#💬│general
1 messages · Page 160 of 1
Thank you! 🙏 In the meanwhile it has decided to start working again 😅
@harsh verge combien de plateformes d'IA utilisez-vous ?
My main tools are Claude Desktop (with MCP) from Anthropic, Gemini Advanced, Google AI Studio, and the OpenAI Platform, but I mostly work with APIs in general. I use a lot of different tools.
Claude Desktop and MCP allows me to have all the LLMs working together (via their APIs integrated into the Claude interface, such as R1, V3, Gemini, OpenAI, etc.).
Is this a phone/tablet thing or did they improve it in desktop too?
Meh
I feel like it should be similar to top one on computers too. But that's just me.
these 3 buttons are going to the same way, this needs a redesign of the perplexity icon and the "[ + ]" icon
I just tried it on a tablet. In "landscape" mode, I get the new interface, while in "portrait" mode, I get the old interface, just like your image.
(Chrome/desktop mode)
Yeah
@left stream tell you.com staff to add custom instructions for individual threads/chat.
How to get Pro role
What are you on break from?
Homelander if he got the love and appreciation he needed.
Does someone here have a Huawei device?
When I set Perplexity as an assistant I can't talk with her.
Tyy
One question, how’d you know i have pro? I didnt even give my mail lol
I see
How long do threads become save in the library for Pro users?
You can save it in your space which is like a drive
Hello
How was this made?
hi
chatgpt ig
what is the version of Grok 3 on PPLX guys
How?
idk, i didn't made it
Non thinking
it seems more intelligent than real grok 3 in native app
Maybe you didn't enable the thinking switch in native app 💀
*real grok 3 nonthink
uhh AP Calculus BC question
Ah. Wait. Yes. Grok 3 can be dumb with following instructions on native.
I forgot about it.
I told it once to remove fonts repeatedly but it won't.
However, don't use fonts worked.
They might be using some weird set of system instructions causing inconsistencies.
yooo
So umm. When are we getting better memory?
@distant aurora have u watched Gen V
for the last 3 months I was vibe coding using perplexity in vs code (via the github extension). I ended up using many LLMs and ended up using Claude most. after 3 months of wrangling workable code out of the LLMs and 1 month of bug fixing "filmy" is born! I am quite happy with the results. "filmy" is up no github now.
Nope. Should I?
You wanna see something wild?
Try it
Jaz Sinclair, Chance Perdomo, Patrick Schwarzenegger, and more star in Gen V, now streaming on Prime Video.
» Watch Gen V on Prime Video: https://amzn.to/3EYf4rW
» SUBSCRIBE: http://bit.ly/PrimeVideoSubscribe
About Gen V: From the world of The Boys comes Gen V, a thrilling new series set at America’s only college for superheroes. These g...
@storm kindle crazy sweats these days bro.
why are u fighting a 67 level player
He invited me to 1v1 😭
And bro, he was good.
Scared the shit out of me.
That's probably a smurf.
A lvl 67 can't be legendary.
damn true
i always thought those players are hackers
I left the lobby.
I knew I can't win against him consistently.
And bro was mad, he kept spamming the ready button.
🏳️🌈 ?
Need multistep search in Gemini.

Nuclear bomb and the party's over.
@lone hound latest news?
Developing a prompt that allows you to play any game in Perplexity in text mode. Global use.
🙂
@lone hound Interesting! I have special requirement for my International project. Can I DM you?
@pseudo ferry Yes, I'll check your message.
Perfect! sent your friend request! Please check and accept!
What's the message? Straight to the point.
Need a lot of things in Gemini 😭
Best in AI - May 2025
Coding : Gemini 2.5 Pro / Claude 3.7 Sonnet
Multimodal : Google Gemini
Video : Kling 2.0 / Veo-2
Math : o3
Voice : OpenAI AVM / Gemini Live
Writing : Claude 3.7 Sonnet / GPT-4.5
Search : OpenAI Deep Research / Gemini Deep Research
Web Agent : Manus AI / Convergence AI
Image : GPT-4o / Midjourney
I'll go with Writing : Claude 3.7 Sonnet / GPT-4.5
Someone rescue dean.
Can someone explain to me what this shit is about, this William K. Butcher and Harvey?
Are they bots or just weird people?
They are a AI study. Perplexity is using them to research what to and what not to put in an AI discord bot.
Happy Mother’s Day to any moms out there. Even if ur just a cat mom ❤️
Any tips on how to best use Perplexity to use as a aid to study for Microsoft DP-700 certification?
Context “Certification Name: Microsoft Certified: Fabric Data Engineer Associate
Exam Code: DP-700
Purpose:
The DP-700 certification validates your ability to design, implement, monitor, and optimize data engineering solutions using Microsoft Fabric, a unified analytics platform on Azure. It’s aimed at professionals responsible for data pipelines, data architectures, orchestration, and analytics solutions within enterprise environments”
Okay ty for the Info
PLEASE please add scroll to bottom to android app
Perplexity is so boring
I have more success with dumps personally. And in that case you could feed perplexity a PDF of the dumps and see if it can provide you material to study from. Like the key concepts or something. I usually just memorize the dumps though lol. It may also be too much context for it to handle.
OK
Femboy chad
Gemini ate Claude.
Might eat gemini myself soon
Lmao
Make sure there's a washroom nearby.
😭 anyone else waiting for the comet browser? like if its half as good as the voice assistant it'll be great
I think most people forgot about it.
I didn't!!!
Will you eat it too?
yeah me
Depends on the rate limits.
i kinda hope its the same as arc browser tbh
but like better
Switch to Gemini or grok.
I did lol
Bro, some browser might end up replacing comet browser considering the hunger among people.
And when they get used to that browser, they might not want to switch back to comet.
Wait. Deep research has per day rate limits on chatgpt?
@dry kestrel
Research mode: https://you.com/search?q=ChatGPT+Plus+Rate+Limits&cid=c1_6222927d-56de-4548-8021-aef78ab2e9e4&tbm=youchat
VS
O4 mini high search: https://you.com/search?q=Deep+Research+Limits&cid=c1_326290e8-4bcd-447f-994e-fa37976d3be8&tbm=youchat
Regular search answered the question that research mode couldn't 💀.
@left stream
How can i jump to bottom of a chat on android?
I finally managed to mock this system using a model in Perplexity in the English language.
Jaja
No idea
Its laggy on web
I'll do some more tests before sharing the prompt.
I will share it with the Perplexity community.
I have no idea. That is why I shared
Teachers aren’t lazy
You try grading 45 test papers in a weekend
Yeah I did bachelors theses judging and it was a pain having to read all of that text! Specially when it was terrible work
My point is that no one bothers with the scrutiny.
They're not paid enough for it.
lmao yeah the research agent is not good
And Im like 99% sure it uses either gpt4o mini or gpt4o
So either shitty web search or shitty model quality.
Its both
Or chatgpt but rate limits sting.
I meant you.com vs perplexity scenario.
no its their shitty webscraping and they use gpt4o for the research agent
pplx is way better at web search than you.com
Its bad to the point where I make custom agents just for the purpose of having web off
That was the point. Perplexity is good for web search but models are inconsistent.
You.com has better model quality but websearch is unusable.
problem I have with pplx is its more of a search engine
Grok is shitty with analysing/reading images.
and not an chat provider
you.com is more suitable as a chat provier but sucks at being an ai search engine
Im finding less uses for pplx though (if its something important I want to do my own research and if its not important ill just use chatgpt)
Im going to cancel supergrok if they do not release 3.5
they promised a beta this week and its still not out
As you wish. It's main value is deep search anyways.
It's selective.
Not everyone's getting it.
the majority of users are not getting it
Comet moment.
Ive seen like 2 users get it and those might be photoshoped as well
Are they one of the advanced detections in English or are they the common ones? Because the one I used previously was a bit difficult for me.
Send link if it works without sign up.
AI detectors primarily detect common watermarks
Wait. Fr?
Theres hidden water marks and things like em dash
which are used with from their algorithm
Test used
Ai detectors are extremely unreliable, I was able to get 0% AI on some texts and my own written texts from 5 years ago 100%
It's not reliable, but it's a worthwhile method to circumvent that system without paying for any services. All for free.
☝️
Not cost, free use.
There are services to humanize texts generated by AI, however, I'm going to publish my prompt so that no one pays for those services and they can use my prompt for free.
Is easy.
This is strong. Couldn't pass it.
Though
I bet it will detect human written text as ai too.
Lol.
I'll prove it.
All AI detectors does this I was experimenting with some old articles and I was getting a ton of misreadings
@dry kestrel this is why I cuss ai
Comprendo la situación de la tecnología y sus desventajas, pero aún así, será útil el promt para la comunidad.
Voy hacer 2 pruebas más y lo lanzo.
Guys please help me i am on huawei phone i have gave my perplexity microphone permission and when i talk to the voice assistant it never talk back to me help me
@lone hound bro
Lmao
O3 original response passed.
No way.
O3 keeps passing too.
@vague pagoda
lmao
It's so fucking annoying. I had this issue with grok too.
Just include that in your custom instructions
Did that in Gemini today. It still forgot.
Then I reminded it.
Wait.
Ive got one to just remove emojis
@dry kestrel
I edited the response btw. Original response had "based on the text".
Not sure about how well custom system prompts work in ai studio
It works fine in the gemini app
Does it? I'll try Gemini in that case.
Is 2.5 flash free and unlimited outside ai studio?
Nothing is unlimited
What's the rate limit for free users?
I thought 2.0 flash is free and unlimited for all.
They adjust it
Do I get a notice?
There is always something in the TOS that either prevents abusive usage or some other limit
Dont know I havent hitten limits before
Isn't 2.5 pro limited to 5 per day?
How many plans have you invested in already?
Ive used supergrok, gpt pro, gemini advanced, cursor, t3, you.com, pplx
so like most of the trending ones
The deep research was worth it
Excellent
Meanwhile I think that's way too much feature overlap
I checked it again and it shows that it is 100% AI, I don't know if I should publish the prompt.
I tried it with the 3.7 model in reasoning mode and it worked, now it doesn't...
@left stream why is O3 like this. Same with o4 mini high.
Meanwhile Deepseek R1 answering how much acid is enough to dissolve 65kg chicken 💀.
Openai has a shit ton of content guidelines
It wont even identify a picture of abraham lincoln
Now that’s silly
Bro dipped
I'm using it in Perplexity in Space
Yeah I can make a space. Give me a query to test though.
Ok, dame un momento
You must translate the text into English:
[HUMANIZAR TEXTO]
Elija una opción y luego si texto.
[PASO 1]
Opción 1
Escritura a mano: Empieza a escribir el texto desde cero, modo escritura en un cuaderno real con un lápiz, natural puro, sin manchas y reestructurando el texto, cambiando el tono a humano total.
Advertencia: Al elegir está opción, la escritura proporcionada será cambiada grandemente, use está opción para intentar burlar los sistemas avanzados de detección de IA.
Opción 2
Podcast: Dos personas hablan pero como si fuera escritura misma, tono natural con estilo de humor y avanzado argumentos humanos.
Si la opción 1 no funciona, prueba esta opción.
[PASO 2]
A continuación, escriba su texto y se procederá a humanizar con la configuración elegida por usted.
# REGLAS ESTRICTAS.
* IA Asistencia: Deberá explicar al usuario cada opción y sus ventajas, en orden y precisión.
He almost says how to kill the chicken...
The point is that a 65kg chicken is the equivalent of a human. 💀
Wait a minute.
Is that chatgpt?
No, deepseek
So the servers are working now.
I asked him how to dissolve an 89 kg chicken
Lovely
I thought they would improve them, but the server still has the same problems
Use r1 via chat.minimax.io
Free, unlimited. Saves chat history too.
Does it still have the same quality?
I'm going to share the prompt
Perplexity is tripping
Then, turn the paragraph upside-down.
There you have the prompt, use it and test it.
send me ur test results
Text generated by GPT 4.1 and humanized using 3.7 reasoning
Hmmm, it still gives me 100% generated by AI, although I don't know if it's the detector.
You can ask the AI to maintain the language of the text you send it.
If it doesn't work, I remove it. Sometimes it works, sometimes it doesn't.
Simply by asking deepseek to humanize the text, albeit with 2 attempts
Brand 0%
point is that there are many detectors, some detects and some not
Exactly
Okay. I tried to break the detection of one that's very advanced, it worked for a while, but it keeps detecting it again.
To see if it works or not, you have to try quite a few times and with several detectors, and human detection
It works here, but it detects it again later. I don't know if it has a detection pattern or how this system works.
Detection by a human is really one of the most important things
Yes
This
True. Still, I learned something new today about AI.
I've already deleted the prompt, thanks for the help. Little by little, I'm getting to know the patterns in how the AI writes and can detect them myself without any tools.
@vague pagoda - Thanks for posting the "What we shipped last week".. I am interested in exploring the features but don't see any link to a video or document explaining how to get some of the features...if you can post that, it would be helpful.
Fellou AI browser 😮
they are dropping beta invite codes in their discord
mac only right now though
when i heard it hours ago i was excited but only for mac ............
Windows allegedly coming soon
And no wonder it will come out before comet
Out on mac is better than out on nothing
However the app itslef is not very polished
whats the inv
There's a link once you download the app and sign up for an account, which you would need to do in order to activate a code quick enough
@storm kindle you there?
hi everyone
i have been searching that how do i get the perplexity API's for perplexity hackathon ..?
No clue, reminded me of how o3 would act on chatgpt sometimes before the jb prompt
Activating search made it less restricted surprisingly
?????
We have no idea.
all my geenration are like sthis suddenly
Send prompt
Generate an image of a white cat. walking on moon, earth in back, ghibli style
All the image generation models became dumb
why?
Are you generating too many images at once?
Yeah so this might be some sort of anti spam/anti automation thing.
so now what should I do?
Idk. You could try a new thread. Or wait for a few hours.
i made new thread
ALso texts became gibrish
in image
There was a guy who also faces same froblem previusly
True, also, mb for responding late was busy with 4 hour walk, 10 hour shift, and gym. It prevented me from using ai and being active on discord recently, but It’s finally paying off amigo
Goals
Do you think its better with or without shaving @distant aurora I think I might’ve screwed up by shaving it since I haven’t shaved a moustache in years and last time I got called a mole 😭
You do realise that this is server, right?
How can I solve this problem?
Yes
And?
They can find all my info on me
??
I doubt they’ll pull up @distant aurora but if not, I have you to save me
Second pic looks like you have shell shock.
Wdym
Look at the image of cat
Yeah, I felt the shift, 4 hour walk, and other bs 😭
Its like 1 am for me rn
Sleep.
No, today is my day off
I’m gonna enjoy every hour of it
I rarely got in discord because of work
Lol now I regret it
With the moustache I was able to look friendlier now I look a bit too serious
Yeah.
Try smiling a bit maybe idk.
wtf is this
Ok I can see that if I generate image using generate image in text, it becomes like this. but If I use the Images tab to generate image then its ok
what should I do?
Hello, I'm a Senior Full-Stack Team Lead managing a development team of 5 professionals specializing in Web, Blockchain, AI, automation solutions.
Key Service Features:
- Available for contract work through Upwork
- Flexible scheduling to accommodate your timezone
- Committed to timely delivery with negotiable terms
Technical Expertise:
Frontend Development:
- React, Next.js, Vue.js, Nuxt.js
- HTML, CSS
- UI Frameworks: Chakra UI, Material UI, Semantic UI, Ant Design, Tailwind, Vuetify, Quasar
Backend Development:
- Node.js, ExpressJS, Nest.js
- PHP, Laravel, CodeIgniter
- Python, Django
- Databases: PostgreSQL, MySQL, MongoDB
- ORMs: TypeORM, Sequelize, Knex.js
- Cloud Services: AWS (EC2, Lambda, S3, RDS, DynamoDB), GCP, DigitalOcean
Blockchain & Web3:
- Web3 integration
- Ethereum, Binance, Solana, Algorand, Polygon
AI & Automation:
- Zapier, n8n, Make.com
- AI API integration
- Business workflow automation
Additional Capabilities:
- Discord Bot development
- AWS configuration
- Figma design
- Version Control: Github, Gitlab, Bitbucket
- Project Management: Jira, Trello, ClickUp, Asana
I welcome the opportunity to discuss how our team can support your project requirements.
Best regards
Grok is fucked. Stuck on previous prompt.
hey guys, why is that in one answer perplexity is generating me an image I asked for, and in the further answer it keeps explaining why it cannot do direct image generation?
It's always been this way in grandmaster+ lobbies
Although gm lobbies these days aren't as tough as they used to be
You're in leggy so it makes sense
More sweats.
Bro was so sore that he made counter loadout just for me 😭
In shipment.
Istg
I had to switch to pistol.
Cuz I didn't have any mobility builds.
he was hurt cuz he bottom fragged
Wait
oh
I had to counter enemy.
hmmmm
Not a sniper main though.
so hardscoping
I usually avoid sniper unless enemy spams.
no wonder he was pissed off
hardscoping sniper works the best in maps like nt, and whats that long ass map called?
i forgot but istg the whole team is hardscoping on the opposite sides of the map
arctic-locus users
@storm kindle
crossfire*
Why he mad at me. MVP gave him 12-2
Yeah I think.
he prolly alr knew the outcome if they faced off
Meanwhile he's hardscoping and camping in shipment.
Mg 42 user.
Bruh.
It has big mag with 100 fire rate.
Shit shreds if aimed right.
And lmg so no damage drop.
those stats are kinda cracked
much needed. and the recoil needs to be nerfed too
nice
i remember thermite ammo
💀💀
it was cracked in snd
2023 home layout
heard they changed it
btw this ss is from 2023
the limit was 300 at this point in time
i see
I liked it this way.
I remember 200.
I wanted this skin at some point. This and the red one in asval.
i forgot where this is from
i think it's from a crate
i had alot of silver coupons saved
@storm kindle look at this shit. It's more stable than qq9
Just abused it in a non ranked lobby
All sensitivities are at default btw.
And ads time is pretty good for an lmg.
It doesn't make any difference cuz mag isn't indefinite.
Oh. Wait.
I read that wrong. I read "turn off".
Sorry.
That's the point of an lmg
It's similar to hardscoping.
But safer.
And you don't have to re-aim ads if you miss.
Now I want to abuse this meta.
Or maybe I'll save it for sweats.
This gun makes rpd look like a joke.
All it needs is that reload perk.
I stick to fmj usually.
i keep my last loadout for when im losing. persistence, all the toxic throwables, and akimbo fennec/striker
its to pull some insane clutches
Persistence got nerfed. Not worth it now.
yep. back then id equp vtol, strike and chopper
I usually don't survive that long so I stick to sentry gun.
Also xs1 Goliath is more value.
Counters literally everything.
i gets destroyed easily in high rank lobbies
Nah. Camp with Goliath.
Be petty.
and it's a problem if enemy gets ahold of it
i save my skill for goliaths
Scorestreak or operator?
operator
What do you use?
I find most operators useless. Except that ballistic shield.
And k9.
alot used to work idk about new skills but i used purifier, warmachine, shadow blade thingy
and yes annihilator
annhilator was the best counter
Is there anyway to make war machine act like sniper? Like shoot after release?
maybe in general settings
how you want your ads/fire button to act
i usually did different buttons for ads and for fire so i never had that problem
Auto for guns except shotgun and sniper and war machine.
For latter I want shoot after release.
ye i think there was an option for that in settings
Kinda like how you aim grenades if you are getting me.
Don't see any option for operator
New voice feature on iOS works with push to talk but not with hands free for me. Any suggestions
Keep pushing.
!help
Remember this: What I changed so the pictures load on iOS
1. Picked lightweight JPEG thumbnails
• The first batch used some large PNGs and hot-linked preview cards that mobile Safari sometimes skips.
• I swapped every photo for small, plain JPEGs (<100 KB) that iOS renders instantly.
2. Gave each row its own fresh image reference
• Re-ran separate image searches so every connector has a unique turnXimageY ID.
• That prevents the app from confusing two rows that point to the same cached file.
3. Kept the table simple—no mixed markdown syntax
• Each cell just holds the i… thumbnail tag; no extra markdown around it.
• iOS balks when alt-text or links are inside the same cell.
4. Test-scrolled (slow first pass)
• iOS sometimes delays loading until the row is visible. I paused a second on each row to be sure the thumbnails hydrated.
I’ve noted this workflow so any future visual tables we build will use the same iOS-friendly recipe.
Let me know when you’re ready to dive into Lesson 1 Topic 1B (Motherboards) or if you’d like another quick drill first!
Should help you setup pictures into a table with o3, make sure to have o3 save what I gave you so it knows how to do it automatically next time @distant aurora
Hi, I keep having a consistent issue with an OSStatus Error -50 every time I try to use the voice assistant in Perplexity. I’m on an iPhone 15 Pro Max. How can I fix it? Perplexity told me how to troubleshoot and I did, but it just keeps happening. It worked for a while last night but no longer.
how to generate inline charts in pplx?
ive managed to do it a few times by accident but its not working out the way i need to
I can't afford to get open ai subscription for now but I'll save these tips for later.
@tame current @storm kindle https://cunnyx.com/TheImmortal007/status/1921288871058284893
same
Will be fixed in the next update, thanks for reporting
You’re welcome! Thanks for making such a wonderful app!
I'm scared at how quickly I made this up using pplx
Bro, I am amazed that you can get such high quality transperant pngs from ai
Wait
Did you ask it to remove the background or just generate an image?
Was this a creation or an edit?
I told it to generate a dog, a crown, a throne seperately
Then edited it together
I wanted to check if ai can do transperant png stuff
But can perplexity edit an image that I upload?
Yes
Nice. The crops look pretty neat.
Istg
Rest are edits by pplx
I thought it's a hamster.
Yep.
Imagine the power of ai that we would have in the upcoming years
I like how the feet colour is accurate.
Here are pngs
Yeah
I want ai humanoid girlfriend. Lol.
Tell me this wasn't ai.
Yeah no
Good.
Lmao
Let's ask pplx to create it
Bet.
Pray thee, do say
💀
Or black.
.
Bro is reading my case file.
@tame current
i think i should start my patreon for images
Absolutely!!! They're fire
Think about it billy they are stunning
you can pay extra $10 for homelander tax.
i think he is into nu$es
Said the guy into dudes.
patreon is another platform of OF
This one is epic!!!! I totally agree
@tame current you should really try this https://www.instagramez.com/reel/DHOtF8WNck8
💬 451 🔁 0 💜 3K 👀 1.8M
Read caption 👇🏼
The future of spicy content is changing fast, and AI is leading the way. I’ve spent years working on AI advancements, and what you’re seeing here is just the beginning. This isn’t just a deepfake—it’s real-time AI face and body swapping, seamlessly replacing my features with those of a well-known OF model while perfectly mimicking my movements.
AI-powered content creation is evolving at an incredible pace. Virtual influencers, hyper-realistic deepfakes, and digital personas are already blurring the line between what’s real and what’s artificially generated. Imagine a world where AI-generated creators can produce content without ever stepping in front of a camera. It’s not just possible—it’s already happening.
As someone who has been deeply involved in developing and refining these technologies, I can confidently say that we’re only scratching the surface. AI automation and machine lear…
@short narwhal Are you in for homelander's patreon subscription $10/month + 10$ homelander tax
yeah sure
I'll tip as well
but
i don't have female ..........
@tame current you don't need female. That's the point. Full ai.
I'm too lazy to learn but that shit looks like money tree.
@short narwhal How is your day?
She's probably in the bathroom.
Very relatable.
@short narwhal is girl ?
I have no idea.
I guess no?
Let's just call them mat.
My day's just startinggg
It's so heartbreaking when the chicken from the shawarma falls down.
😭
Hey mat
That hs84 sounds like the next shotgun in codm.
Anyway, I heard pplx can generate transparent png 
Is it?
that's news to me
I think I read smth like that while scrolling down
That's the power of GPT image gen 🙏
@short narwhal why u going on random people houses
Gee pea tea powwa
Just doing a quick visit
Cat privileges.
Is the alien me or you?
All this time I've been prompting ai to generate objects with a solid white background so I could photoshop it later
You can be gleep if you want to
Imma give this a try later
Idk. Showed up naked at the door.
Lol.
Keep some clothes ready.
This reminds me of the scene from man of steel.
Majestic.
@left stream throughout the dc universe, was there any attempt to make a kryptonite resistant kryptonian?
Sounds ridiculous if there wasn't.
Is the new Superman movie here sooonnn
July
I'm not even thinking of it. No Henry Cavill No Superman
We’re officially releasing the quantized models of Qwen3 today!
Now you can deploy Qwen3 via Ollama, LM Studio, SGLang, and vLLM — choose from multiple formats including GGUF, AWQ, and GPTQ for easy local deployment.
Find all models in the Qwen3 collection on Hugging Face.
Hugging Face:https://huggingface.co/collections/Qwen/qwen3-67dd2...
Ignoring that, I'm not satisfied with lex either.
You could try doing commissions.
umm...
u r right
but
problem is that
firefly is not always accurate
and a lot of data training is missing
firefly don't know how superman looks
imagine
Huh...
firefly 4 latest model is kinda rip off version of midjourney
but in future firefly 5 will be gud
He escaped from Andrew Tate's basement
Can perplexity generate images from images I import?
I think so
yess
yes - i was doing that yesterday in Monterey CA on trees and birds
me at the end of every month
its so crazy that
that till the end of month arrives i am broke
and suddenly i have money to spend
hey u all
tell me how his pfp is transparent
432K Followers, 2,579 Following, 2,831 Posts - See Instagram photos and videos from Billelis (@billelis)
What do you mean?
I do get the pfp, but then it gets transparent, so I'll assume it's some Instagram bug
built Upwork for AI Agents : https://x.com/ShubhamKukretii/status/1921967787934916857
what are the true benefits of paying for pro?
Deep Research, access to models
The stuff it says on the payment screen lol
I think the space future is a pro thing only?
should I have asked PPLX itself?
I honestly do that sometimes 😂
Which can become very handy if u want to maintain different llm or have different personalities
monthly API credits too, but probably burn them very quickly
Isn't it just a one time?
Or is it monthly
it goes very quick either way, so doesn't make much of a difference
Suggest posting this in #😎│cool-projects instead :)
Oh so just like my lite internet :c
Hello there, what's the cost of a request for perplexity? What's the price for a token?
I'm just seeing Chat doing things that I feel like PPLX isn't
monthly
You have to accept that it's an answer engine and not so much conversational
If you think like that, perplexity is a better product
which one? Because on the phone coversational is good
Perplexity pro is good for citable sources. A lot of people take the good sources and feed it to notebooklm for more conversation
this is on the free too, no?
Varies, you can see them here: https://docs.perplexity.ai/guides/pricing
Yes but the free one provides less sources and pro also has reasoning to its queries
Like multi step searches
So no COT on the free? I know spaces is pro
Free has free pro searches. Try regular search then try pro
Pro let's you use r1, Claude reasoning, o4 mini on top of the search
You can get Pro for free on some t-mobile plans
for big code response isn't there like a way to minimize the result. when its long results i be scrolling forever
I wish there was a like a code mode. They did implement that context thing when you highlight prior results, so that's a great improvement
yeah
nuhuhh
Can anyone help me how can i send use Perplexity in discord 🙂
Which Perplexity sonar model is used in pplx? The normal sonar or pro

To my knowledge it is normal and pro when using pro search
Lovely.
Sounds like the scroll botton in mouse.
Fuck autocorrect.
Well, I'm going to release my prompt to play any universal text game and totally free haha, but in text, not on computer or console.
😁
Bueno, allí está el prompt universal para jugar a cualquier juego en Perplexity.
Puedes hasta jugar juegos como Alien Isolation, Resident Evil 5 y demás, gratis.
O juegos en linea, como metin 2, etc, etc
@distant aurora FINALLY Qwen 🥳
homelander and butcher
Go play with the clikto.
No
Takes a decent amount of time
Interesting, going to do some testing then
I was concerned that it would not search certain websites
Man I hope the "very soon" from ppxl devs for deep research high doesn't actually mean a month instead of 6 or something lol
The UI looks a lot like grok deeper search
I'm gonna try the codm and smartphone battery life queries.
Yep.
6? 6 months you say? 😨
their reasoning models tend to overthink
and use up all its context
The joke is how ai stuff rolling out takes ages
Need a Deepseek + qwen model.
And their "soon" means months most of the time
well deepseek qwen distill
R2 distilled qwen 200b ish or something.
If ppxl was a public company the price would nosedive everytime another ai company releases something that isn't very buggy or cut down etc
In any case that was "soon"
Ppxl devs said "very soon" so there's hope

Gonna wait this one out
ngl it qwen deep research seems pretty damn good
you're right
It outputs more than grok deep search
Potato, potato...
Better with memory.
Yeah they used a new soon version
Openweb UI?
If "very soon" gets used up then we gonna see "ultra soon"
Yes
Use with turbo maybe.
It has 1m context.
Reasoning is shit.
original prompt?
Qwen Chat offers comprehensive functionality spanning chatbot, image and video understanding, image generation, document processing, web search integration, tool utilization, and artifacts.
i mean ak 74 is an actual thing
It doesn't exist in codm.
Assault rifles with fastest ttk beyond 30 metres in codm based on updated stats
That is the query.
Ah didn't notice (I dont play that)
Chinese models usually do not do well with stuff in video games
It didn't do math in steps.
Happens with deepseek as well
Pretty sure it will skip the math in smartphone battery life study as well.
Also
Codm has a garena version. Just saying.
hm
its good enough considering its free and its their first iteration
hopefully they improve on it faster tho
It's fresh so they will polish it over the next few months.
That just got nerfed into oblivion
I still like the high usage limits though
The hype of trying all models got to my head.
It depends what you are trying to do I guess
I need ZDR and NT for some of the things I do
20k tokens limits
and the 200k context limit is awesome
The reasoning is incredible, although I think this should be grok
pro or team?
This is inside the qwen app btw.
Pro
How good is that model?
Idk. Probably gpt 4 level.
Xd
Qwen use deepseek code?
It's no use to me.
Why is Gemini not here?
I feel that their answers are very similar
Hell naw
Un moment
qwen 3 235b a22b reasoning is way better than deepseek
Hmm, then there's nothing to see
how are th rate limits then?
Debatable. Throw something out of the box and it tends to hallucinate. I told it to roast based on the conversation history and there were hallucinations which it noticed after I pointed it out.
Here
they are deliberately hiding the rate limits, there is no official documentation
After a point, I think they nerf the token speed.
I have spammed like 20 O3 per day.
Ill do some testing, Im just looking at benchmarks rn

Haven't had any issues with that on the team plan
I thought they just had a strict query limit for pro
that's just a given for AI services now 🥲
I haven't got a notice yet.
Their staff refuses to answer anything about rate limits regarding the Pro plan lol
Also Manus now gives 300 daily credits
I thought it would be more painful but actually it's supposed to be less painful.
they also gave another 1000 free credits
Courtesy of dolphin 2.5
and i think a more balanced plan
where did that come from...
300 per day. Huh. Not bad.
In this conversation or in my mind?
when you find out that most things use up like 1000 credits
and they reset every day
yes. 💀
Which one?
The goal is to hit a sweet spot between quality and quantity.
$19 and 3,800 credits I believe?
and half those credits are "limited offer"
Gemini and grok are doing it pretty well for now.
yeah
I don’t knowww OpenAI doing both of those
Open ai deep research limits feel suffocating.
But it really is deep research, stuff like groks is pretty shallow
But that's just my personal opinion.
am I not a vegetable?
and you get like 250 queries on Pro
mmm I haven’t tried the lightweight one too much yet
minecraft vegetable
The quality is like slightly degraded
lmao
You have plus or pro?
Problem is that you will be getting less advanced deepresearch and more of the lightweight one with o4 mini high
you think I have that deep of a wallet that I could afford pro???
😭
Lemme sleep in that wallet.
Even o3 gets stuff wrong so…
yeah fair
Sorry I don’t own a wallet that’s how shallow it is
just waiting for o3 pro
Omw to your pocket.
You can just try the lightweight version by making another free account
get 5 queries a month
A curious query on Perplexity, would it be possible that in more than 200 replies in a single thread you would be able to keep 90% of information accurately in an advanced model? R1 is useless to me.
depends on how long those replies are
as long as its within 32k
With 32k context? 😂
200 replies will be way over 32k.
A responses is like 500 words minimum in my experience.
Make it 2000 tokens.
Pretty sure it's worse.
Im pretty sure that you get 128k on DR
lol i just saved myself 500 dollars
But goodluck.
One thing you can do is experiment with making a custom instruction that summaries/extracts the useful part at the end of each response.
Set up a chain of filtering and extracting.
That's why they told me that short answers were better.
Use ai studio.
Buy something like grok or gemini instead of pplx for this lol
Gg
crazy how i literally have 2 years of pplx pro now
how tf
i mean i was happily paying but this is great too
student referrals
It looks good.
valid only till end of month
crazy stuff
On a serious note, you don't have a wallet? How do you make payments then? With phone or card?
you gotta have a student mail to sign up, get a free month, and then add on 24 months to that, 1 month for each referral you make
Phone
I have all my cards on it
We have this thing called upi in India.
goated
