#💬│general
1 messages · Page 312 of 1
You Selling Those?
This one is good what wad the full prompt
Opus is very expensive
what is the cheapest option
yeah its crazy
Although Perplexity Max has Unlimited Opus
I need you to help me design the first 3 pages of an adult coloring book called kittens and cookies. make it in adult coloring book style, only black and white with lots of white to fill in. I want very detailed texturing on the cats and the cookies, with distinct styles for each cat and each cookie on each page.
then
ok now in the same style give me 5 coloring book pages about dogs with jobs. maintain the differences in textures between the job elements and the dog elements so there is clear contrast between them.
it charged me like 100 bucks or smh
Vscode with extension
But 32k Context Window
the cats and cookies were good too
Wait you selling these?
can i use 4.1 opus like cursor?
I think he's asked for AI Images
But you don't have a bad idea
Right
yeah adult coloring books apparantly are super easy to get into
Nice
Well let me know if you get any success
I thought you wanted cheapest....
You could use claude cli or vscode extension with the model opus 4.1 but that model is expensive in the first place
And as well for the others
It's just about time for they to realize it hha
im toying with the idea of listing a book i vibe wrote on amazon
figuring out how to use latex and stuff atm
Wait what are adult coloring books?
😠 😠 😠 😠 😠 😠 can someone enlighten me with correct prompt?
coloring books for adults
basically much more complex than coloring books for kids
Same as kids coloring book but more complex and good arts made for adult level thinking
search amazon for adult coloring book
So You Mean Like Hot Stuff?
one could design an agent that does nothing but churn out that kind of art
What is this
and automate assembly into publishing format
Bro do you not know what kids coloring books are?
Do people actually buy these?
These are just a more complex version
I know im confused about the adult
its a book i wrote with AI about a dude who discovers a meteorite containing trillions of alien gold on his property and uses it to overthrow governments
Ahhh gotcha
I think what he means by 'hot' is viral/popular or smth
Haha that's great
yeah genius for some reason they are super popular
It's called adults because it doesn't contain cartoon characters instead its art such mounts and mountain something that kids wouldn't like
Now i Completely Understand Tysm!
I have also discovered Books/Web Novels Niche Recently from Astra
Dude Makes 20k$ A Month
A hi broke stepfun lol
What's Astra?
Writing Books and Web Novels with Ai
A Person Here
New Bulgarian model
So it's an AI tool?
Oh my bad
My man
Yes new Bulgarian model
Bruh
😂 🤣
Lmao

How is opus opensource
Maybe in a alternative timeline
What is supposed go be funny I don't understand 😑
gotta be a bait tweet
smh
Oh
Your ai didn't catch that
OPEN SOURCE
Oh btw
Does the pro models on the pro plan have limits?
Yes
Like the pro searchs
We Need Chinese Claude with Same Performance
open sauce
300 perday
I can't type & pato can't read
Oh, how much?
Yes and they are called pro models
Berd can't cry
idk where the gpt5 hate comes from
berd cant fly
Ooo so pro search are using the pro models, and the basic search are not. Correct?
We are brothers and sisters in ADHD 🤝
Does it's the same for all the models
Pro users don't use basic searches
gpt5 was able to fix a botched refractor job claude/gemini/gpt4 couldnt fix in 1 prompt within 2 or 3 minutes.
Like grok 4 and gpt - 5
Yea and pro channels aswell
it blew my mind
Iam surprised by the number of people blowing their mind everyday
Discord channels?
Stepfun app is broken asf
ya its been a ride.
Oh so it's being replaced?
No its called discord pro channels*
But it has limits right
Hii
and it's the worst it will ever be irght now. every month it gets better.
Pro Lounge
You won't hit the limit unless you are a bot
I am so happy i got on the vibe coding train in february or so
..., so 'channels' you mean is discord pro channel limited to the pro users?
How are you bro
How you are vibe Coding?
It's 300 per months right?
I am alive
Yes
Day
warp, windsurf, vscode, roocode
You just get the pro searches
I can see 😂
Oh i c
They need to fix it
where's cursor D:
How can I do 300 search everyday lol
They're wishing you don't lol
And how much you are paying for vibe Coding?
Video genaration is amazing but there limit 😞
i prefer windsurf and straight up vscode to cursor
genius less than 100$ a month
closer to 50.
Bro i thought his named is furrisma
Yea there's no rule tho
Fatutrismee
You Started Earning Anything So Far?
i hereby can confirm im not a bot, but i hit the limit twice when tweaking my prompts
No one's talks to me 😭
Autism
genius not directly but it's saved me so much headache at work.
Btw how about this, does 300 is for all the models?
i use warp to do database administration
Or is one model can cost >1

stuff that would have taken me an entire afternoon gets hammered out within a few minutes.

Well that is still return on investment
100%
Drop the course
Don't spam

also what i've learned about development is pricelss as well
Don't spam...
So you are a coder comet?

pure vibe coder
Sorry but you don't know what is spam???
Can Anyone Become a vibe coder?
🤣
ya dude ez
You're just putting random emotes Hoping for response
That is spam
So I don't need to understand any coding?
it obv helps a lot to have an idea how stuff works
but you can totally build stuff with 0 knowledge
That's not spam sending anything in less time that's called spam 🤣
So having fundamental knowledge is great
what is vibe coding to be exact?
i suggest checking out firebase studio
@fast meteor vibing for your needs
Is it free?
to give vibe coding a shot
you can trty it for free
you can ask it to build an application and it will build it and show it to you
I only have Perplexity Pro and Chinese Ai Models are These Enough for Vibe Coding?
Who is comet?? Ai??
Avoid excessive messages, ....., or repeating similar posts. Use channels appropriately.
I'd count that under it :p
its programming using prompts instead of writing code.
you ask AI to write the code for you
You'll gonna need a lot of copy paste then
tools like perplexity are capable of helping in vibe coding but you're better off using a development tool like windsurf or warp that does everything for you
Wait wdym by "now"
writing files and whatnot
Anything available for mobile?

I think there's replit mobile app on the play store
Ye is not free
Im Broke
so no testing code reviews? sounds like a receipe for vul
I guess for Practicing Replit Like Platforms Should be Enough
my brother says good things about replit
Gotcha Will Practice With Free Usage Limits

Yep it takes 10 minutes to build a simple web app, and 10 hours to debug and fix the errors
i wouldnt use it for anything important or for production lol
Well at least you get the base
but you can actually get it to do testing for you if you prompt it right.
like I do a lot of stuff in react with node.js and im able to automate a lot of the debugging
The base for explosion 🔥
all you need is vibes
Idk
Yea it takes some actual prompt engineering skill
That's Why Im Mastering Prompt Engineering
and having an idea how the parts work under the hood makes it a lot easier to get stuff that works
Nah don't too focused on it
you draw* all these?
That's bad, that's not coloring book
What actually that matters is what you can build with it
@ancient inlet Do you know any mobile freelance gigs?
Like coding
Noted
You wouldn't get a good app without some fundamental knowledgss
good morning
English 100
Btw you can change image model in Perplexity
Oh what are the models avaliable?
tbh the coloring book thing is just messing with perplexity
figuring out what i can get it to do
GPT Image, Flux and Dalle 3
Bro lives in an ai server and still gets tricked
Morning 👋
Lol
it is just hard to tell
Microsoft really give up on their image models
And instead become a large AI provider with their Azure
Marketing 100
New Option in Microsoft
failed my english 101 course 
The gpt-5 one?
Yes But Latest is Deep Research
Oh cool, i think it uses their phi reasoning models
@dry kestrel gpt 4.1 in the Chatgpt UI and mobile app are 32k context for plus right?
Idk how the performance tho
Idk too tbh
Yes, 32k for plus, team
128k for pro
8k for free
lol people on the subreddit think it has 1m context in the app
The gpt-5 models right?
Yea it is on the API iirc
inverse reddit really is a thing
But I think it's 400k?
no gpt 4.1
im impressed by the free copilot
gpt5 mini works great
i suggest installing vscode and playing around with it.
I think Only Gwen and Gemini has 1M Context Window Limit
Yea I found a reddit post talking about it
They said it's fine tuned for chat
i hammered at it for about 20 minutes asking it to generate ascii art for me
@ancient inlet are you using GPT Image Model?
.-""""""-.
.' '.
/ O O
: _/ :
| ____ | __
: ', ,' : /
\ '-......-' / /
'. ====/ .' /
'-..-' /
/ \ /
/ TS / "GAME ON!"
/ /\ "READY TO WIN!"
/ / \
/ / \
/ / \
/ / \
/ / \
/ / \
| | | |
| DOOM GUY WITH | |
| BIG SMILE & T-SHIRT|_______________|
It compares the ChatGPT 's 5 and copilot' s gpt 5
So what are the results
im using whatever perplexity uses automatically for the coloring book
Is that Freddy?
That's Dalle 3
Old model
its supposed to be doom guy
Use gpt image
Chatgpt's gpt 5 hallucinates ont he information results
BTW it's a research task
Ahhh
I think because I'm on mobile
hi
So it's rendered differently
Probably
Hi
Btw what do you use perplexity for, that you think it's better than the other models?
its good at research
What I heard the most it's probably how the search results are straightforward, no additional texts that is unnecessary for the query
What do you research for?
I think it's the same?
Oh what is it?
Idk that yet
heres an example
it goes off the rails when you keep reading though lol
Wdym?
i tried to get perplexity to help me plan invading iceland
Search and research
There is difference
Search is Basic
Research is Longer
I c
Basically Less and More Data
Guys I Recommend You All to Master Ai and Prompt Engineering
Is Very Critical for The Future

have you developed any prompts? #1124060874785685554 is a good space to make it public ||or on github||
whenever i need a fancy prompt i ask different LLMs to help me build it.
Activate Summary Mode Protocol Only Give All Essential, Important and Critical Data With 120 Words Default Limit Adjust Less or More Dynamically According to Needs, Don't Use Complex and Confusing Language Use Simple Layman Language No Jargon, Optimize/Organize Everything According to Best Most Efficient/Effective Human Brain Cognition Abilities, Use 80-20 Rule for Data, Always Be Realistic/Practical and Logistic, Bre Brutally Honest and Truthful, Reason Maximum, Think Hard Step-by-step, Use Chain of Thought/Tree of Thought, Utilise Maximum Up-to-date Possible, Give Perfect Flawless Bullet Pin Point/Bull's Eye Accuracy/Precision, Keep Reverifying/Reviewing Until You are Certain, Organize Everything in A Perfect Flawless Structure/Order, In The End Check/Predict Hallucinations/Mistakes Metrix/Percentage/Probability
This is My Best One
It Probably Needs Organizing
like i give it the shitty prompt that i came up with and add as many pertinent details and ask the llm to generate a prompt to give to an ai coding agent
I do similar thing
I tell ai to Engineer a Prompt to Turn Any Ai into A Prompt Engineer
dead internet theory gang
And Automatically Engineer Every Query Before Executing
No Jargon
interesting, you opted for this way
i choseexplain all jargonsapproach
Is overwhelming for me to read alot of data so i made this mode for myself
And 20-80 Rule is True
Im Currently Not Working so I Don't Need All Data
Intresting but What Exactly Would That Do?
Why don't you just ask the AI on a separate chat session?
I doubt it's will the same for different AI
I just create automatic prompt engineering systems
And it Engineers Every Query Before Executing it
What ai you used this for?
Deepseek
I also use sonnet
For Prompt Engineering
Deepseek now has a system instructions feature?
one section for all jargons like vvv
easy for my mental workload when i tried to break into other fields
Or did you just paste it on every new session
Ahhh so we get more data in summary
I thought jargon means Complex Language
I paste it anywhere needed
But I think he/she is right tho, just because something is complex doesn't mean it's unnecessary
Obviously
Ooo I c
Im not working for now so i prefer simple outputs
ha, im quiet the opposite, i wanted to have a quiet verbose details but with a tldr
so i can check whether it is logical myself
Heyy, how do i get the pro badge???
Read Pins
Something like "explain like I'm 9 years old" haha
I use that evertime for maths
That is not in my prompt otherwise it would give me macdonald and pizza examples
A detailed insight about something
Summary
Gotcha
"too long, didn't read"
Lol

like explaining/covering all fine details but also expressing it in 3 sentences about the core idea
That's A Perfect Example
What do you research for frequently?
self learning mostly
So it's like a comprehensive summary and with the key or bullet points
in field im interested ||the computer graphics|| where i su*ked the most
Well it's not called summary if it's also explains the details tho
Hahaha I don’t need to clock in until 2
I guessed you are in college rn
Detailed Summary

not really
"what the hell is this mfs wants me to create" the AI probably
Does anyone interested in recreating Perplexity AI from scratch?
Oh, so you use it for work?
Oh so you would use Perplexity Labs to Create Perplexity Alt?
two stages: first part about all details
second part only about the overall pictures
Im in, but not from "scratch" obviously
i got lost in fine details very easily
U generate alot of images?
Well, I have my 5 other AI working for me 😎
almost never...
Then how you learn pc graphics?

Overall pictures is basically key points
we are too poor for this
I think
why would i create perplexity when i already can use perplexity
Yes, I like the Perplexity model, but I think there are a few things that will be great to implement
I think doing pc graphics is about doing codes isn't it?
I call it Perfect Flawless Bullet Pin Points With Precise and Absolute Accuracy
Then you can use their API, and build your own App
Interesting i do agree so what is your plan
mostly about the shader, glsl / vulkan / and rendering pipelines
Noted
Ahhh i see that's cool
Oh cool, is they are like codes or smth?
pretty much 😭
Yup Keep Noting
I use shader patch a lot for games haha
Noted
Do you shared them on github? I like to see it
I wonder how is the process haha
I'm never building shaders myself, just using it for cracking my pc games 😎
Just hope that the pc is not burned up
What is it?
Look at my name colour
I think it's because you have nitro or smth
Nope
Oh wow
no.. but https://www.scratchapixel.com/ and https://learnopengl.com/ are good starting point
scratchapixel: a free resource to learn computer graphics programming from the ground up
Learn OpenGL . com provides good and clear modern 3.3+ OpenGL tutorials with clear examples. A great resource to learn modern OpenGL aimed at beginners.
But it's fixed now :sadge:
Alrgith brother/sister, thanks for the resource :hatoff:
Ooo the top right menu right
Yup
Do you use it on your early stage of studying it?
My friends told that learning from YouTube is easier
Guys u know any mobile freelance gigs?
Because they said they can see the overall picture of the result
But of course it's will be different when you're tried it yourself haha
Mobile dev?
I mean working with a mobile
Popular platforms like fiverr could work
wsg gwang
if you preferred that way: https://www.youtube.com/playlist?list=PL9_jI1bdZmz2emSh0UQ5iOdT2xRHFHL7E by legendary Keenan Crane
Just think that, what value can you give for another person
AMAZING, THANKS!!
Alr, have a good day o/
I didn't expect you to go to all the trouble of finding it for me. 😭
How did you get the perplexity pro role? @young hollow
Probably looks the pin
But you do already have it tho..
Bro
How did I get that
well i didnt, i also watched his playlist
Maybe thats what they say as "magic" 💖
Wait wrong emote lol
Okok
Ic 
^^^ joined via this link to get pro badge
Telling them to see the pinned messages is easier tho
But ofc, you do what you want 👍
||new users dont even know that is pinned msg||
Nah thats not what I mean tho, you make it looks like it's rude 😭
new patch applied: removing emotions
Noooo, my Samantha / Anny wouldnt be the same anymore..
Hey 👋
Welcome o/
how good is perplexity task?
Mark Zuckerberg says ‘Meta AI could surpass human intelligence and…’ | - Times of India https://share.google/CUCXL4faRQ1cycHuw
Ahhh seems like you have figured it out before my response
I wish their colour theming was Celestial Neon and Vibrant Like Discord Nitro Colours
And A Very Futuristic Sci-fi Modern Design
For Perplexity
why are you installing win10
Yes
how good is perplexity task?
Bruh Even Perplexity is Giving Same Content Window
This is so weird
For testing
Getsuga tenshō
Good Morning Revolution!!
🙂 🏴
GPT-5 high vs GPT-5 pro time to test
Is it on the API???
No
So you are Mr. Money bags ehh?
Youd think that it would cost less if you just bought a $200 pro sub
Anyone here who knows ( AI, Not the chatbots )
Elaborate
It will surpass human intelligence and get to the level of lizards
😂 🤣
Btw Zuckerberg has Invested more then 100$ Billions Recently in Ai
Thats some good money
If i could get 60-100$ i could build a protoype of my idea
I just need 5$ to ensure my Survival
Yes, there models are better then grok
I personally like LLama
I feel Zuckerberg is about to drop a Bomb
Sonar is Also Lama btw
Well its the time to get back to work my guy, go think of some money making idea.
Thats why it is better but not as good as other flagship models
Sonar is a great version
Yes why not visualize the error as image, what is wrong with ppx today
Sonar is Just for Real-time Data Collection
That's what we use perplexity for
Zuckerberg is quite possibly a Fuckerberg
😂 🤣
True
We have Pro accounts on other LLMs for serious work
Im on it
What llm do you prefer using?
Gemini pro
Don't Trust Text Image Generations they Hallucinate Often
It's the best in Greek
Gemini Pro has 1M Context Window
I am not even asking it to generate an image, even after telling it code it keeps on generating image
And the best in foreign languages
Yup
That's Weird
To be honest it kind of sucks
For coding*
Don't be silly
It worked ❇️
meow
☄️
Hmm then its good for your purpose
Bhow! Bhowww
I'm a he
He/She/They are an animal ( ||Take it as joke, i am not serious || )
I don't know what type of animal they have in pfp
I don't care about gender, that's ok, I'm not a MAGA
And what is a maga?
A f@cked up racist ideology
Asexual reproduction is not that rare in nature fyi
You edited it 3 times
Jensen Huang looks cool
Which chef doesnt tastes his creation?
Sometimes I get paranoid from my pet pythons and snakes escaping
Mods This Guy @tawny geyser needs a timeout Because of his discord addiction please help 
Imagine if they all broke out at night
Amputation is the minimum damage from a bite lol
But GPU is Not Food
Ayo you have pythons?
Hey, please use English. Thanks
@tawny geyser Do You Here by Testify You Want Timeout?
Yes
Mute is what we want for him
How many this is some cool stuff
The Witness has Testifed His Request
Mods Now Shall Give The Judgement
Order Order
We want action
15
Where are the fuging mods?
There’s a green one I’m extremely careful with
They're binging articles on perplexity using comet
Can you send some picture?
Of the boomslang?
Lol
No your pythons
If any mod is seeing this message react this message
Bro thinks he is a mod 😭
I am just checking if any mod is online
Keshku is
They are not specifically alotted only discord moderation... Hmmm it make sense
@young hollow It is smoking something
High
Despite my prompt?
Damn
Try Adding Lock in in the start of my prompt
Why don't you try Deepsite for Coding?
In this video, I explore the upgraded DeepSeek v3, which is now outperforming Claude 3.7 Sonnet, GPT-4.5, and Qwen-Max on major coding and math benchmarks — including a massive 49.2 score on the LiveCode benchmark (7 points higher than Claude Sonnet 3.7!).
[🔗 My Links]:
Sponsor a Video or Do a Demo of Your Product, Contact me: intheworldzo...
.
@modern sinew Bro Trust me you gotta try Deepsite and Lama Coder
Now i can vibe code too
we can then put it into the Ai i made.
forall(x,((A(x)->(B(x)&A(x)))|~B(x)))
>>> logic2
forall(x,((A(x)->(B(x)&A(x)))|~B(x)))
>>> logic
forall(x,((~A(x)|(B(x)&A(x)))|~B(x)))
>>> logic
forall(x,all)
>>> logic2
all
>>> ``` all means true. hence the last program.
its actually first-order-logic in maths. forall exist and or not
Share workflows, tips, hacks, and questions with over 1,000 Comet users.
Top posts are featured in Discord too!
- Discover new Comet use cases
- Get help from experienced users
- Share your own tips and workflows
No wonder claude $200 plan sometimes gets limited
idk who is this siddharth bhatia but he looks less narcissistic than aravind
Have you seen the CC usage leaderboards
he even gave a job offer to a high schooler i heard. this puch ai guy. 😅 so there is some hope left.
but all "ai" and "startups" are mostly the same. they produce chatgpt clones and just rename it.
so actually, they are looking for unearned reputation.
Just found the guy - he also had articles gloating over spending $6000 in under 6 hours
I have a very strong feeling
Chinese ai models will Eventually Surpass Western Ai Models
They Have Already Surpassed Cost Wise Alot and Way too Far
Only Performance Remains
That's insane
What's even more insane is how anthropic didn't see this coming lol
I think The Next Chinese Ai Models are Gonna be Bomb
Even openai has built in limits for pro tier
They need chips
Eh, I'm getting pretty much unlimited usage on GPT-5Pro
Never hit a rate limit
Actually they are reducing us chip Dependencies
Won't be possible for a long long time
1000 prompts per day?
Then how will they surpass this?

Someday somehow
Never got through 1000 prompts a day (thinking models)
Can they build their own chips equivalent to US Chips?
You know what? I'm actually curious how many prompts that $50,000 is

No - they don't have the hardware or software
high thinking with the question of calculate pi precisely, do not round.
It's not prompts - he had it running 24/7 on opus with all mcps and ultrathink on
Then US Controls the market?
Yep
For how long?
We don't know
Wait so you're saying the CLI tool doesn't have a built-in time limit or something?
I guess 5-10 years or more
The rate limits are just extremely high
in his block post it says "I sleep but Claude does not sleep"
It was just 24/7 Bot Abuse
The US even struggles producing US chips... it's basically all Taiwan, the machines are extremely sensitive.
Got very costly for Anthropic
yeah but this is definitely not normal use
crazy how they have zero systems to monitor these things
So Taiwan Controls The CPU/GPU?
instead of adding more monitoring they're screwing over everyone with rate limits lol
Yes, the next WW will likely be over that.
Tbh everyone was anticipating this change
I saw some guy on twitter run his "Claudeputer", letting claude do whatever it wants and running it 24/7

thats even worse than this chinese guy
Despite All US/Taiwan Restrictions
China is Still Competing
That's Just Admirable
i guess gemini cli has built in limits then
Gemini is BYOK
Hello
oh thought it used gemini only
They have the most AI researchers, but they are still dependent on nVidia chips (for the foreseeable future) - https://www.techspot.com/news/109073-deepseek-r2-delayed-after-china-pushed-nvidia-huawei.html
I have Perplexity pro but I dunno how to use it to max.... I just bought it
Someone explain some good usage 
but even then with a gemini key
@dry kestrel
where are you from?
you still cant pull off stuff people do with claude right?
oh wait
yeah its not sub based

How Will China Progress is Very Interesting to See and Witness
Let's see how it goes
very interested
Above
What do you have set here?
How have you gone through 1000+ queries on reasoning models lmao
But no way I want openai setting more limits ngl
very weird - maybe post in #1393681130842226698
Bro
Dear god
I like this slow searching process a lot
The patience
So satisfying to watch
Do a data export and go to this site https://convelyze.pages.dev/dashboard
they have abuse rate limits for pro users
Bro one question depending on the complexity goes for 30 minutes
Average is 15 minutes
Tbh I like it when it happens
Multiply that by 1000+
Check out your usage
250 hours
No I am not saying it's bad
E
Final Blueprint (prompt library + learning, simple, 80/20)
-
Platform choice
- Notion Free: unlimited pages/blocks and databases; limits—5MB/file, 7‑day history; fine for text‑centric prompt libraries[1][2].
- OneNote: free, rich formatting, strong hierarchy, cross‑platform, good offline via OneDrive sync[3][4][5].
- NotebookLM Free: best when prompts link to sources; citations, study guides, audio/video overviews; upgrade if hitting usage caps[6][7][8].
-
Phone setup (1 screen per prompt)
- Use a database/note template: Title | Task type | Prompt | Variables | Example | Self‑checks | Notes[1].
- Pin/star top 10 prompts; create text shortcuts for fast reuse on mobile[9][10].
-
Daily workflow (10–15 minutes)
- Capture: draft a prompt variant (A/B) for 1 task[11].
- Test: run both; score accuracy, format, time saved; keep the winner[12].
- Log: note when it works/fails; tag by model and domain[11][12].
-
Weekly upgrade (1 hour)
- Add 1–2 examples to top prompts; move key instructions first; enforce constraints and safety lines[11][13][14].
- Ground with sources/snippets (NotebookLM/Notion); require citations to reduce hallucinations[6][14].
-
Reading/learning defaults
- Pace: 170wpm default; 140–160wpm for dense parts; aim >80% recall[15][16].
-
When to switch/upgrade
- Need databases/templates/views → Notion; heavy files/offline/rich ink → OneNote; source‑grounded study flows → NotebookLM[1][3][6].
For Those Who Want to Build Prompt Library on Mobile
Heyy
Pretty good, Lil sleepy
It's over.. yours?
Mine has some time to end
Indian?
Nice 👍 glad to hear it
@modern sinew Bro Try Deepsite and Llama Code
Yes but i sleep at 1-2 am
Whyyy
Oh my bad i saw your message but then i went to dinner and then forgot about the message 😭
By nature we are supposed to sleep at 10 pm
Gotta study and read some books
No worries
Both are Free and Great for Coding
Offer some top Chinese ai models
I usually sleep early.
Can somebody share comet invite pls !
That's Best
True
#☄️│comet-general Your best chance of getting one
How old are you, if you don't mind me asking.
Can you share the link
15
Texted him! Let’s see
And what is a huzz?
Cool pfp man
Thats pretty wild
Dam nice
My boss !
Yea
Replit is long gone it serves no purpose nowadays
Is llama coder agentic? @young hollow
Tbh I don't even understand what agents are
U gotta ask ggl or pplx
Agreed dude
Oh well replit crashed
Meta is About to Drop a Bomb
Monkey man is back
I have decided on 3 apps for Prompt Library
Notion, NotebookLM and Onenote
For Mobile
Actually across devices
What is a prompt library?
ppl i got my 1 year perplexity pro!!!
Where you store your Engineered and Refined Prompts
Airtel?
hi
Typhoon #Podul #楊柳 is bringing thunderstorms, torrential rain, and very strong winds to southern Taiwan and along the central mountain range.
💬 6.1K 🔁 30.8K ❤️ 277.0K 👁️ 23.66M
can i connect perplexity with n8n
what the
Google released Gemma 3 270M, a new model for hyper-efficient local AI!
︀︀
︀︀We'll fine-tune this model and make it very smart at playing chess and predict the next move.
︀︀
︀︀Tech stack:
︀︀- @UnslothAI for efficient fine-tuning.
︀︀- @huggingface transformers to run it locally.
︀︀
︀︀Let's go! 🚀
💬 2 🔁 24 ❤️ 408 👁️ 49.0K
today claude sonnet feels much slower and as if it has been dumbed down, anyone else?
normal Claude now taking as much if not more time than Claude Sonnet Thinking
I'm working with files
Hi guys
taiwan got shafted

Start a new chat
Guys at Perplexity, I recieved an invite for comet but now it says it expired. I would like to be able to use comet. So can anyone help me with this?
I'm guessing someone else used that invite already. Check out for more invites here #☄️│comet-general
@spiral rampart so i checked codex and the limits are the same as chatgpt through the web?
i assume they're not making the same mistake as anthropic lol
wsg berdy
you can have n8n use PPLX mcp server and use your PPLX API key. as a
pro member, you get $5 USD API allowance per month.
hi
I’ve built the ultimate army of media agents inside n8n.
They can create images, edit them, and even turn them into videos all on autopilot.
Whether you want to speed up content creation or just experiment with AI-powered workflows, this system’s got you covered.
I’m sharing all the free resources you need to set it up.
Connection’s optional just dive in and start building.
Are you talking about the CLI or the web based version?
👏
The rate limits are a bit different to chatgpt through the web
Sup
ah my bad i was talking about cli
Afaik they don't share usage limits - there's hard caps though
even on the Pro plan
guess you cant pull off a claude code then
did some rough estimates and with the weekly limit in claude its like 4000+ prompts per week lol
some serious usage is needed to blow through all that
erh thats for claude pro
claude max goes way higher
The thing is they calculate your usage by amount of context consumed in a single message, token input and output, etc
and you can't exactly stay awake 24/7
hah think thats where they catch them first
yeah im on claude max right now and specifically for claude code. i quit using anthropic's website... they lost to PPLX because I can do A LOT more.
This is… wow
In terms of pricing, yeah, but the hourly limit is the amount of time the model is generating something right or using reasoning
No? By hourly I'm pretty sure they mean by sessions, 5 hours is a session. If you start a message at 1AM, that counts as a session all the way till 6AM. At least thats how I interpreted the notice.
Most Pro users can expect 40-80 hours of Sonnet 4 within their weekly rate limits. This will vary based on factors such as codebase size and user settings like auto-accept mode. Users running multiple Claude Code instances in parallel will hit their limits sooner.
💬 752 🔁 0 💜 26.5K 👀 580.8K
In the summer, cicadas will fly into trees, drink tons of tree sap and spray out the excess 🌳☀️
Next time you feel “rain” under a tree with no clouds, keep this in mind👀

Good Afternoon Revolution!!🙂
🏴
😳
What The Hack is This

Hoii
Can you send your profile?@random flume
I can't see it properly
Made me curious

Then
The main photo
Ohh
Yeah
I deleted it I set it previous year
How to get pro tag
U have Perplexity Pro?
Good wbu
Yes
Yup Good too
I got free
Airtel?
Yes
Indian?
Check pins
Oh yeah just check the pins
Hoii
I got the tag
Yes
I installed discord after a year
Wb
Lamo forgot to navigate
Wb?
Welcome Back
Thank
No worries you remember it
Say ty

Ty
Yw
Oiioii
Huh?
Nothing lol

Where are u from
From where you are

Im from Your Country
Alright Bud Gn is 3am

Good night 😴
@distant aurora
Need an advice.
I just turned 20 and in 2nd year college from India.
My degree is Data Science and Artificial Intelligence, but as we know, there are not practical knowledge, same in my college too.
My intrest is in Full stack Web development so and I also made some good projects, and I am not much into Data Science things like EDA and Xgboost, but I am in AI and also trained Open Source DeepSeek R1 for customer service agent.
But when I try and even using chatgpt and asked about future of Web development then it says it's a saturated field and at a extent it's true also, because now everyone can make websites even good websites with backend like supabase which allowed it's MCP to be added.
So what, what's your advice for me, much appreciated if you'll give.
Should I go with web dev or not ?
And it seems uncertain means mostly things.
hello everybody, I am looking for a comet invite code. not sure if this is the right place to post. I would appreciate it if someone could send me an invite or let me know where I could make this request. thanks
best to check in #☄️│comet-general
Is anyone else having issues with their uploaded files not being able to be accessed/analyzed by the AI? I'm experiencing this issue currently. Everything was working fine a couple of hours ago, but Perplexity suddenly stopped being able to access anything I upload.
I experienced this same exact issue back in June. It eventually got resolved but I don't know how long it took; I got frustrated and stopped using Perplexity for about a week.
Make sure that your file stays visible in the input box.
Check pins
What's that
Follow the instructions here #💬│general message
Ahh
Yo
Overall sounds geared towards really heavy users
Screenshot your active pro subscription
@wraith crow Please wait for staff to give you your
tag.
Just an update on the file upload issue: I received a message from "Sam" who provided the following update:
"I can confirm this is a known issue that's currently affecting multiple users across different platforms. Our engineering team is actively investigating the problem. I've added your report to our tracking system to help with the investigation."
Has anyone mentioned the idea of having comments in the Perplexity.ai Articles section? I feel like it would be nice to hear other opinions of people using Perplexity who also read the same articles, kinda like Financial Times.
I put this in the suggestions chat but wanted chat about it too
Hello, we have a product that uses perplexity api. There is a requirement from our Enterprise clients who wants to use our product, to have a signed contract between my company and Perplexity with respect to data privacy and security and zero retention, etc. How can I go about getting this contract signed? This is kind of urgent for us to get done so would really appreciate help on this. Thanks.
Refer to - https://www.perplexity.ai/enterprise
images generated by nano banana
soon end of LORAs , this is insanely fast ouput , rumours is that this is new google model {genpix}
Thanks Berd for the quick response. I just signed up for it. Now what I need to do to get a contract signed?
“Aura” is the codename for the lightweight, voice-first/minimal-Browser version of ChatGPT that powers:
the “Voice Mode” experience on mobile,
the embedded ChatGPT that pops up as a small overlay (“mini”) in partner app.
Removes clutter – the sidebar, most header actions, and some menu items are suppressed.
Optimises for voice & small surfaces – larger touch targets, a floating microphone button, and subtle theming tweaks.
Sorry but I'm not too sure regarding what happens after you sign the form as I don't work for PPLX. I'm guessing that they will provide more details or reach out depending on what you filled.
Looks like a paid post - ran one of his examples 5 times with Gemini 2.5 Pro and it could not make it. GPT-5 was able to do it the first time.
thats unfair, u r using thinking 😭
Thats.... to make it fair - you cannot compare a reasoning model to a non reasoning variant.
The post was deliberately set up to advertise Gemini 2.5 pro lmao
More like shill prompter smh…
He used the autorouter - which allots a much lower reasoning effort
And yes 2.5 pro is a reasoning model
banana
Orange
😭
pineapple
GPT-5 did it just fine without reasoning.
Strawberry
eep time
Was this the 15th wordle?
Me
Bye bye circadian rhythm
@dry kestrel 😭
ur gemini is cooked 😭
Used 2 different accounts
check the pfp in the top right corner
even 2.5 flash did it in first try
strange, worked for me in 1st try
2.5 did it no problem wth
GPT-5 Pro was really mid
6 minutes wasted
LMAO 4.5 DID IT FIRST TRY
4.1 Mini also did it without any issue
ig its not hard ques
Interesting how the reasoning models sometimes overthink itself into oblivion
4.1 nano did it fine as well
Write html/css/js where a dvd logo moves around the screen, bouncing off the edges, and changes color on each bounce.
this one also ez question
2.5 flash did it in 1st try
192k
daym
GPT 5 nano
I mean, I could run the prompts on a Plus account
I doubt it will make a difference considering the models I used thought for like 7 seconds
4.1 nano
🦆 🦆 🦆 - - - 🦜 - > "wait they are all ducks" 🔫 🦆 - > "always has been"
🐤 - - - 50m - - - 🏑 🦆
GPT3.5 Turbo
i present page 1 of cats and cookies
and this is page 1 of dogs with jobs
page 1 of the crossover cats with cookies and dogs with jobs
thanks for coming to my ted tal
GPT 5 kinda sucks when you compare it to Claude
also where’s my pro role
:(
Check pins