#💬│general
1 messages · Page 347 of 1
So do i
im PRO
Im slow
lol
no does it have badge
ohh
maybe no
how get community veterans bage
Yo
@wraith crow how get veterans bage?
yep
i never had the pro perks
where
what were in the pro perks
discounts for random stuf
ohh
they had cursor pro free for a yr at one point i think but they got rid of it
oh, idk

or like 3 months
one of them
i didnt rlly care too much for it i share a cursor max acc alr and i main vscode 😭
By being active and helpful in the chat. It is not given out lightly since Veterans also have some perms to moderate the chat.
i should saved it tho jus to use it for fun idk lol
Guys how do I get pro role, I have pro
is there a channel where i can ask questions to a perplexity employee
oh, i think it is the "old" tag
I did
If you have a question for the support team you can ask directly via the intercom on the perplexity website.
Mr. Mod, can I have the pro role
anyone know fellou is good or comet.
which to use
Please check the pinned message on how to obtain the role. Let me know if it still doesn't work after that.
Is there any channel through which I can complaint to perplexity employees?
gpt-5 codex edition?
what im interested is
when will openai add their own nano banana lol
the watermark is annoying for advertisement usage etc..
Ty
Ig u have to just ping them
Is there no way to request an album or playlist to play via voice assistant?
I mean you can on YouTube
Grok 4 is an incredible model.
But it's best experience is on their website itself.
The Grok in Perplexity has some weird system instructions.
If Perplexity adds a feature in which they don't provide any system instructions and give users the right to give it custom instructions, then Grok / I think so all models, would perform a lot better than they do now.
Grok 4 generates soo many tokens that it ends up being way more expensive than the task needed
Plus it doesn't show the thinking data lmao
I never wanted to say but,
T3Chat >> Perplexity
Hello
hello
Guys a doubt.... Can anyone give their opinions?
what is it
Yes
@pine burrow I've been wanting to try out t3chat. What is so good about it? Also what about rate limits?
if anyone want it
whats this
How much would live dynamic website with ticket booking, admin dashboard, staff dashboard with all user authentication cost?
Ask Perplexity!
Hmm 🤔... I asked and I saw like kinda huge numbers for me
@wary geyser whats sthat link
a comet browser invitation
You Really Need to Ask A Expert in That Field Tbh
Hmm...
BRUH
Drop The Needle
They offer various models + better UI & UX @ 8$.
The current rate limit is based on messages sent, for premium users it is around 1.5k messages monthly + 100 premium messages per month.
You can check it on their website and then decide.
Well I tried to claim but, it already has been claimed@wary geyser
someone claimed it really fast
Yup gotta be Fast
There are Comet Invitation Hunters All Around The Server
i calimed it 🙂
Because is so Hyped
lol
thank you 🙂
Well thanks for info and time
Some of the models which are available, the ones which are in white are free (20 messages daily)
I don't think web/app development is expensive now because of Ai
There is Also You.com
But Their Search Features are Bad
The Models are Good
it says joined waiting list
Oh then i think somebody else actually has already claimed it
T3Chat is best, ngl.
Hmm that's what I have been thinking, but still I want to know what's the price in market at present
send another one
Couple of $
Not much tbh
It depends on what quality you need ans who you work with or hire
Hmm it's not for hiring actually.. it's for charging
Im confused whether I was less paid or was it good
Aah gotcha. Thanks
How do I click the sign in link
Just check what others are charging who are at similar or same level as you
Hmm thanks buddy!
Just click on the Sign in Text i assume
Yw you are a freelancer or something?
I tried that
I think around $10k to $20k.
This range is fine.
Hmm yep
cool
Then try to sign in from settings
Wait u mean dollars????
You are a coder or vibe coder or use both?
Both
If what you made is absolutely good and works fine seamlessly without any bugs, then I think yea, that much is pretty much people get paid.
You mean monthly right?
For once, lol.
That would be for Professional Enterprise Projects not General Ones
Hmm 🤔 there are no bugs and I have implemented good security too for it... And did some good design
What level of project it was?
If it is minimal, ~ 5$ is fine.
It depends on the scale of the project
Yea
Hmm it wasn't much big I guess. Had 3 level of users admin, staff, user. It was for an event ticket booking. So had payment integration and content showcasing about event and things. With admin dashboard
Beginner, Intermediate, Expert, Professional, Business and Enterprise
That's it I guess?
Seems like a intermediate project to me i assume
Oh, event ticket booking.
$5k is good deal.
Yep
But i think that is pricing for US Freelancers not Indians
I thought it is like some complex like may be flight booking or something, that is a bit complex.
Wdym, everyone should be paid same for the same work.
Hmm nope not that much complex
If Indians Would Charge Same as US Freelancers They Would Just Prefer US Freelancers Of Course
Since they don't get any Discount
It didn’t link.
Freelancing Pricing is Country Wise Not Global Wise Not All Country Freelancers Get Paid Same
Hmm 🧐 then how much would it be ideal at India?
Maybe search this problem on YouTube or Google you will find something
It depends on what the Clients Think and Agree to
Exactly
According to fiverr, then $100 is more than enough.
South Asia Countries Don't Get Paid Alot for Freelancing
Hmm 🤔 I see
The Rich Countries Treat South Asia Countries Freelancers Like Digital Labours
How much did you get?
Give it in a range.
Not exact
Just curious.
As Long you get more then your local jobs is worth it anyways
That is so unfair, tbh.
OSS 🫡
Hmm it would be around same $100
True but that's how the Freelancing Industry Works for Now if The Rich Want to Pay High they Would Just Hire Their Country Freelancers
Yea, that seems fine. How did you get the client? Fiverr?
Pretty Decent Ngl
And while today I was searching around browser for other things I found the work I done was pretty underpaid....
But I couldn't believe whether it was real in market price or not
In India The Average Salaries for Local Job Developers is Like 20-30k
That's why I have been looking around whether should I increase prices for my service or not
That is why most of them want to go to US or UK.
Even With Low Wages as Long You Get Clients International Freelancing is Worth it
Like, the thing is those IITians get into news that they got like 100k $ package in USA.
If You Increase Im Pretty Sure there are Other Indian Developers Who Would Offer it for Lower In fact there probably should many offering for lower
I mean they gonna live pretty bad life with that salary in USA.
That is bare minimum to live in US.
Cost of Living Matters The Most
100k $ in India is awesome, but the same in USA is pretty bad.
🤔 hmm considering this all fact what is the thing?
Is it worth for that price or undervalued?
Its fine for start. I mean do more projects, build some portfolio on you, and then give the clients a reason for why are you charging the amount.
That's Why Even with Low Wages Freelancing is Viable in India
Because The Cost of Living is Low
Hmm so now
Just Check Fiverr, Upwork, Freelancer.com and You Know it
These are Standard Freelancing Platforms
v0 is actually pretty good at work.
V0?
Sorry idk about it
How much would an e-commerce app for Android and iOS be?
It is by vercel.
Is it Free?
Like bolt, lovable.
Vahhh bhai vahh
Ohhh
5$ limit per month, Ig.
Hello
Like an app for some store to display all their products and display whether in stock or not and booking them for online and offline
Hopefully
It can do the back-end, integrating with supabase or whatever you like.
Overall better than replit and all.
realtime probably means voice stuff :/
do you remember me now?
thats good
cmon man
i was in arc browser dc :(
free subscription is insane
but then they made dia and i left ah
Noted Noted
I want to start learning vibe Coding
Even if is 5$
@young hollow @ember bramble
hi guys
I have dementia
goldfish memory
Codex is out of the way sooo :P
No harassment, hate speech, or malicious behavior. Keep things positive and constructive.
Bro Im Not A Coder or Freelancer as i Said Just Check it on Fiver and such Platforms
Codex has been amazing
OH! okiee thanks for the info!
I actually don't have much idea. I would suggest to check on freelancing websites for others how much they charge and then you decide.
^^ Yea
Hehe
Is v0.dev Best or There are Better Alternatives? @ember bramble
what would be really interesting is a gpt-5 browser like comet
Hope the stuff they announce today are cool too 🙏
Yeah those are Pretty Much Standard Freelancing Pricings Globally
I mean, they made a browser landing page alreaedy
Yes
I think they are already making such a browser i assume
I mean it is one of the best, you can test it, they give 5$ for free monthly.
There are some more, but they don't allow free plans, they have paid plans.
already?
lol then its not far off
I see no other free tools?
Lovable probably
OV, Lovable, Bolt, Replit
I didn't get anything in manus for free when signed up
chef, emergent, flowith, minimax, kiro, jules
Minimax but that is paid
They give free credits everyday
At this point just mention PPLX labs and ChatGPT canvas as well
U sure about that?
Manus is very slow..
no
cursor, windsurf, cline, haisnap, comate, trae
I have it now when i had just signed up i had got nothing
Man Anything Free Works for Practice and Exploration
Just checked again, it was removed :/ https://chatgpt.com/new-tab-page
dammit lol
well edge showed how it could look like
its not far from comet
i assume they're going for the more apple aesthetic tho
hi guys!

hi zro
At this point might as well mention Most Ai Chat Models Since Most of Them Can Code
Pen and paper next
Just got the pro version of PPLX but it is dumb 😭
If you are a bad prompter almost all Ai are Dumb
Hi guys
Ai is Just A Reflection of Your Input According to it You Will Get The Output
I swear man alot of These Ai Content Won't Make Sense
I Swear there is so Much New Ai Stuff Everyday i can't even Keep Track of it...
I was already Overloaded by YouTube and Other Content
Now all this Too
In past there were softwares everywhere
Now there will be Ai Tools and Software Everywhere
For Almost Anything I Think

bro
why is perplexity so bad at this point
i swear it used to be better
we're being given less tokens ig
especially because of those free airtel users
the servers are overloaded
:/
ai gf when
skill issue
fr
Lol
Anyone watching the OpenAI livestream?
the openai engineers are so awkward talking on cam lol
Soon lol
That means they are the real devs and not some fake PR devs lol
hot
No NSFW content, inappropriate usernames, or profile pics. These will result in removal or a ban.
kold
yeah
idek wtf i watched
Tmobile people talk way more easily lol
drink it
Take it out of the lake woulda be a starter 😄
these DOOM BOTS VIEGO EVIL ON LEAGUE IS HELL
fucking stupid riot overtuned em for no reason
That would mean going into the lake
lee sin is a menace fucking nerf him now
unless you have THE CLAW
wait until water evaporates
I'm making Kesku get it for me
Oh thats a good idea
I'm impatient
drink the water like its female water
Bro.. you good?
It's awesome
not yet
if it was cheap i wouldnt have complaind
I've been using the extension in Cursor + claude code + codex CLI
Im so happy they released it
You gotta try it out
Anyways we have like 7 versions of Codex now
codex nano when?
lol
never
I mean, they are switching over to GPT-5 in Codex
2023 cutoff?
bruh
I Swear i have never really seen such high and fast level of cut throat competition before
Everyday there is Like some New Ai Code Stuff

so it cant pronounce "skibidi"
...
$64 
cheap
Can't choose the model on the playground
Yoo
eh will it even be available in advanced voice mode?
They didn't say
I hope it will be
hmm
the wording makes it seem that advanced voice uses it
since it uses the same realtime api
gpt-realtime ≠ av
That's how I interpreted it at least
(Not now, and still no on September 28. Did you make another choice? 😅)
They will have to pay me to turn on model training
We agree — they could at least offer something similar to the sharing feature in OpenAI’s Data Controls (You’re enrolled for complimentary daily tokens.)
Wow, so many people use perplexity
Yeah, they have something similar to OAI's complimentary token program, although its not as good.
yes, only on CC
Of Course Perplexity is A Great Ai Assistant with Main Famous Ai Models in A Unified Interface
It’s there, check again.
I had similar issues. Not on such a large scale though.
Did chatgpt get it right?
Yeah
Before deciding to possibly tell me that it's a typo. It actually does a web search to check by itself
Grok just assumed it was my mistake lol
Hi there guys
can anyone help me in my Open Web UI set up?
its not connecting to ollama
even my own web UI connects to Ollama fine
what the hell is going on with opne web ui?
Nice, finally lol
You might want to ask this question in their discord server - https://docs.openwebui.com/troubleshooting/
I already did]
whats up yall?
Bro that's like the other planet orbiting around the sun 😭
Also i thought perplexity would have the Indian flag
And i don't think this graph contain China at all
Why? It's a US company
I genuinely didn't know that
...
And i think it would be very normal to assume its an Indian company. It's not like India is behind in technology
PPLX's HQ is in SF
👆
It's not just Aravind running the business
smh
It's not all about him
The user base have also a lot of Indian individuals and naturally i assumed it's an Indian country
Ah
I'd be very interested in seeing their backend user stats
Close but not quite
Iraq : 1 individual
Lmao
hi
I mean if you use a VPN it wouldn't even count
HI
Iam surprised by Korean and Japan but then I guess it makes sense because r1 1776 was great at translating south Asian languages and it was the non ccp party version
guys is perplexity just bad at image generation or am i doing something wrong?
Choose GPT-Image 1 under settings
perplexity has an image generator?
damn ok thanks
Is this before or after airtel promotion
It's right now
i dont find
Probably set to Flux or default
oh ok
thanks
the stats differ by website quite a bit
Is this Perplexity User base by each Country statistic??
It's just people visiting the site
Ohh
Its free
i know its free is there any difference in normal perplexity pro and the airtel one and is it not a lose for the company perplexity
Probably including crawlers
Hello Friends ! xd
"Presearch: Redefining Search in the Web3 Era - The Decentralized Alternative Challenging Google and Perplexity AI"
I think you really need to get out of LOL Mindset and Go in Nature
Don't be so obsessed with a Game i Do Understand Your Enthusiasm but We are Not eSports Players Yet
bro i didnt do a thing im not even toxic on league i cant use their supp website the shit says page failed to load and ive cleared cookies and everything. the website is still having this issue and everyone mentoned this too
i play leaguefor like 3 hrs or so then chill on yt or whatever
i swear its that doom bots match i lost over 10 times u cant win it ive tried lol towers are like 10k hp
Ahhh my assumption is wrong my humble apologies
the lux bot and corky and lee sins are on steroirds their ult is batshit
game gonna make me quit if it keep happening
That's FVCKED UP
Stay on-topic and helpful in each channel. Use #💗│sharing to post tips, content, or cool discoveries.
theyre "busted as heck" riot says
Maybe Try Some Other Games that's More Fun
ye i been on gmod and dead island
Yeah just Switch it up
And Have Fun
The Gaming Universe is Like Infinite
Countless Games Available There
ye
trying to get out of low elo ik its alot new members but i be losing alot lol
our shaco jg didnt do a thing once and i was fighting ksante felt like some marvel movie and later on he won the lane because his jg got most of the dragons
okay...
很好

@green crystal this shit lacks initiative. https://chatgpt.com/share/68b0c489-d494-8013-bbf1-d8ff7a9222f7

You didn't enable it?
anyone from US here?
yes
You can't enable thinking and web search simultaneously manually.
Usually enabling thinking does the search too.
Even if I ignore that, what bothers me is that it still didn't search despite me pointing it out and the complexity of the query. Ended up trying to defend not searching.
Just tell the model to use the websearch tool?
What's new? Did Chatgpt fix his system prompts?
Does they give free or gift promocodes?
There's $1 codes that Theo shares in his videos
You only get 100 messages a month with premium models though
I don't understand why I need a t3chat if I have perplexity with GPT5 Thinking, Grok4, etc and Gemini separately
Not impressed, even with unlocked models
Is anyone interested in setting up their own server with Open Source llms?
@green crystal grok can browse now?
This answer is way better than what I got. https://sharetext.io/6ea485de
you don't
also sonnet 4.1 isn't a thing
I did. It still asked for confirmation.
Chromium being funny, also Comet comes with Adblock
There's this shield icon thing
Just click on the URL bar and it'll appear
is perplexity down?
Ah free tier
Try to turn web on and adding to think carefully
Yeah not for everything tho
My instructions probably helped lol
Basically makes it a bit like o3's personality
The last one killed me lmao
Hello guys. I have 5 invites of comet. Does anyone want to try Comet?
Is the ceo on this discord?
hi
how did taylor even find this lol this is sampled in big poe on his new album and its catchy
i jsut put roked roked and this popped
indian music go hard fr as an amercian who doesnt understand a single word i imagine theyre spittin bars
Why it freezed? Sonnet 4.O thinking
Again and again
Hello is there a perplexity API option for GPT5 thinking
What model is sonar pro
PPLX's in house model.
No it’s not
This is false and misleading.
They have a modified version of R1 thats it
You can learn more about the in house models that PPLX offers here - https://docs.perplexity.ai/getting-started/models
They’re sonnet, R1 etc underneath
sonnet? sonnet is closed source afaik
No they are not.
Read the website I sent you
How can I automate using perplexity GPT 5 thinking with a preset prompt for example
I have, ive implemented sonar pro in my bot
But i want gpt5 thinking with perplexity search
Perplexity's in-house model that is optimized for answer quality and user experience. Built on top of Llama 3.3 70B, Sonar has been further trained to enhance answer factuality and readability for Perplexity’s default search mode.
I said it was their in house model. You said it was sonnet.
Sonar isnt that good. Sonar pro is better, that one is sonnet or something
It is most def not sonnet
Sonar pro reasoning is their fine tuned R1 just like sonar is fine tuned LLAma
How do I make a preset prompt i want to quickly run every time with the app since thats the only way to use their gpt 5 thinking
Is there automations or something
Model
Used o3 because Claude Sonnet 4.0 Thinking was inapplicable or unavailable.
Why Sonnet 4 thinking is broken?
Hi guys. Any idea how to get the pro role?
In house model
Lol this doesnt tell you
Sonar is llama, sonar reasoning is R1
Sonar pro is believed to be sonnet
Fine tunes or whatever you want, but those are the base models
I can tell you for sure that Sonar Pro is not sonnet
Premier reasoning offering powered by DeepSeek R1.
Sonar reasoning pro
Sonar is llama
Sonar pro is something that is NOT theirs likewise
Lol
Ik the pricing is similar
Correct its the same pricing
Grok 4 is the same pricing
One is llama, one is R1, one is sonnet
Sonar pro precedes grok 4
I'm just saying you cannot be claiming its sonnet 4 just because the pricing is the same
Yea i can
I just did
And so have other people
This is baseless, false, and misleading.
You’re wrong
"same pricing" does not mean same model
that is not the only reason it is sonnet…
That was the only reason you gave
So what, there’s other reasons people believe it is sonnet. It is 1000% NOT their own model like you stated at the very least
We’ve established “their” models are just fine tunes of other models
The other reasons being?
He might just be rage baiting ngl
I'm not seeing other people confirm this lol
👍
Google it
I don’t owe you any answers lol. You claimed its their model and that’s a blatant lie they dont have models
hi
sonnet is closed source, so perplexity cant build a model on top of it, because they dont have access to it
Correct, so they are just using it + their search = sonar-pro API
The others are open source so they did a little fine tune
Benchmarks don't support your claim, if Sonar Pro was just Sonnet, it would get similar results.
that's an assumption, right?
i mean, no one knows which model is behind sonar pro
If you look at the benchmarks, Sonar Pro severely underperforms any sonnet model.
Especially in coding
Which are?
S there a deepseek cli tool like Claude code and Gemini cli?
Darn
You can use a deepseek API key with claude code though
3.5, and haiku
You can use diff models that aren’t from Anthropic?
pricing won't match.
Yep
try this (randomly found it lol) https://github.com/holasoymalva/deepseek-cli
Thank you
you can also plug deepseek to claude code via claude code router
Here's with 3.5 added.
but thats a bit more complex
No need
- 5 sonnet is same pricing as sonnet 4 lol
Deepseek has their own documentation now
Was referring to haiku
I’d like to try that tbh
oh, really?
They have instructions on how to use their API with claude code
Works great
is deepseek 3.1 good in coding?
heard it often spits random chinese stuff when coding websites
Both Haiku and Sonnet 3.5 shows no resemblance to sonar pro.
At best you could say Haiku 3.5 but then the pricing is different
Its okay
Sonnet 3.5
Oh
Typo
Damn
I've checked this a long time ago
Is their api expensive?
Cheap but slow
.maybe it's a limit thing?
?
They said it was just sonnet with tool calls from PPLX
Sonnet 4 thinking is broken, stopped unexpectedly
Please make a bug report
(you are probably hitting the output limit)
Grok4 and GPT5 Thinking doing well
i personally dont recommend using perplexity for coding. They limit a lot the context window of the models
Tbh any chat based websites aren't made for coding
This is very similar to sonnet models
You start hitting output limits extremely fast
Would not explain the huge decrease in performance
I'm just build prompts
Try using deep research or labs
true. Claude is the closest to being made for coding tho
Yeah, and even with Claude you need to be using their CLI or in the IDE
Can u add, deepseek v3.1, original r1, llama 3.3 70B etc
I JUST CLOSED THE TAB
wait
all caps
lol
i dont think scicode is a good benchmark to measure serious and actual coding skills. o4-mini-high is the leader in that benchmark....
try with swe-bench if it's there
Unfortunately not there on artificial anaylsis
dayum
Berd
thats a big difference
Also add
Other sonar models
I KNEW IT
What
Lmao pplx hard loosing models
They didn't run some benchmarks for r1176
or the other ones
so it just does not show
i mean, they are designed for web search lol
Beautiful people, how are we doing this evening!?
Not r 1776
Use sonar reasoning pro
And sonar reasoning
desperate for the weekend. I'm already tired of school lol and its my first week
Sonar reasoning is based off R1
they say this in official documentation
For deepseek r1 0528 prob not sonnet
They said its a bit different
They said similar
So i assume
It's r1 0528?
Wait how do we get to acces sonae reasoning pro in pplx?
Doubt it
And sonar reasoning non pro
You can't choose
Based on what I understand, deep research automatically routes you to the best model.
api
i think
When it first released it was most likely powered by R1
since the CoT wasn't obfuscated
and it always started with "Okay....." pretty common r1 trace
I gotta recharge my openrouter credits to test out the sonar models, one thing I know for sure is that they aren't sonnet or haiku
But it has its own model
Afaik it used sonar reasoning pro before at launch
perplexity/sonar-deep-research
Then it was switched to an OAI model
(or the auto router might just be giving me that)
I still think its a multi model
Doubt it
Because oai only acts when i used web
Wouldn't make any sense in terms of cost.
I had web off
This mode automatically selects the optimal models for your research needs, so you cannot manually choose a specific model when using Research
My best guess would be it routes to OAI models when it detects a hard query, then uses sonar reasoning pro for simpler ones. Pure speculation
they should add an option to independently choose the model for deep research
would be super cool
It was a thing for like half a day. Likely internal testing that people somehow leaked
I think I saw a post by TestingCatalog regarding this, it was quickly removed
So its like a hybrid?
I'm not too sure at this point
and I don't have time or a better way of testing this
Might get some Openrouter credits later this week to test out sonar
ig they are not gonna add it because it would be more expensive, so they choose the more cost-effective option
Spend that
take a wild guess
why I need to buy credits from openrouter
Limit increase?
Aside from being more expensive, it's more infrastructure, and just more work for the devs
|| I ran out of API credits on PPLX ||
Gonna cry
NEW: Anthropic will start training its AI models on user data, including new chat transcripts & coding sessions, unless users choose to opt out by 9/28 (it's a pop-up window that will give you the choice). It’s also extending its data retention to 5 years.
https://t.co/DCYvFh0Iqu

We are heading into the territory where they need more real human data because there is so much slop AI being created lol
I think it went a little off track
ehh, it was bound to come. It's good that you can at least disable it
and they give you a popup
I'm glad we have the chance now to make their models even better! 🎉
And then we have people using ai generated prompts, then they train on those
Going to be a huge issue going forward
Lol true
They are eventually going to run out of training data
and you get dirty data with mixed ai thats impossible to sort and clean
I don't think LLMs are the future of AI tbh but I guess we will find out
Real
So many kimi 🔥
Hehe, been using this for a year now
I cant read Chinese so idk what im getting 🥀
There used to be Long Kimi that had 2m context
O
ppt, medical search, scholar article search, kimi explore, translate
Was almost a year ago when they released this
Got removed I think in May
🔥
Nice
Kimi Customer Service 🗣️ 🗣️ 🗣️
prompt engineer, IT helper, rednote article writer, office work, paper edit, paper writer, trending website writer
pro user, but couldn't download the comet. any idea?
Whats the exact issue?
went to comet webpage, tried to download, the page did not show up as a pro user
only option was goes into the waiting line
pro user should downloads the comet immediately, did I miss something?
in us of course
No DMs promoting your products, services, or servers. You may post your own project in #😎│cool-projects only if it adds value and follows the no-spam rule.
Thanks
We need these soon in english 🙏
Fr

🤦
Happy Browsering
😄
I love you
nvm
honestly you can use venice.ai
completely uncensored
guys after so many tries I managed to make comet assisstant do this and actually do some useful work
can anyone tell me what is term used for this type of tool calling so that I can design prompts like that in future.
Note: This is not the blue border one. This tool call is it open a separate tab internally and does the work and shows the preview as seen in the image

Thanks, lol I contacted support but I used the wrong discord account for the ool link
X)
how exactly do i copy without citations / links?
I don't think it's possible to remember
If you speak in the text, export it to markdown files.
You will just have to remove the sources at the bottom
hi
not just the sources at the bottom, it copies like this. its all over the main text too
Weird, it's showing for me
wow wats that from
is that the web version? i am using the web version.
Yes
this doesn't show on my end
prompt?
is this a newer version?
Wdym
weird I don't have the icon with the i and circle around
is that complexity?
Bingo
oh
so its not there in the default version
yea doesn't count tbh
Just get the extension
yes not there
it has some other QoL features as well
What's special about this?
yea it has its own discord too
hmmm, i actually use pplx in ferdium. not sure if i can add extensions to that :/
Yeah idk about that
wow is ferdium good?
ferdium because i can have all the AI apps in one place like this
yeah i think so
hello guys
◉ Got Impressed
|
◉ Downloading ferdium
|
◉ I am more interested in trying new features so going for the Nightly Build
|
◉ Download complete, Starting the .exe file as administrator
|
◉ Oh No! Windows Smartscreen defender blocked it, anyways I trust this file Run anyway
|
◉ Installation process initiated
|
◉ Installation completed successfully. Launching Ferdium
|
◉ It prompts to create a new account. I need to fill in my details
|
◉ Adding different AI providers
|
◉ Oh No! Copilot service not available, I need to create a custom service
|
◉ Successfully created Copilot service, Now I need to login every service
|
◉ Someone known as "i love xlm it will flip eth" sent a message. but i must ignore until my current task completes
|
◉ Finished logging in to all AI providers, but seems like copilot and perplexity is missing the logo, I need to upload the custom logo
|
◉ Finished setting up everything
|
◉ END OF CoT
hello manager, I would like to speak to the manager
One because that one was free and the other because I forgot ngl
seems to be
ok people hello i am back

what the 😂
what did it reply with
best settings?
and are they any similar utilities?
is it still reasoning?
haha. btw you don't need a account to use ferdium
it prompted me so i did
maybe AGI-2 will understand the contexts better
xD
Hi @everyone
Hi
How do I get the pro role
Meta Confirms Llama 4.X & Llama 5
Mark Zuckerberg says:
- Llama 4.1 + 4.2 arriving this year
- Llama 5 coming next year
Good luck with fixing 10m input context 😭
MAI-1-preview doesn't seem quite good.
For a knowledge question it gets wrong, even Mistral Small 24B can answer correctly.
For lunch wheel of fortune it fails too (left), while mid models like Gemini 2.5 Flash can do perfectly (right)
my pplx gave right answer
I'm having video bro
u have pro ?
Yes
share thread, there is share button below
Check it out
@languid star
look at these sources lmao
💀
ask this question again in new chat
noice, ig earlier it was hallucination which occurs rarely
Is this sonar api free for pro users ...?
Oh but it happens to me a lot when asking trigonometry questions
then select a good model like gpt5 thinking in menu
K thx for ur help
nope
Sonar Pro API typically costs $5 for every 1,000 searches, plus fees based on the number of input and output tokens processed
Can someone explain to me how the search function operates? What is it actually searching for when I enter a keyword, because I think on average, maybe 1 out of every 100 searches I’ve done actually comes back with a result when I KNOW the keyword is in the thread multiple times.
What model did you select?
Even the lights are fans lmao
Claude platform has quite limited tokens wow
it's not even what they describe too i think
Perplexity is very very good tbh
i don't think many people have tried uploading PDFs and compared between Perplexity and Claude
"Context Window Size:
The context window limit, which dictates the maximum amount of text a single prompt or conversation can contain, is generally the same across different plans, including the Max plan. "
even the Max plans are just as bad
Yeah i have been saying that they should at least increase context window limits for max plans
it's actually quite interesting
PPLX apparently has less context window than Claude righT?
@gloomy cipher had tested the context window of reasoning models and he came to a conclusion is at least 64k
but PPLX allows way more tokens to be used in prompt
than Claude...at least in practice
PPLX Spaces even accepts 2.5 million word PDF in Spaces
So reasoning models have context window around 64k or possibly even more is just not clear
very very eye opening tbh
good to see what other platforms can actually do
i think most people don't realise this
Got
Docs which tell everything clearly
no one does that tho
Claude's docs are actually wrong too
PPLX has a whole lot of docs
which are also often outdated
That's The Problem with these Ai Companies

PPLX is a Decent Platform that's Why People Use it
Also i would Like to Mention The 1 Year Free Subs Given Recently
wow PPLX isn't far behind Gemini
200M Diff for Now
that's shocking tbh...considering Gemini's free platform is already very good for a lot of use cases, especially Google AI studio
i'd have thought Google would be much higher
Most don't know Ai Studio or How it Use it or They Think it's Complex
As Long GPT Hold The Benchmark Leaderboards others Will Keep Having Hard Time to Grow Their Userbase
PPLX + Claude 4 Thinking seems to be similar performance so far too
Yea but think for a second even gemini at full power can't read that much
So you could Deduce that the file cap isn't related to the models power
I would also like to mention technically Perplexity Pro is Free if Someone Can Hunt Referrals We Just Need Them to Add A Payment Method for 1 Month Free Trial
Both Sides Get 1 Month Pro Free
PPLX Max might be the best
i can't imagine how strong PPLX + Opus 4.1 is
it might be worth it just for that alone tbh
Claude's search isn't as good
If used referral both parties get 1 month free if both buy one month max
Just Use PPLX for Search-Research-Deep Research and Use other Ai Tools for Other Strengths
has anyone actually used this with complex PDF analysis
problem is most people who have Max are either just quiet, or don't actually know how to test it fully imo lol
they gotta do some giveaways for max, ppl gonna go crazy ☠️🤣
That is Surely not Happening
200$ is Expensive
