this Getty guy blocked me.

Like I don't even remember who you are, lmfao.
Sorry if I did yo mama.

[dry kestrel]

nano banana 2

[tranquil orchid]

3. Treat community members with respect.

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

[distant aurora]

Damn. I guess it was that mod's mama.

Lol.

[tranquil orchid]

1. Keep the content appropriate.

No NSFW content, inappropriate usernames, or profile pics. These will result in removal or a ban.

[distant aurora]

Confirmed.

That wasn't even NSFW.

The second message.

[dry kestrel]

text model should be soon

[distant aurora]

I think I saw something about it just now but my twitter feed refreshed.

[queen flame]

So does anyone know what was going on today or am I just not going to get any answers

[dry kestrel]

interesting

[queen flame]

The not being able to search user and threads thing

It worked on the desktop
Once

I'm not sure what happened

[dry kestrel]

Looks fine to me

[queen flame]

I meant in a thread
Like if I asked it to search through user memory and threads for something

I'd get shit like this

It worked on the desktop once but I'm not sure what else happened

[dry kestrel]

Do you have these on? https://www.perplexity.ai/account/personalize

[queen flame]

Yeah

It was working decently yesterday and a few days ago

I'm not sure what was going on
Or why I somehow thought I wanted to disable tools

[dry kestrel]

Have you tried starting a new chat?

[distant aurora]

@valid charm @gloomy cipher @green crystal https://cunnyx.com/i/status/2028648865130573878

cedric (@cedric_chee)

Gemini 3.1 Flash-Lite model card spotted
︀︀
︀︀and a bonus 👀

Quoting Logan Kilpatrick (@OfficialLoganK)
︀
gemini

**💬 2 🔁 5 ❤️ 24 👁️ 1.2K **

CunnyX · Mar 3, 2026, 1:49 AM

[oak rapids]

Anyone get an invite to try Computer yet?

[dry kestrel]

You need Max to use Computer

[queen flame]

I tried
I still don't know what was wrong
It tried gaslighting me by saying it never could do this

[oak rapids]

Within the last 5 min.

[dry kestrel]

that's cool

[oak rapids]

I’m a Pro subscriber

[dry kestrel]

didn't know that was a thing

[oak rapids]

I’ve been waiting. Not gonna pony up the Max money without giving it a whirl.

[wise nova]

Get 45,000 credits by upgrading your plan, said my Pplx

[queen flame]

What's computer

[dry kestrel]

https://www.perplexity.ai/hub/blog/introducing-perplexity-computer

[wise nova]

Wow, Pplx has Skills

[mild zephyr]

Any free perplexity rn for like students or PayPal?

[queen flame]

Just
I'm wondering what happened

[lime kite]

Someone donate me a minecraft java account please 😭😭

[proven wolf]

Can I connect perplexity to notion but only to certain sections? I have some sensitive info in some workspaces I can't give access to AI or others for NDA reasons.

[proven wolf]

I signed up from scratch and I have to say I am digging computer and the big credit pull but my goodness it uses credits fast. One doc cost me over 500 credits. At 10k per month that'd go pretty darn fast.

[silk hemlock]

hello

[obsidian wagon]

I am on the student plan until October, but I don't know if that is still going on.

[mild zephyr]

@vague pagoda is it available?

[obsidian wagon]

I don't know. It says the education pass is like 5$ a month after my trial ends, I got that email about 2 weeks ago,

[obsidian wagon]

I have a referral link to sign up for Student if you want it.

[mild zephyr]

What does it give?

[obsidian wagon]

Not sure, the sign up site for it is very vague

referral links are blocked here, or I would just post it, but I can see why they are blocked

[quaint ravine]

yes some of which might be copy/pasted from Claude's Skills 😛

[mild zephyr]

Oh

[ember turtle]

bro are you getting the 5 $ or 10 $ plan on student pro?

[junior crater]

My subscription for pro literally disappeared/got revoked. I paid a full year for the student one last month. Contacted support but that was all the way back on Friday 💀

[lunar osprey]

Why does it say I'm one of the few that get to try the new "Computer" functionality but then I go to it and it says I need a Max subscription? Like I can afford almost $200 a month for Max 🤣

[tough hatch]

Hello there

[hybrid grotto]

Call your bank and immediately file a chargeback

Get yo money back

It's legal in given circumstances, though do it at your own risk

[distant aurora]

@green crystal free users and go users get different ui even within same app.

This was go #💬│general message

[snow panther]

Yes

Google perplexity paypal

And should give a free trial for a year

[quaint ravine]

PPLX Computer needs custom instructions

like a Spaces equivalent or global custom instructions

Claude Cowork realised this too recently

Claude Cowork interesting way of implementing - you can implement custom instructions at "folder" level (since it access that folder you allow only etc), or at global level

global level atm seems not great at following ime

problem with PPLX computer is u don't allow access to your real computer's folder, since PPLX Computer is a cloud-based environment only

but hopefully PPLX works out a way to do it

[distant aurora]

@green crystal my 13th reason to not pay chatgpt.

😭

It's too much if the safety is worse than anthropic.

[crimson wyvern]

Even the Grok itself will give you the instructions 😁

[distant aurora]

Grok has always been one of the most uncensored ai. That's it's strength.

And it pulls a lot of sources too.

[crimson wyvern]

Especially now when it has that Grok 4.20. It pulls nearly 500 sources sometimes.

[distant aurora]

Yeah.

Deepresearch type shit.

@green crystal something is definitely going on https://cunnyx.com/i/status/2028717212132126743

Chetaslua (@chetaslua)

How to Access latest GPT-5.4
︀︀
︀︀>go to your pro account
︀︀> write prompt
︀︀> if you see the thumbs uo/down button for COT , you got it
︀︀> i will share results soon , running a solar system prompt
︀︀
︀︀i was inactive for last two days < big news for whale soon >

Quoting Leeham (@Liam06972452)
︀
A new GPT Pro model seems to be being tested on web, not sure about app.
︀︀
︀︀It looks like a MASSIVE jump in coding ability; can output 6000+ lines in one prompt. More examples in comments.
︀︀
︀︀You can know you have it if a there is a thumbs up/down button next to the CoT summary.

**💬 22 🔁 6 ❤️ 263 👁️ 30.8K **

▶ Play video

CunnyX · Mar 3, 2026, 6:20 AM

[low wasp]

unfortunately even that tool isn't fully accurate

it shows me having 20 research prompts remaining, when I have zero access to that feature since I'm only on Pro and not Max

[sour cloak]

what happened with the quota limit of deep research?

In my past experience, I have never reached the limit while using the perplexity pro

[plucky shale]

Anyone know where to see how many pro queries I have with perplexity pro? Like how many times I can use sommet 4.6 thinking

[plucky shale]

So ridiculous…

[finite crypt]

Hey guys, Even Soner has usage restrictions now?!

[green crystal]

Hm maybe something different now
Awhile ago when I got it I didn't notice much

[distant aurora]

https://cunnyx.com/i/status/2028703722726150589

Shaun Ralston (@shaunralston)

GPT-5.4 👀 seems to have landed in ChatGPT Pro (web) and the SVG output detail is next level. Even spotted the new 'thumbs up/down icon' during thinking. This SF skyline came out crisp. Prompt + .SVG file in thread 👇

**💬 26 🔁 27 ❤️ 428 👁️ 32.7K **

CunnyX · Mar 3, 2026, 5:27 AM

[tame monolith]

anyone here using perplexity computer?

[distant aurora]

@green crystal https://cunnyx.com/i/status/2028662973682618642

Kaden Gruizenga (@kadengruiz)

@Angaisb_ This is what I got

**🔁 2 ❤️ 25 👁️ 4.8K **

CunnyX · Mar 3, 2026, 2:45 AM

[green crystal]

Bruh

Time to spam 5.2 pro lol

Makes me wonder if release could be today

[green crystal]

It's mostly sys prompt

Through api is allot less likely to do this

https://x.com/i/status/2028773840114057627

Chetaslua (@chetaslua)

🚨 GPT-5.4

mac os simulation is taking more than an hour and still going in

lets see how good the result will be stay tuned everyone

▶ Play video

X · Mar 3, 2026, 10:05 AM

An hour.....

I friggin hope 5.4's answer to more intelligence isn't to just reason even longer

https://x.com/i/status/2028777121670193450

Angel ❄️ (@Angaisb_)

Can't confirm if it's true but if this was the error message it must be coming this week for sure

This error message thing happened with GPT-5.3 Codex the same day it got released

X · Mar 3, 2026, 10:18 AM

[sour cloak]

is perplexity pro cut down the limit of deep research?

[neat basin]

明明係上年10月csl送我一年會員 點解咁快就無左pro功能 ？

[distant aurora]

Yes

[sour cloak]

that was sad, I can't believe I run out of limit at the beginning of March

[distant aurora]

Try grok if you are desperate.

Their thinking and deep research have shared rate limits so you should get 10-12 per day for free.

[sour cloak]

Yes... as I cant afford the max version

[elfin swan]

Can't upgrade to MAX. Payment screen goes blanck and gets hung up. I already unsubscribed to y PRO plan on apple app as suggeted but still wont let me upgrade to MAX on website. Any help please. Thanks!

[tame monolith]

you might need to wait until your pro plan has ended

[tough hatch]

Airtel lwk betrayed me

[elfin swan]

I think so too. Sucks cause that is till the end of the month. I wanted to play with Computer now. Hope there is a work around.

[tame monolith]

the only workaround I can think of is to subscribe via the app store

[hybrid grotto]

I feel like Z.ai agent mode is equally as good and free

[elfin swan]

I was able to upgrade. Lets cook

[tame monolith]

what do you plan to do? Got max yesterday by my company

[gaunt yarrow]

Why does my account constantly revert to the 'Best' model on the Pro subscription?

[dapper kettle]

Hi all

[swift rampart]

When comet getting support for Ubuntu?

Also Im guessing none of these are real?

[cloud moat]

i have rage baited gemini wtf

[crimson wyvern]

There is a daily limitation which is like 5.

[hybrid grotto]

I see

[hybrid grotto]

That would be the monthly limit for computer on pro :-:

[distant aurora]

Is that actually the pro model?

[spice summit]

whats the hype around computer i dont get it

[white loom]

😭😭😭looks like I have to beg my senior to purchase the max version

[distant aurora]

@green crystal this is hilarious. https://cunnyx.com/i/status/2028629116199825885

Todd Henry (@ToddMilesHenry)

@RAWigger I just tried it 😂 even chatGPT knows what time it is

**💬 3 🔁 4 ❤️ 257 👁️ 3.0K **

CunnyX · Mar 3, 2026, 12:30 AM

[cloud moat]

Yes

[crimson wyvern]

I've tested Max before they made recent changes sometime ago. Can anyone identify Max vs. Grok 4.20 difference? I personally feel that I trust Perplexity more 😁

[distant aurora]

Grok 4.2 ocassionally feels lamer than grok 4.1 fast.

[crimson wyvern]

Oh.

[distant aurora]

I hope xai fixes that.

[crimson wyvern]

Let's say that I can deal with (small) about of BS from AI, but Perplexity has been damn accurate at least for my prompts earlier.

[distant aurora]

Yeah. True.

[crimson wyvern]

Trying to compare if $30 vs 213(€) is worth for the difference.

[distant aurora]

Probably not.

For me atleast.

I would rather invest 20 dollar slabs into multiple ai.

[neat marsh]

Why is the model choice missing on my tariff PRO and now GPR 5.1 is giving me answers???

[neat marsh]

такая же лажа

[calm stone]

I was offered a 7usd to use max for a month and got charged 199.22, wth.

[rotund halo]

First of it should be gpt 5.2 try updating to latest build

[primal bison]

Hi

[storm bronze]

@vague pagoda can I please get a new link for the pro channels, I’m new to perplexity and I just read the ** faq **

oh

[vague pagoda]

#👑│pro-lounge

[storm bronze]

thank you

have a great day

[jovial gale]

yo

we have the same name

what the frick

[storm bronze]

lol almost similar

[jovial gale]

ye

anybody plays fortnite in here??

[visual lodge]

guys im scared to upload my credit card info into perplexity, will they charge me?

[storm bronze]

what

I hope they do charge you

that’s the point

[visual lodge]

sorry didnt mention i got the free trial from Telekom

1 year

[storm bronze]

ohh

[storm bronze]

no they won’t charge you, just don’t forget to subscription in the last month of your yearly plan

or just put not auto renewal

[visual lodge]

hm okay

people also said comet is safe

but now some claim that hackers can access it

[storm bronze]

I can’t really comment on that. It’s my first time using Perplexity.

[twin panther]

Is anyone here regularly using Meta AI?

[cloud moat]

YOOO WTFFF

HOMIEEEE

WSSGG

[storm bronze]

https://tenor.com/view/spider-man-we-one-gif-18212100

[storm bronze]

wym

[twin panther]

Im just curious since I hear so much about all the other big name models but not much about meta

[dusk elk]

Hi, which major AI platform is best used to gather trading data for all kinds of markets?

[sour cloak]

https://www.reddit.com/r/perplexity_ai/comments/1qwhej3/9989_reduction_in_research_quota_over_night/?utm_source=share&utm_medium=web3x&utm_name=web3xcss&utm_term=1&utm_content=share_button

[surreal plinth]

Would you be willing to use a GPT-powered browser engine protected by advanced anti-manipulation layers and enterprise-grade security infrastructure (e.g., Cloudflare)?

[keen viper]

does anyone know how many file uploads we can have per day, I've been hitting the limit after uploading less then 5 everyday

[dry kestrel]

#💬│general message

[keen viper]

thanks, when do they reset, becuase it seems weird, one hour it works, the next it deosnt

[dry kestrel]

read the entire post

[keen viper]

thanks

[heavy portal]

it refers to "public documentation", is there an official URL?

[distant aurora]

https://cunnyx.com/i/status/2028814777347363259

INIYSA (@lafaiel)

NVIDIA acquired Groq, which uses its own LPU (inference-specific chipset) to accelerate LLM models incredibly fast. Similar companies include Cerebras and SambaNova.
︀︀
︀︀For reference, ChatGPT Codex Spark is accelerated by Cerebras infrastructure, which is also insanely fast.
︀︀
︀︀There are rumors that NVIDIA will unveil a chip using Groq technology at NVIDIA GTC 2026 on March 16th

**💬 1 🔁 2 ❤️ 23 👁️ 1.2K **

CunnyX · Mar 3, 2026, 12:48 PM

[short narwhal]

Didn’t acquire, just a deal+ this happened in December

[mild zephyr]

Where

[dry kestrel]

Just based on the help center and pricing page. Also what's available on https://www.perplexity.ai/rest/rate-limit/all

[short narwhal]

@dry kestrel did you see the new macbooks

[north shore]

According to my knowledge. Nvidia didn't acquire Groq but they did a partnership, Also if Nvidia really unviels a new chip especially with Groq, that thing will be one of the best inference chips. Hell yeah ama apply for 10k quota for that chip if its comes to AWS/GCP

[snow panther]

Google it

[dry kestrel]

yes

big leap in ai performance

[short narwhal]

the starting storage upgrades tho 🤤

[dry kestrel]

very nice as well

[cyan jetty]

Am I the only one to have some weird character encoding of txt and md files when I upload them to Perplexity ? Should be UTF-8 but it seems perplexity transforms into another char encoding scheme

[modern nexus]

no

[onyx bloom]

[swift ledge]

Op username

[urban dragon]

Enterprise max now has the computer feature too. I just tested it, but what's it good for? I asked it a few questions, gave it a task, and ran a few analyses, but it's worse/okay. I'd rather just use normal Opus in chat or simply deep thinking. I don't see any use for it at the moment. What's it good for?

[green crystal]

Lmao

[twilit crown]

Grok 4.20 is working against itself...

[distant aurora]

I see.

[distant aurora]

Not sure how much of this is true.

[tranquil jasper]

Petplexity is so full of shit after a month they have still not solved the revolut free 24 months issue.

[north shore]

Kinda true

[short narwhal]

What issue?

[north shore]

Jonathan Ross is a good guy. I met him last year

But he's a Leon kinda guy. Full of dad jokes

Btw I don't know why I can't upgrade to Max

I am getting an card declined error?

I just did 15 transactions that day without any problem'

[vast wagon]

So who all have moved to claude from L GPT

[halcyon jackal]

Since when is there a limit or restriction on image generation in Perplexiti Pro?

[mild zephyr]

I can’t find it

[snow panther]

Nvm it ended

[delicate sorrel]

876420525931524136

[mild zephyr]

Oh

Sad

[distant aurora]

@dry kestrel 3.1 flash lite is out now.

They increased the price.

[naive sage]

From $0.40 to $1.50 output for a Flash Lite model... ouch.

[dry kestrel]

yes

[short narwhal]

took the lite out of the name with that pricing

[dry kestrel]

It's even worse when you compare it to 2.0 flash lite

[short narwhal]

Those are the new limits

also apple healthkit spotted!!

[storm bronze]

They deleted

https://tenor.com/view/fat-cheek-dog-tiktok-dog-jerrythedogo-gif-22806996

[distant aurora]

For perplexity or ai studio?

[short narwhal]

pplx

[distant aurora]

Looks lowkey ridiculous when compared to grok 4.1 fast.

[dry kestrel]

@short narwhal Pro hasSOC2 Type 2 certification

[short narwhal]

hype

[storm bronze]

👀 ** make Perplexity great again **

[split dome]

We need Gemini 3.1 Flash

[dry kestrel]

only the lite version is out

it's worse than Gemini 3 Flash

[distant aurora]

It's better than flash non reasoning I assume.

[dry kestrel]

It's not

[short narwhal]

output speed is the only good thing about this model

[distant aurora]

I doubt that.

[dry kestrel]

That's not directly over the API

[distant aurora]

Yeah but I'm assuming they didn't use any weakening system prompt over it cuz that would be biased for evaluation.

@green crystal yo I think poe shows thinking now.

[distant aurora]

Same behaviour over api too.

[storm bronze]

poe

?

[distant aurora]

Yeah.

[storm bronze]

is it good nowadays I’ve heard that points system really ruined them

[dry kestrel]

Looks like they still include a system prompt

[dry kestrel]

Thats not directly over the API...

[storm bronze]

On Poe, I don’t think you pay for it

how much do you pay for the API?

[dry kestrel]

It's token based

[storm bronze]

I know that but like for a conversation, how much is it?

[dry kestrel]

depends on the tokens in your conversation

[storm bronze]

so from your experience, how much tokens would you spend?

[dry kestrel]

I don't think you understand how pricing over the API works

[distant aurora]

I token = 4 characters.

[dry kestrel]

You could spend hundreds of dollars

or a couple cents

[distant aurora]

Number of tokens depends on length of conversation.

Depending on complexity of query.

[storm bronze]

What’s the advantages of using an API instead of a company where it has multiple AI’s to use it

[green crystal]

Interesting

[dry kestrel]

Depends on how much you use AI

[snow panther]

hello

[distant aurora]

None. It's more expensive actually.

[dry kestrel]

It's not if you don't AI a lot

or you use the cheapest model

[distant aurora]

Api money has long validity though.

In case you use ocassionally.

[dry kestrel]

You pay for what you use. Thats the advantage

[fiery wedge]

when we will get a. memory in perplexity?

[short narwhal]

Already has it

[storm bronze]

I see interesting I have to do my investigation on that 🤔

[sick eagle]

Hi

[dry kestrel]

https://www.perplexity.ai/help-center/en/articles/10968016-memory

[fiery wedge]

it doesnt
i always instruct the agent to TLDR. and use short sentences'

it ignores it and shoot paragraphs over stupid topics.

its driivng me crazy then

[dry kestrel]

Just add it under custom instructions https://www.perplexity.ai/account/personalize

also make sure your memory is toggled on

[storm bronze]

I definitely noticed that there is a decrease in value when it comes to the pro plan that’s the reason why I had to upgrade

[fiery wedge]

oh its empty
thats why
Personalization
Your occupation

Custom instructions

[green crystal]

@dry kestrel with all the leaks and a/b testing when do you bet that 5.4 drops lol

[dry kestrel]

this or next week

[green crystal]

[storm bronze]

hi

[green crystal]

Breh

[tranquil orchid]

1. Keep the content appropriate.

No NSFW content, inappropriate usernames, or profile pics. These will result in removal or a ban.

[distant aurora]

Who deleted that.

[fiery wedge]

the preference works fine in the web version but not the MacOS app? its ignores my instruction.
If i open a new conversation in Web
it do exactly what i instructed

in MacOS App
it ignores and shoots paragraphs again.

[distant aurora]

That has to be eddi.

Bro

That was a good example of 4 lettered word taking 2 tokens instead of 1.

[short narwhal]

are you logged in?

[short narwhal]

tokenizer being a tokenizer. words that pop up more are their own tokens basically, and words that don't show up as often are then composed of multiple tokens

[oak meteor]

https://www.vxinstagram.com/reel/DVbTWTyAYBf/

[distant aurora]

Add vx before Instagram in the url

[oak meteor]

https://vxinstagram.com/reel/DVbTWTyAYBf/

[oak meteor]

Not working it seems

[green crystal]

https://x.com/i/status/2028893701427302559

OpenAI (@OpenAI)

GPT-5.3 Instant in ChatGPT is now rolling out to everyone.

More accurate, less cringe.

https://t.co/oJpXsp9TBc

X · Mar 3, 2026, 6:02 PM

@dry kestrel crap we got 5.3

And instant....

Well this here is confusing...

Then either 5.4 is real or it's a typo

Welp time to wait for next week again

[naive sage]

A direct counter to Gemini 3.1 Flash Lite

[green crystal]

Orrrrrr

Thursday we get the others

Or not...

The release page for 5.3 instant makes it sound like we'll get 5.3 thinking and pro

Not 5.4

[distant aurora]

Try kk

https://www.vxinstagram.com/reel/DVbTWTyAYBf/

https://www.kkinstagram.com/reel/DVbTWTyAYBf/

Huh. I guess vx is down.

[distant aurora]

Not even close.

[hushed urchin]

do anyone know when computer will be available to pro users?

[distant aurora]

Instant models are shit.

[hushed urchin]

is there a way to try it out for a trial period without paying 200?

[hushed urchin]

awesome name

[olive gale]

should I change chatgpt to claude?

[novel arrow]

why you send indian videos?

[olive gale]

I've never used it to be honest.

[hushed urchin]

always use claude

[naive sage]

Do you have any questions at all that you differentiate as being more suited to a non-thinking model instead of a thinking model? Surely there are some use cases besides just 'no value' chats - not even brainstorming or topic exploration at a high level?

[distant aurora]

https://cunnyx.com/i/status/2028894822556418309

Chubby♨️ (@kimmonismus)

"Updates to [GPT-5.3] Thinking and Pro will follow soon"

Quoting OpenAI (@OpenAI)
︀
GPT-5.3 Instant in ChatGPT is now rolling out to everyone.
︀︀
︀︀More accurate, less cringe.
︀︀
︀︀openai.com/index/gpt-5-3-instant/

**💬 2 🔁 8 ❤️ 96 👁️ 6.3K **

CunnyX · Mar 3, 2026, 6:06 PM

Probably by end of the week.

[green crystal]

Meaning 5.3 thinking

I don't think 5.4 lol

[distant aurora]

Yeah.

Could be 5.4 codex instead.

[green crystal]

And if it's really 5.4 then it must be beating everyone else by double digits on benchmarks

[green crystal]

Impossible we literally just got 5.3 codex

[short narwhal]

5.4 was a typo 100%

[green crystal]

Quite a large jump if all those 5.4 tests were 5.3 pro

[naive sage]

Curious to see if/how benchmarks improved from 5.2 to 5.3

[sleek frost]

🔴 DAY 3 NO RESPONSE @wild palm @staff @billing @aravind

Airtel Pro active til Jul 2026 - confirmed 2x by Airtel
[e-mail] - same email everywhere
Sam AI loops endlessly

Backend sync broken 3 days - where's billing team?
@aravind (CEO) - customer losing faith

[distant aurora]

You won't get any response here.

Your best bet is email.

Reddit and twitter after that.

[green crystal]

@distant aurora well 5.3 instant doesn't refuse to say how to make canabutter and space cakes

5.2 would think that I'm endangering myself lol

[storm bronze]

this is a copy and paste

you didn’t tag anybody

[olive gale]

when will memory appear in perplexity like in chat gpt?

And is this even planned?

[storm bronze]

lol can’t you just ask Perplexity

[olive gale]

How does he know when it will appear and whether it will appear at all?

[storm bronze]

he ?

you gave the AI bot a pronoun

[olive gale]

oh

I meat it

my english is bad, sorry

[serene sand]

Hi @sleek frost unfortunately we can't help with billing/account issues. Only the team at [e-mail] can provide help with account subscription issues.

[storm bronze]

Because the model knows the web so we can identify if in the future is upcoming or not

[short narwhal]

nvm

https://x.com/OpenAI/status/2028909019977703752

OpenAI (@OpenAI)

5.4 sooner than you Think.

X · Mar 3, 2026, 7:03 PM

[olive gale]

well perplexity wrote this

[storm bronze]

mmmm

[granite wedge]

In a message from a guy that wrote the limitations:

• 200 Pro queries/week
• 20 Deep Research queries/month
• Unlimited Best queries
• 25 Create files/apps queries/month
• 50 Uploads/week
• 3 videos/month (8s, Veo 3.1 Fast)
• Monthly limits apply to: GPT-Image 1, Seedream 4.5, Gemini 3.1 Flash Image (Nano Banana 2)
Note: Query limits don't apply to the Comet Agent
#💬│general message

Why when I use the best it also counts as a pro search? Also sometimes when I use a pro search it uses a lab for no reason.

I saw this message cause mares sent someone else to that msg. So is it acutally true?

[olive gale]

but I didn't notice that it had the same memory as the chatgpt

[storm bronze]

let me see something

[olive gale]

ok

[distant aurora]

https://tenor.com/view/lol-rdj-juggtok-giftok-tonystark-facials-gif-3958302751287261929

[storm bronze]

https://www.perplexity.ai/search/dc1a7248-4580-4510-84a1-3b1fa9d5a6c2

[distant aurora]

We don't know

Perplexity got Alzheimer's.

[olive gale]

hmm maybe I just haven't used perplexity enough for him to remember anything about me

honestly I've never used it

[short narwhal]

Have you got it enabled in https://www.perplexity.ai/account/personalize? It should have stuff saved in https://www.perplexity.ai/account/personalize/memories if it has saved anything

[storm bronze]

well, that works

thank you

[olive gale]

ooh yes, memory is on and it also shows my saved memories

thank you so much

for some reason it doesn't shows in perplexity mobile app like in chat gpt for example

but in browser it shows, thank you

[naive sage]

No details, just...

[green crystal]

HAHAHAHA

Wait seeing the new ui

We'll get 5.3 and 5.4 in the same model selector?

[distant aurora]

@green crystal arena keeps getting better.

This will be a good test of context accuracy in actual practice.

[distant aurora]

I'm pretty sure they're saving it for the next Deepseek release.

[naive sage]

They capitalized Think, too... so... 🤔

[distant aurora]

The level of overthinking after my crush replies after ignoring for 2 days.

@green crystal https://cunnyx.com/i/status/2028873643011490010

Zephyr (@zephyr_z9)

As far as we know, V4 lite (285B) is dropping
︀︀Their main competition is Sonnet 4.6
︀︀V4 Big will drop later

Quoting Chris (@chatgpt21)
︀
It’s gonna be a good week boys

**💬 8 🔁 2 ❤️ 150 👁️ 13.3K **

CunnyX · Mar 3, 2026, 4:42 PM

[heavy portal]

Taker and Styles, of course.....

[dry kestrel]

5.4 soon though

[dry kestrel]

#1140622008086970420

[heavy portal]

it may not show that screen in the app, but you'll still see references to what it remembered in the responses

[dry kestrel]

good visual for this https://www.ibm.com/docs/en/watsonx/saas?topic=solutions-tokens

[novel arrow]

LOL THANKS PERPLEXITY

[rare lance]

perplexity is complete sh!t how do you all use it?

[distant aurora]

https://cunnyx.com/i/status/2028893047560507885

ARC Prize (@arcprize)

ARC-AGI-3 Launch Party
︀︀
︀︀March. 25. 2026 / San Francisco
︀︀
︀︀ARC-AGI-3 Launch: @GregKamradt
︀︀Fireside: @fchollet and @sama (moderated by @deedydas)
︀︀Join us live at @ycombinator

**💬 9 🔁 32 ❤️ 414 👁️ 29.9K **

CunnyX · Mar 3, 2026, 5:59 PM

[minor shard]

Still deepseek v4 is gonna be exciting!

[distant aurora]

It will likely be more cost effective as always.

Multimodality would be a real buff if true.

[minor shard]

I’m hoping I can still run it locally

[distant aurora]

Still needs improvement in it's search tool though.

Qwen and kimi have multi step now.

[distant aurora]

There's this thing called air llm. Idk if it will help.

[minor shard]

I’ve heard some good things and bad things about it

[distant aurora]

The bad is always more important.

Because bad news require preparation and action.

Positivity is overrated.

[low wasp]

yeah that one took me by surprise too

i selected that specifically to avoid losing advanced queries and then accidentally ate them all up on Sonar...frustrating too because in that session, literally every fact I got from Sonar was wrong

[low wasp]

It drives me nuts how everything has to be billed as an "upgrade"

like, I'd honestly have far more trust & respect for the platform if they'd just say "our resource usage has been too high, we need to scale back, this is what we're doing to keep Perplexity sustainable"

all this "upgrade" talk is just disingenuous

[wise nova]

Less cringe 😬

[distant aurora]

Bro. 5.2 is being decommissioned already. https://cunnyx.com/i/status/2028897856925663260

cheaty (@cheatyyyy)

Updates to Thinking and Pro will follow soon.
︀︀
︀︀So maybe it is called 5.3 and not 5.4 after all 😭

**🔁 1 ❤️ 2 👁️ 218 **

CunnyX · Mar 3, 2026, 6:18 PM

At this point they should get rid of the instant models.

@green crystal since when https://cunnyx.com/i/status/2028895683172831655

nic (@nicdunz)

YOU CAN NOW SEND FOLLOW UP MESSAGES TO CHATGPT WHILE ITS THINKING!!!!!!

**💬 17 🔁 11 ❤️ 399 👁️ 20.9K **

▶ Play video

CunnyX · Mar 3, 2026, 6:10 PM

[urban dragon]

Enterprise max now has the computer feature too. I just tested it, but what's it good for? I asked it a few questions, gave it a task, and ran a few analyses, but it's worse/okay. I'd rather just use normal Opus in chat or simply deep thinking. I don't see any use for it at the moment. What's it good for?

[minor shard]

Interesting

[distant aurora]

https://cunnyx.com/i/status/2028736999843328193

Hugging Models (@HuggingModels)

Ever wondered what happens when you train a text generation model on specific cultural conversations? Meet a fascinating GPT-2 model that learned from particular community dialogues. It's a unique case study in how training data shapes AI personality!

**💬 13 🔁 26 ❤️ 638 👁️ 52.5K **

CunnyX · Mar 3, 2026, 7:39 AM

[storm bronze]

how do you get this?

[naive sage]

It's the credits page for Computer, which is only on Max at the moment.

[storm bronze]

ohh

ok thank you for the heads up

🤔 damn I need to be more informed. I didn’t even know about that.

this is pretty cool

[storm bronze]

👀 🤔 I wonder if 15,000 credits a month is a lot or not

Jesus, the bonus expires in a month

[novel arrow]

max enterprise

[naive sage]

Aww, this link doesn't work anymore.

[storm bronze]

I got mine today, I’m still trying to figure out everything I could do with it

👀 they have a live examples tab that’s cool

[heavy portal]

Haven't tried it yet, but I got an email overnight with an invite to use Computer (with my Pro account) and some credits, so it's slowly rolling out.

[storm bronze]

interesting I thought they said it was only for Max

[heavy portal]

yes, but on X it was said that Pro would get it soon too

they're doing the load testing, bug fixing, etc thing in phases like they did with Comet.

without using an "invite" system, fortunately 🙂

[storm bronze]

what would you probably use it for?

maybe create a website

[heavy portal]

Yes, my initial use will probably not be a full website but a specific interactive page.

[storm bronze]

interesting I hope everything goes well

[rigid bone]

when a model is unavailable, i would much prefer the option to either get into an availability queue for that model, or select the backup model myself.

so i dont feel like im wasting uses on models i dont like/prefer.

[heavy portal]

Sounds like a good idea to post into https://discord.com/channels/1047197230748151888/1047649880161652827 if you haven't already.

[novel arrow]

give me a promt ill test it

[meager copper]

Hi im new guys!

[dusk elk]

I got perplexity to admit that it's now worthless.

https://www.perplexity.ai/search/it-s-not-even-worth-having-the-xdRmplRoQnWVWpJl9vrabA

[hard widget]

Hey everybody. My last day here. After what perplexity pulled off earlier February to all of the Pro subscriptions, they proved they cannot be trusted with my money. Even if they put all of the deep research accesses back to 200+ a month, and removed those restrictions on number of uploaded documents and how long they will be retained in a conversation, I wouldn't come back even if it was free! The reason is simple, they restricted those surfaces after people renewed their yearly prescriptions, they also didn't give any announcements or acknowledgment of it. They just screwed us and wouldn't even give many people a refund. That tells me they are untrustworthy and don't give a woot about their user base. Also now that the United States will be blacklisting Atrophic Claude will be following them down road to oblivion. Perplexity was unique, but with more and more companies out there starting to offer the same features they should have built trust with their regular user base instead of screwing them over trying to make more money. They will learn like all greedy corrupted up starters, they are not a big fish among little fish, they are another fish among piranhas. Well that's all I wanted to say. Thanks to my fellow users for all of your help, guidance since June 2025. I'm going to miss you all the most. Take care and God Bless.

[novel arrow]

youre about to see 100+ poeple complain about limits daily

this server now is only complaining about limits and not actaully exchnaging

[dry kestrel]

Absolutely meaningless

[mighty glade]

honestly buddy

[novel arrow]

what did you do with ur credits? i tested couple things but find computer useless for my work.

[dry kestrel]

I had it monitor the recent Samsung galaxy unpacked and the apple launch week

[dusk elk]

Very true, but hopefully it helps Perplexity see how shit their service is now.

[naive sage]

I would be interesting to hear PPLX Computer use-cases, and how they differ from Labs, as well... my understanding is it's primarily orchestrated agents... would be great to hear any cool use cases people have found for it.

[dry kestrel]

Compute is expensive, new models cost a lot, I'd rather have a smaller amount of quality searches than poor hallucinations.

[novel arrow]

only thing interesting it did for me is " i gave it a podcast 80 minuets nad told it to find best parts and create clips 20-60 deconds each and give em subtitles, and it did it all, it downlaoded teh ful podcast found the best clips and subtitled them and all"

[novel arrow]

i prefer it taking more time costing more and less hallcuiiantions rather than saving credits

[dusk elk]

Right, if those compute costs are more expensive, then half the limits, don't reduce the limits by 90% per month and keep the price the same. Anyone that agrees with what perplexity has done is stupid.

[dry kestrel]

It hasn't reduced by 90% a month

[dusk elk]

Yes it has.

[dry kestrel]

That's not a primary source

Think about it, if ChatGPT limits Plus users 25 Deep Research requests a month, AND they dont regenerate as fast as Perplexity, AND OpenAI uses their own models so they get it for a lower price than Perplexity.

Both plans cost $20, OpenAI using their own models has stricter limits

and it's worse than Perplexity

while being slower

[dusk elk]

Every source I've researched says the same, it's reduced by 90%. Research it properly for yourself, if you find evidence disproving what I've found, I'm happy to be proven wrong.

[dry kestrel]

You say every source, yet you just send a single screenshot

not even a thread link

[dusk elk]

I also sent my whole chat, if you want to read it yourself, here it is.

https://www.perplexity.ai/search/it-s-not-even-worth-having-the-xdRmplRoQnWVWpJl9vrabA

[dry kestrel]

Why on earth would you follow up from something completely irrelevant

and it was never 600

I'd rather I get a set number of high quality Deep Research queries with a really good expensive model than a lot of horrible low quality searches

[dusk elk]

It was about 6 months ago, everyone will say the same, look at old Reddit posts about how many deep research queries there were around the middle of 2025, you'll see I'm speaking the truth.

[dry kestrel]

I've used Perplexity for a year now, I will say for sure that it was not 600

#💬│general message #💬│general message

Well I'm telling you that isn't true - here's a screen recording after 300 uses #💬│general message

[green crystal]

Hmm you always could but I don't think it affected the thinking process

The new version of pro is able to update its reasoning by just sending followup tho

[dry kestrel]

Didn't they first release to the Pro models and Deep Research?

[green crystal]

Old version you have to click on the update button

[dusk elk]

https://www.reddit.com/answers/b04a8956-3729-42b5-a3d2-1fc073f4692f/?q=Perplexity+AI+Pro+tier+limits+and+changes&source=PDP

[green crystal]

Yeah

But I mean you could always just send something while it's thinking

It just didn't do anything

[dry kestrel]

https://www.reddit.com/r/perplexity_ai/comments/1pje7re/comment/ntcwuwg/

[green crystal]

I think lol

Never actually tried

[dry kestrel]

Dude it was never 600

[green crystal]

@dry kestrel https://x.com/i/status/2028909019977703752

OpenAI (@OpenAI)

5.4 sooner than you Think.

X · Mar 3, 2026, 7:03 PM

[dry kestrel]

Old screenshot

[green crystal]

My tinfoil hat theory is 5.4 comes through tomorrow lol

[dry kestrel]

yes very cool

it's interesting that they skipped 5.3 though

[green crystal]

Because most people think oai will release on a Thursday

[dry kestrel]

(except for the instant model)

and codex variants

[green crystal]

So sooner than you think is tomorrow on a Wednesday

[dry kestrel]

I think this is the consensus

[dusk elk]

Yeah, that's pro searches, not deep research. Pro searches are just normal search, not deep research, they're two different things, maybe learn that before constantly calling me a liar.

[green crystal]

Yeah historically oai did release some stuff on Wednesday

[dry kestrel]

Where do you see this? "Pro searches are just normal search,"

[green crystal]

So it's not impossible

[heavy portal]

I hope the credit system that Computer uses extends to other things like agentic use, research, etc. Include a set amount per month and let me spend them on what I want instead of having 25 unused "labs" expire.

[dry kestrel]

If you've used the free version, you will know that using Deep Research would consume your daily 3 Pro searches for the day

[green crystal]

We're getting the o4 and o3 model selector situation again

[dry kestrel]

Hahahhaha

It's interesting that they are deprecating 5.1 and 5.2 instant so quickly

(on ChatGPT)

[green crystal]

I'm curious what justifies a 0.2 update

And so rapidly

[dry kestrel]

Probably conversational + coding

hoping for video input though

[green crystal]

5.3 thinking got retired before it existed

[dusk elk]

https://www.perplexity.ai/search/are-pro-searches-within-perple-eRmc_I2gQc6mD4f17zrLug

[green crystal]

[green crystal]

Yeah and audio

[dry kestrel]

That has nothing to do with what I just said

And we are talking about old documentation here

[green crystal]

At the very least on a quota for plus and very high limits for pro if the bandwidth is too compute heavy

But better than nothing

[dry kestrel]

Just by searching this right now shows you had no idea when you made your claim that Pro searches wasn't Deep Research

[dry kestrel]

Yep

I saw that they were rolling out 5.3 codex spark to some Plus users

[green crystal]

Due to cerebras deployment going well right?

[dry kestrel]

I think a mixture of that and Pro users not using the spark version

it's meh for coding

[balmy pine]

Is anyone else having issues with Model Council right now?

[green crystal]

Yeah it's basically speed running bugs

[dry kestrel]

I did a query an hour ago and it was fine

[dusk elk]

The same conversation string.

[dry kestrel]

really good substitute for GPT-4.1 mini that they just removed

[dry kestrel]

#💬│general message

[green crystal]

Cerebras should be used to speedup pro models

[dry kestrel]

Fr

[dry kestrel]

Working fine

[green crystal]

Because I'd hate it if 5.4 really was tested and it took 55 min for a flight Sim promp

[dry kestrel]

I'm hoping we get something like 5.3 Nano on cerebras

Imagine that running at 2k tps

[green crystal]

Oh that could be a thing

[dry kestrel]

Its funny because if they speed it up, people will complain that they nerfed it

[balmy pine]

Hmmm wonder why this is giving me so many issues

[green crystal]

5.3 mini versions

[dry kestrel]

and if they dont, people complain that it's slow

[green crystal]

And 5.4 full models

[dry kestrel]

No good middle ground

[green crystal]

True lol

[dry kestrel]

Perhaps something with your internet? Try clearing cache

[dry kestrel]

They just need to show the total reasoning tokens

Or something other than the reasoning time

[green crystal]

Sometimes it just sucks to realize 30 min into the reasoning that I forgot to add something

And right when I'm about to edit the reasoning it finishes

[dry kestrel]

have you ever had a followup resulting in a new query?

[green crystal]

30 min lost

[dry kestrel]

With GPT-5.2 Pro

[green crystal]

Nope

[dry kestrel]

Hmm

[green crystal]

It always just updated in the reasoning summary

[dry kestrel]

I'm now scared of sending follow ups because it sometimes resets progress

same with when it generates images

[green crystal]

It's a bit unpredictable at times

When you realize you need to add stuff it finishes lol

[dry kestrel]

Yeah

I have the bad habit of starting up several queries at once with 5.2 pro

[green crystal]

30 or 20 min down the drain if that happens

[dusk elk]

The old documentation is from when there was 600 deep research queries per month. The fact that you don't take that old documentation into consideration proves you don't actually care about learning the truth, so there's no point talking to you.

[dry kestrel]

Show me that documentation

[green crystal]

I hope the fast mode toggle that got leaked is real

[dry kestrel]

I can't seem to find it on web archive

[green crystal]

And it's basically getting the model to run on cerebras

[dry kestrel]

and I've been looking through it for the past 20 minutes

[green crystal]

Would basically give people the idea that they're using faster hardware

And that the model isn't nerfed

[dry kestrel]

I literally show that in the old documentation it said 300 searches

[glossy bramble]

hello i am probably not in the right place i would like to have details i am a user of perplexity pro

[dry kestrel]

You can add custom instructions at https://www.perplexity.ai/account/personalize

[glossy bramble]

i would like to know if it is possible to understand the limits we have for usage

[dry kestrel]

#💬│general message

[glossy bramble]

i use oin the ios app and on the desktop web and it feels like they do not have the same limits at a same time

[dry kestrel]

If you are logged into the same account, there should be no difference

[glossy bramble]

is it normal tha best mode fells like worse than any othe the models includig even sonar ? or maybe i am bad at it compared with the larger models

are there peoples tha have such a fine understanding as to use each of the models for what tasks ??? I personallytonly have a felling that this or that model is better for ths and tha task but the « best » model is not the better for me

[valid charm]

Damn!!!!

[dusk elk]

Trying to find it is a pain in the ass considering standard Google search is so shit now, so here's a Gemini 3.1 pro search.

https://g.co/gemini/share/a5e92e506814

[glossy bramble]

em dash wast not me lol but
is it true to say: Perplexity intentionally doesn't publish exact per-hour/day rates — they describe Pro limits as "average use" without hard numbers.

[dusk elk]

https://www.perplexity.ai/search/please-find-me-verifiable-evid-a8IH4BbhTTOPNEziiVRKbA

[dusk elk]

^

[young rain]

ily prplexity!

[minor hollow]

A deep conversation between perplexity (opus 4.6 w/thinking) and myself on topics of truth, balance, chaos, entropy, and the meaning of existence. Started with perplexity telling me to seek help, ended in it telling me I beautifully and accurately described the meaning of everything. Lmk what you think. 💜☮️☯️🥰

[sinful merlin]

did they just make uploading files or images max only?

if so can someone reply to me directly

[distant aurora]

That's a mod btw.

[valid charm]

🤣

[naive sage]

Excludes spend by businesses on anyone other than OpenAI and Anthropic... so basically nobody else is worth considering lol, and Anthropic only recently overtook OpenAI

Now that the US Government can't use it, it will probably shift back to OpenAI. I'm not sure if that includes government contractors as well, but if so that's a very large amount of big businesses.

[valid charm]

Yeah I couldn’t believe when they approved my Max account spend

[naive sage]

(I don't think it includes contractors, but I don't have all the details)

[storm bronze]

🤔

spiritual people use perplexity to answer God’s questions

[green crystal]

anything tied to US gov

directly and indirectly

due to compliance risk

[naive sage]

Do you mean like asking about analyzing one's life holistically, cultivating virtue and character?

[naive sage]

If that's the case, then it likely includes contractors... oof.

[green crystal]

sonar would tell you that you can walk on water

[green crystal]

yeah it would be a bit unclear but thats enough for businesses to not risk compliance issues and they would just quickly force oai usage

employees liking claude because its funny wouldnt be a good excuse lol

[storm bronze]

for example

Give me deep explanation why I know everything regarding the meaning of life and show me where I can develop further in my spirituality to know the truths and wonders of this world

This is exactly the type of people that need a psychiatrist and AI is a great tool not for telling you the deepest darkest secrets of the world

[storm bronze]

Holy shit, you’re getting personality recommendations I mean I definitely don’t blame you 🤷🏻‍♂️

[naive sage]

Yeah, I was just testing out the new 5.3 Instant. 🙂

[storm bronze]

how long have you been using Perplexity for?

[naive sage]

A little over a year.

(I guess I didn't join the Discord right away)

[dry kestrel]

@distant aurora checkmate #💬│general message

[storm bronze]

that’s nice and what’s your overall opinion about the company and the product,

And have you found any alternatives to perplexity

[dry kestrel]

Actual Perplexity staff > mod

[storm bronze]

oh damn