#💬│general
1 messages · Page 116 of 1
Nah like regular links
Gemini 2.5 Pro available now? I see it on Pro search
Yes!
has anyone tested gemini 2.5 pro on coding tasks?
How do you cancel your subscription? It's completely unavailable for me on Desktop
what is that
Only on the web you can 🙂
I actually had to cancel from mobile, the option to cancel on web isn't available. Likely because I subscribed initially from mobile
It was broken until perplexity servers where fully restored
They ripped off deepseek r1 for their r1 1776 model 😭
Its an open model
Is it possible to use any other model with voice chat? I want to talk to claude voice/voice
Ik
Claude doesnt have a voice model
But it doesn't need one. Perplexity already has that
It just needs the model connected to perplexity voice, which is probably connected to sonar now.
Actually a good way for perplexity to eat some of that anthropic lunch
It's a very popular forum topic. No solution on mobile.
Anyone here have access to voice mode on Web?
Will the Android app have the same voice features as iOS soon?
Same thing, need to clear app cache for actual updates or latest results
Thanks but i am referring to the hands free mode
Hi.
Is this telegram bot official?
@askplexbot
If this is official enough for you, then yes. 😉
https://x.com/AravSrinivas/status/1910345955838234899
Just found this tweet. Thanks, mate!
You can activate pro using “/pro” j think
@distant aurora
google will buy all the projects because they have a lot of money, it was already expected. im sure these developers are happy to walk away with the money and make more garbage to be bought later.
the devs could be google themselves and google just showing its money power in the media buying its own projects
Wouldn't that mean it's probably accessible if someone messes around with the app?
Not yet but on the way. Reviewers say cursor with sonnet is still better.
How do I see that thing in the bottom which shows system instructions?
Note that each space has a 50 file limit
Anyone have a missing message In the threads?
Check your msgs box mate @tame current
Gemini 2.5 pro on PPLX or on native Gemini app is better guys
is the context of Gemini 2.5 Pro decreased in PPLX?
Yes
Native will almost always be better for most models I believe.
Imho google ai studios is better than both
why
Great UI, video and audio uploads, more settings, ability to turn down censorship, etc
ok
But try both, that’s just my preference
but i got the gemini advanced, ai studio only let the 2.5 Pro free for a while right?
Perplexity is meant for web search, native one is better for everything else
Gemini 1.0 and 1.5 were free until 2.0 and 2.5 pro replaced it, I don’t know if they ever plan to remove it as free usage, but so far they have not
AI studio is not going to have any models deleted like that
They're removed only once new versions are out
rate limits are a thing though
no i got Advanced through the 2TB plan
ahh
but it's not meant for daily use right
i don't see it saving chats
you can link your gdrive and it saves the chats there
Is it meant for daily use in the way the gemini app is? No. Can it be used still that way? Yeah
My chats get saved, I think I had to connect to google drive or something before it started saving by default
Also if you wanted a more chatGPT like setup you could always use Open WebUI with the google free api
Guys, seriously does anyone used to get messages In your thread disappeared before?
I got half of my messages from days ago disappeared somehow.
by far genspark ai is the one who was able to read github repos properly not sider ai, not monica ai, not perplexity, not chatgpt, not gemini, not blackbox ai
Claude also has this capability
daym i missed claude
I don’t have this issue with my Pro plan. 😉
rich people (feeling sad)
I also prefer working in Google AI Studio, but for certain tasks, I switch to Gemini Advanced —> the 2.5 Pro model includes the Canvas feature and, more importantly, Deep Research with 2.5 Pro. 😉
Gemini also has its drawbacks. You have to call things by their names.
I use perplexity and it's great.
Perplexity also has the option to reduce censorship.
Yes, of course, I’m not saying otherwise — I’ve been using PPLX for a long time as well. The discussion was just about Google AI Studio or Gemini Advanced. 😅
Each task has the tools that are best suited for it.
Where do you get this message?
At top right corner in claude website
I'm talking about the message length limit warning.
Good morning PPLX gang
I imported my github repo to claude then told it to make a dockerfile by reading my repo
Then this limit error occurred
gif working for me
can you send this? https://ten * or.com/view/gang-signs-gif-23644682
I think I'm special but not in a good way lol
is there any special place in this server who have pro role ?
Why is perplexity like that tho
How?
Void of character in comparison to gpt
@distant aurora have you ever heard of this? A guy I helped told me to tell him what my chatgpt email was and so I gave him an alt email I had for chatgpt and then time passed and it had chatgpt pro 1 day later, I didn’t give him a password or a code or anything only the email 👀
I got rich thanks to a prince in Somalia
I hope the plan is to quickly resolve these findings
https://www.darkreading.com/application-security/11-bugs-found-perplexity-chatbots-android-app
Hardcoded secrets is concerning
those starting with pro- like #👑│pro-lounge or #1111765634267742408
maybe I am too old, but I studied security and half of those vulnerabilities sound not even relevant to the app. a lot of those are fault of the user or OS
Insecure network configurations
Whatever this is referring to is the app
No SSL validation or certificate pinning
This is the app
Exposure to the even older vulnerability CVE-2017-13156
Yes, this is rooted in Android but it means the app is exposed to the vulnerability, because it failed to implement then-standard changes to protect against it
Weakness to "clickjacking" attacks
This is the app
Cross-Origin Resource Sharing (CORS) misconfigurations in its application programming interface (API) responses
This is their backend
No detections for Android Debug Bridge
This is the app
And the hardcoded secrets is DEFINITELY the app
The others are either duplicated (no implementation of SSL pinning) or in my opinion, opinionated (bytecode lacking obfuscation)
Hmmm sounds legit but I have no idea what most of those terms mean
there’s so many issues with this ai
ahh, like these places
whats the process to get pro role
i've done it last month 2-3 time and sadly it didn't worked
@zinc quartz can we have a pinned message for how to get pro discord role?
Oh. I see.
You have to leave first.
ik
Then rejoin via the link.
lemme try again
didn't worked
@halcyon coral can you help him?
Please check #👑│pro-lounge 
Lol
Thnk u very much
You know what's funny? On you.com, on the default model. If you ask what version of (any model) are you, it will say it's the latest version of that model.
So they are misleading users?
I think it just hallucinates based on search results.
If you type gpt, it searches gpt and so on.
@signal hamlet @zinc quartz
I've already uninstalled the app just in case. I'll use the website, and hopefully the Perplexity team will confirm they've fixed those issues so I can reinstall the app on Android.
Thanks, we'll check the report.
Be careful with the links you click on your email. I smell phishing or something similar.
I never got an email or a link
What is he even gonna do with the email then.
And how did it get chatgpt pro?
How did he even log in?
🧠💡 ChatGPT just got a major memory upgrade. Now, it can automatically remember and reference information from all your past conversations, making interactions more personalized.
This feature is currently rolling out to ChatGPT Plus and Pro users, with Team, Enterprise, and Edu accounts gaining access in the coming weeks.
But, it's no...
5404
When is an update to deep research coming to bring it on par with ChatGPT’s
you ... do realize that OAI uses o3, immense context window and probably doing live crawling via real browsers, and it costs $200 for 120 uses per month, or $20 for 10 monthly uses (approx 1/3 [third] of use daily, so roughly one research every three days)
I think it is more realistic to target Grok 3 Deep Search - not that much longer responses and a lot less hallucinations
Even 10 monthly uses included it’s really good so much better then perplexity deep research like miles ahead in usefullness
well, they too have problems with hallucinations, but yeah, length of the report is astounding.
but I have a feeling when they removed GPT-4.5 from pplx it was because of high cost. they could keep it and give like 3 uses per month, but it may look a bit bad in comparisons and you get a bunch of users who didnt read how many uses per month/week and be disapointed that they wasted it on testing
but as far as I know, pplx team still works on DR high. so maybe it will land somewhere in between?
Wow, this is hysterical... after I flamed perplexity with a scathing review of an absolutely awful answer to my question, it instantly responded by asking me for a gold star... whatever design, coding, and computational bandwidth went into this reply should instantly be burned at the stake.
Grok-3 API Available now ... come on Perplexity
how to share like you do
You right click and click "Copy Message Link" if it doesn't show up you just need to enable developer mode on PC. Same thing applies on mobile with long press
Thank you
Yuh
You do know you can still select a model in a space right...?
Also with extremely recent UI changes, I wouldn't waste my time giving a huge paragraph to an LLM that doesn't know what you're saying just because it didn't find it?
this seems like the old FB users who sent death threats when any UI change happend on facebook lol
You moved my news feed. You will now face a class action, Face of the Book!
Why Perplexity Student Referral not recognising the .edu.... E-mails
Hey @rain anchor! Please forward your account info and this report to our support@perplexity.ai team. We'll help you with it.
No, that button and capability are precisely what used to be there and are now gone.
HiDream-I1 full review. HiDream full installation tutorial for ComyfUI. #ai #ainews #aiart #aitools #stablediffusion #imagegeneration
Thanks to our sponsor Humva. Use code HEYHUMVA to try Standard Plan for 1 month for only $0.1. Available for the first 1,000 users only. https://humva.ai/
Sources:
HiDream official repo https://github.com/HiDre...
perplexity can use this i hope
the game is so strong for this
Actually No... the UI did not change, the underlying functionality, which was foundational to my use cases, was removed completely from the application. As a software tester for over 20 years, I assure you, those two are nothing alike.
half the time it looks like maximum slop
My man, I just took that screenshot
To send it to you specifically
Are you implying that your specific device is not seeing what everyone else is?
Which device? I am currently on my web browser... Chrome, windows
Phone, desktop, browser, app
All of them
Zoomed way out to get the entire dialog in, but here is my Edit your Space window... no option to choose a model...
Same on Android, but I cannot get the entire dialog in one screenshot...
And if your screenshot is from the space non-specific "Enter your prompt" screen... then you are answering a different question than I asked. I already said, having to reselect my model every single time I enter a new prompt is what defeats the purpose of setting up the space in the first place.
Gemini2.5pro is a reasoning model, may i please know why it is kept in pro models rather than reasoning models @zinc quartz just curious to know. Please tag me while responding.
Because you're not looking where I sent. It's not in the edit space section
Yes, this isn't that big of a deal honestly. At least, not one to leave such a "scathing review" to a robot 😂
There's an actual feedback section here for engineers to review
how do I get the pro user role? I have to leave the discord server and come back?
come back using the Pro link in your PPLX settings
Does the new Perplexity API include access to Deep Research? And if so which models power the DR api?
Okay, but your conflating two different things... the scathing review was about the complete factual inaccuracy of the answer from perplexity about perplexity itself, not the removal of the functionality
and " this isn't that big of a deal honestly" -- well, not to you, maybe... but that is a very personal and subjective statement. My context is different.
Sonar Deep Research --> https://docs.perplexity.ai/guides/models/sonar-deep-research
seems relevant, reposting from pro-lounge (I am not pplx staff):
gemini 2.5 pro has one of the lowest thinking coef. I saw, just bellow 3. sonnet has almost 7.5 . also google's TPUs are quite fast, on openrouter 120-190tokens per second. so it feels more like non-thinking model, because it is fast and needs over two times less thinking tokens.
also has lower base price, I wouldnt be surprised if running sonnet was like 4x more expensive than gemini 2.5 pro
plus I believe CoT is not sent via API at all. so from a perspective of pplx and users, gemini 2.5 pro is closer to non-thinking model
edit: by sonnet I mean thinking sonnet 3.7
Here’s the full list of models: https://docs.perplexity.ai/guides/model-cards
Thanks for explanation i got your point, but only thing is it is really thinking well in Gemini app and top model in benchmarks so thought it should come under reasoning.
It says no access to link
Ah yes, Pro! Sorry 😉
Good logic, i got your point, thanks for sharing bro
I am also pro member, but not joining pro group as i am not continuing perplexity for now.
thanks!
Was that $50 just a scam? I signed up but only got $5 because I'm already a Pro user.
Did anyone actually manage to get the full $50?
I have the pro mode but I can't generate a picture
I don't understand
Why can't I generate a picture in pro mode ?
Anyone from UAE?
Cool
egg
Toodles
i think we get it after hearing about the exiting things in the zoom meating
meeting*
insert egg here
Will you update sonar to the new meta models?
love ur pfp
R1 1776 is pretty good!
But sometimes, when I use it, feels like sonar is few steps behind.
How's it hanging folks
on mobile every researches are conducted through Pro even though you select Deep Research.
Hello all
How does Easter Bunny works?
Their just randomly egging people @Denny@shut dagger
😭 why did it become such a thin drop down box
when was image editing a thing
ong
thats been an issue for a bit iirc
real
and istg they removed an image gen
they moved it
maybe it switched it to gemini but since it isn't a choice on it it doesn't show...?
ask it to describe an image the same you would before and the image tab should popup
then just "generate image"
I have a couple questions: 1. With the perplexity telegram bot... that's not linked to your account right? It's just a bot that people can use to try out perplexity a bit?
i put money instead of money 😭
Yall are so fast to edit out image generation from announcement
yeah
Champions League is the biggest highlight of the announcement 🗣️🔥
LOLLL
same as the discord bot
ok thanks
I thought we finally got native image gen so I tried it.
genius AI
that means it's coming thoo
at least it gave you some json. That's kinda neat
Unless you know better
hidream just released and i can't run it locally 💔
what model was that?
Another question I have is, what's a good general model to use for Pro searches? I know the reasoning models take longer... I just want to use something that is reasonably fast and kinda capable. Opinions ok
this was Auto mode, used Gemini 2.5 Pro.
gemini 2.5 pro
it's relativily fast
But it's slow to initiate
Sometimes, model thinks for too long, makes me feel like stuck
true true
flash being removed was sad
it was my main fast model
sonar's prob the fastest rn?
Yeah, it was great model for it's size. Creativity was solid and felt natural & human
yes, 1100tk/s (according to ceo)
It was always been fastest model in perplexity
it's a custom llama 72b iirc?
Yes, LLaMa 3.3 70B*
oh it's 3.3?
thanks. yeah sonnet used to be good. I hear good things about gemini
Gemini 2.5 Pro is current SOTA
Is there any way i can make is to that I always get images for my query’s?
theres a guide in here somewhere
o3 mini is also fast and smart
honestly... I have pretty uninformed views on things but I was pleasantly surprised by grok (hadn't tried it before)
Sonar was disappointing
I think i've used grok 2 a total of 5 times
that's a reasoning model tho right?
it's hybrid
I meant o3 mini, sorry should've replied
I heard that Grok 3's API version is dumber
yes
really? that's so unfortunate
💔
honestly, I don't even care, we got gemini 2.5 pro
and that's fine, just seems like it has to iterate a little more
but Grok 3 is still needed for some unhinged roleplay! 🫠
I think grok 3 thinking has never been released
Yeah I will say that grok 2 really impressed me... I like grok 3 but like... anyway we don't get grok 3, yeah it makes sense maybe it's cos the API version is dumber
It was grok 3 mini the whole time
yeah I like grok quite a lot. I will say that grok 2 was like ... really cool when I picked it
😭
it’s smarter for me
I dunno like... i'm not usually asking very complex things but grok seems able to navgiate social media stuff handily enough and that's nice. I guess perplexity does to some extent in general
Grok 3 hallucinates like crazy
Hallucinations allows creativity
perplexity discord and tele use sonar right?
interesting. well I was asking it for x posts talking about xyz and it was an obscure topic, like it was a new piece of furniture that was just released
I kinda don't blame it but the results it came back with were like "OfficeFurnitureNerd" said
and then I clicked through and it was all made up
kinda hurt my feelings
like you can hallucinate but don't trifle with my emotions
😭
yes
It's like "CredulousShillPosterForThisStupidGrokInquisitor opined"
Image Generation and Editing << It’s a function in the release notes that just uploaded, but can‘t I use it?
they accidentally put that in the release notes
it was edited out soon aftert
Aha, thank you!
unfortunate
in order to reduce hallucinations, models are made more smarter, and being smarter usually means professional and well ordered so they train it that way, which kills it's emergent personality vibes thus reducing creativity
still on the website lol
Image Generation and Editing
You can now ask Perplexity to generate and edit images by asking questions.
What model is used for image generation?
The current default model is Gemini 2.0 Flash Experimental. We’ll continue adding the latest models as they become available.
Can I choose what model is used for image generation?
Yes. You can set your model preference from the Settings menu. Click the gear icon, select Preferences, then choose your preferred Image generation model.
Currently supported: Gemini 2.0 Flash Experimental, FLUX.1, and DALL·E 3.
interesting
well I still really like grok so it's ok
the link is in announcments
you can ask grok like "What's the scuttlebutt in silicon valley? who's been getting fat?" and it'll tell you without being like "Obesity is not a choice"
Finally native image support
That image generation function is good and it adds the texts
🥳
there’s a documentation on Telegram bot?
The image generator crashed or has limits? It generated a more or less normal image, but now it generates simple images :).
🥚
🤣

I don't have gemini 2.0 flash in the image models list
just Flux and Dalle
Is it still in the ab test phases?
Me neither. Aravind! 🗣️
seeing as they edited it out, atleast from the discord message, i dont think its coming this week
oh nvm - #🔔│status
gemini?
^I wonder which ai image generator it defaults to as well
@halcyon coral spam?
What a boring scam. Does anyone actually fall for this junk?
Clearly they do since the scams continue
Hi guys, i try add @askplexbot on my telegram private channel ask it a question. Seem like no response. Did i missing something ?
I had a friend give me a 6 month subscription, it's really nice
Because I had canceled mine right before
🥚
So the egg came before the chicken
Ugh must be nice
Yeah, it's good to use PPLX in case there's errors going on with ChatGPT
Thank you 🙏
I like to use multiple, ask them the same thing, and grade their output by comparing them to each other
👀
Life is good
No way
hello
Hi
Hi
I summon eggie
Eggless egghead
With following link you can get Perplexity Pro for free and unlock features like unlimited Copilot queries, file uploads, and enhanced web search. No catch—just smarter, faster information access. Grab it while it lasts
You need edu mail to sign up
https://plex.it/referrals/WB112GUC
Egg
official response from the CEO
Nice
:/
Give me
egg
Woah eggs
every time i enter perplexity with web, this happens. And due to some reasons it's not passing the validation. what can i possibly do?
Headless egghead
#📰│server-news be in the top 10 of the most eggs collected and get free pro for 3 months
Wrong reply first… :P
Thanks
The top 10 egg collectors
Oh I read it wrong
I replied to your message by accident
2023 👀
hello 👋
Smol and nice server
Here, it's 2024... (server-news)
Chat will be dead unless there's a strong and active topic for discussion.
Or mostly people ranting.
Mhm
Perplexity's Gemini Image Generator lets you add text to your image without any external tools. 😁
bot lives in the past 😳
when egg

Hello, on the website, the page loads and then disappears to a white screen. What's going on?
For all French with Bouygues (bandyou)
C’est inclus dans le forfait !!!
no, missed so many eggs
Lol. Same.
Hi, guys. If I’m subscribed to Perplexity Pro, does it ever make sense to use the "Best for each query" AI model? Or is it better to always choose the one suited for specific purposes? And if it does make sense, then what’s the reason?
Each model has its advantages and disadvantages, depending on your needs. My advice is to use Claude 3.7 in reasoning mode or Sonar for fast mode if you can't wait. You have to test each model and see if it suits your needs, the ultimate choice is yours.
I understand all this, I've been using Perplexity for a long time, I've just been wondering if it makes sense to use the "Best" option all this time.
Here I share with you the promt and the image generation using Google Gemini through Perplexity:
Cinematic render, Unreal Engine 5 style, hyperrealistic, dramatic lighting accentuating contrast. Male detective and female detective, Caucasian skin, back-to-back in alert pose holding handguns.
Male: Piercing black irises, short black hair, medium beard, wears a sharp white trench coat.
Female: Striking pure light pink hair (natural texture) in a high ponytail, wears a fitted white leather jacket.
Details: Photorealistic high-res textures on skin, hair, clothing, and weapons.
Good luck.
awesome
what do you guys think of the new llama 4 models?
Meh
i only noticed the low price about it so far
Disappointing, isn't it?
I don't recommend R1, unless you use it to drastically reduce censorship.
Yeahh
Gemini image is really trash
Lol
dalle is really exaggerating, lmao
Try my prompt using gemeni or flux, I shared it above.
That's from Terminator
fr
Not bad, needs improvement.
hmm
I already found the link to the statistics:
https://www.perplexity.ai/rest/user/settings
these limits are monthly or daily or yearly or lifetime
@distant aurora seems they fixed it!!! 🥳 🥳 🥳
✊ ✊ ✊
You can modify the prompt and use it for whatever you need, it's an idea of use in Perplexity.
I can set a preferred model in android app settings. However, how does this choice affect the models used in:
a) Auto mode?
b) Pro mode? (Does it only use the setting's default, or the 'best' available Pro model?)
c) "Deep Research" mode? (Same question - locked to settings default or uses a specific/best model for deep dives?)
d) When selecting a specific model directly in chat (like "Reasoning")? Does this just override everything else for that query?
They are from the pro version and those numbers may be subject to change over time. The numbers shown is the maximum usage per day on the models and they reset, each has its limits.
per day, got it
Why does the web version allow selecting an AI model in chat, but the Android app doesn't have this feature?
About pro mode, yes, it uses the settings set by the user, it is very different in the Android app.
Is perplexity up right now guys?
On the website, all available models are listed, but in the thread, not all available models are shown.
android app is lagging behind, you can say features will be rolled out and will be tested on web and desktop or mac apps first
android app is at least priority ig
yup
For image generation using FLUX and DALL-E 3, i can say :
Flux make good and realistic images but don't follow prompt properly
While DALL-E 3 follows prompt but exaggerates the image and make it unrealistic
is there anyway we can use custom prompt for image generation? I feel like AI is not good enough to write prompt by itself
Thanks man
Create image
Use this prompt for generating -: 1boy, absurdres, black necktie, black pants, breath, chainsaw, chainsaw man, commentary, denji (chainsaw man), fuunyon, highres, looking to the side, male focus, monster boy, necktie, pants, red background, sharp teeth, shirt, shoes, sneakers, solo, squatting, teeth, white shirt
I figured it out, it needs web mode to be toggled on
I was expecting image like this
daym
is this Gemini or Flux dev?
Which look good
Flux
2nd one
the world is a simulation.
move 40 penup
move 100 force f
EQUATION GENERATED =
(((-1*m*g)+n+f)=0)
((m*a)=0)``` this is a box pulled above the ground with a force. the equation of motions are being generated.
turn 135 pi/2+a
move 400
move -300
turn -90 -pi/2
box m b
ABC c f
EQUATION GENERATED =
(((-1*cos(a)*m*g)+(sin(a)*m*f)+n)=0)
(((-1*cos(a)*m*f)+(-1*sin(a)*m*g)+(m*b))=0)
(((-1*cos(a)*n)+(-1*c*g)+d)=0)
(((-1*c*f)+(sin(a)*n))=0)``` for box on an inclined plane.
😂 nicest kind of arugment ever made for world being a simulation. i know.
What is the code to make the message look like this?
use triple backticks
with single backtick you can do stuff like this -> Hello , world!
test
Now yes, thank you.

Try the following prompt:
Cinematic render, Unreal Engine 5 style, hyperrealistic, intense battle scene. A lone male soldier fiercely fighting during an alien invasion. Set against a dark, atmospheric background suggesting a chaotic, debris-strewn urban battlefield at night.
Soldier: Caucasian skin, piercing black irises, short black hair, determined stubble beard. Wears advanced white combat armor (perhaps with subtle grey detailing). Dynamic action pose, firing a futuristic energy rifle. Intense, focused expression.
Lighting & Effects: Dramatic lighting with highlights from the rifle's muzzle flash, glowing alien projectiles, and distant explosions creating stark contrast against the dark environment.
Details: Photorealistic high-res textures on armor, skin, weapon, and environmental debris. Focus on action, grit, and detail. 16:9 aspect ratio.
This is how it turned out with Gemeni 🙂
Hi!
DALL-E
FLUX
Excellent.
intelligently trim context
intelligently trim context ┬─┬ノ( º _ ºノ)
||intelligently trim context||
interesting dalle response ...
What is that
Sam nerdman
Sam nerd man 😭
Eggggg
Deep research high seems to be rolling out in pieces here and there
How can you tell?
I'm getting more broad source searches and there's way more in-depth reasoning per search step
The amount of total sources and search steps is still quite allot less than the full thing
But it's definitely better than the previous 20 sources and it giving up after 2 or 3 searches
On a PC or mobile? Only deep research high shows its research plan in advance, except the app does not support it
Im not hoping for that sir, its weekend also sir, that rolling thing where SOME of the blessed users experience a glimpse of the pre planned DR sir where 2 or 3 times then back to normal sir is been like 4 weeks like that sir. So no sir, unless kwak says yes sir i might believe sir but if the CEO tells it sir no i wont believe it sir.
Only see it happen in web app
I looked at the threads on mobile and the reasoning doesn't show up
Yep, it seems to be rate limited to once every few hours maybe
The search steps are similar but the amount is much less than the full DR high
This might be how standard will be
And high will be what we saw earlier
Can we make it do that we can prompt with the keyboard and the selected thing goes in the chat? for the android voice assistant
Like draw to search? Otherwise just copy the message, it's simple
Also thanks for reminding me to check, I finally got Gemini live with Astra
I can't believe we got draw to search on Android, but they still don't care to give us the updated voice mode
egg
🥚
🥚
gemini context is so peak
Didn't we already have circle to search?
Draw to search with perplexity assistant
yes, but with google gemini as the assistant
PPLX is still king when it comes to medical research. It has been greatly usefull now that I'm a first time dad
hey guys! new here!
You underestimate grok and chatgpt.
Anyone doing the student referal one?
W
Chalk-o-mint
More like Gemini 2.5 Pro deep research
Nah. I'm free user.
From some South Asian University.
So u get premium
@halcyon coral can you help him?
Oh. Lmao.
But I will have to make a purchase for you to claim, right? I'm free user. For now atleast.
Though I'm looking around.
Wdym
An Easter egg has appeared! Be the first to collect it!
🏆 First collected by feline9655!
- feline9655
- bakedpotato28
- jenyx01
- william_j._billy_butcher
- chocomintuwu
and 2 more people...
Collection window has ended.
If i give u link u get one month of free premium
It only works for sign ups.
Someone will have to make fresh account with a university email that's eligible.
does anyone know an ia that would be able to generate litter announcments? like this one:
lets say i put in the photos of the dogs and the text
can any ai do the rest
???
chatgpt 4o image gen
is it worth buying?
Nice
Besides, it's not that hard to apply filters and font text using some image editor.
im a graphic designer for dog owners already and spend alot of time doing my posters, so i figured maybe...AI could do it
Oh. Alright. Reasonable.
is chat 4 the only option?
Most likely.
You could try microsoft designer.
But I'm not a fan.
Try everything for free before buying any subscription.
thanks!!! i have perplexity pro and i like it...but yk, i have not really gotten used to it yet, i have been a chat user always
Theoretically, using Gemini Flash 2 image should ibe a native image model as well. But I am not sure how good it is. (Which is available in perplexity)
Not as good as 4o
very low amount of effort did this:
Also I believe Recraft AI is built for designers as well
OMG. this is just what i was looking for!!!!!! i owe you big time omg! what did you say it was called?
Still can’t generate pictures on iPhone?
It's available at https://chatgpt.com/.
You might need to create an account, but it should be accessible to free users once logged in. If you're on mobile, just download the app — it works there as well.
i have an account, but do i need pro?
neither do you need pro nor do you need plus as far as I am aware
Damn. Nice.
He's a professional so he'll need to buy eventually.
I'm assuming he wants to know about rate limits of plus vs pro.
My answer was upgraded to pro search without my consent.
Great.
I literally disabled pro.
thanks sm! im a girl btw, my profilepicture is my little brother lol
but thank you sm!!!
Hey guys how do I join the beta? I need the memory feature
Lmao. Ngl, I thought you would be something like Ruben.
No haha, my name is act ruby
Coming from chat gpt this feels very basic
@left stream thoughts? https://www.rxddit.com/r/OpenAI/comments/1ium6zv/perplexity_is_going_down/?rdt=64246
I'm calling it! I'm bearish on Perplexity.
Aravind has set out on achieving the impossible feat of dethroning Google, and what he has achieved so far is close to perfection. So full respect to him for that. But things are looking tough for Perplexity from here and we already have the tell-tale signs.
OpenAI recently launched search an...
realized today that sonar is becoming my most used model on pplx
What are the pros vs cons of sonar?
It's literally a reddit embedder. Here. https://www.reddit.com/r/OpenAI/comments/1ium6zv/perplexity_is_going_down/?rdt=64246
There's a 15 sec window I think.
its crazy quick so it has been particularly useful when I am with friends or in group settings and need accurate answers. also I feel like it has the best contextual memory so when i need to rapidly hone my search it helps a lot. I guess the only con is that it doesnt do multi step reasoning like claude 3.7, but im finding more of my pplx use cases dont need a whole lot of reasoning, for now. I just go to claude for anything i feel pplx cant do
people from openai caring about perplexity🤣
btw only good thing about perplexity is claude, they nerfed drs that was very useful to me
I'm messing around with Gemini. I haven't used it much.
Apart from that, I find grok most versatile.
@sick arch fyi, grok is available at 1/4 th price in India.
i don't want to spend 6k😔
How
grok deepsearch and think is currently free for limited queries, that are enough for me in a day
for indian users
did u subscribed it?
@left stream impressed with Gemini 2.5 pro response length.https://g.co/gemini/share/17a025007252
I never subscribed it but am considering it.
I haven't tested grok's coding performance.
I strongly believe that the benchmarks of grok 3 online didn't use reasoning switch.
if u code too much, it would be useful otherwise i think pplx is enough
Egg
Response length is nerfed on perplexity.
Grok 3 reasoning is not available yet
How did you
^^
I mean, why don't they just test it through the website/app itself?
@left stream grok for comparison https://grok.com/share/bGVnYWN5_f58b96d9-3f49-4f80-bc13-b3a786201e0c
I suspect that the think button uses grok 3 mini, like the reason button is o3 mini in ChatGPT
Rate your fav LLMs
Are we including deepsearch?
For me
Perplexity
Deepseek
Grok
Claude
ChatGPT
Close enough 😩 mb
Gemini 2.5 pro deepsearch isn't available for free so I won't comment on it.
2.0 ain't it.
I can run some Gemini deep research for you
google should atleast give some queries for free for deepsearch
I believe that's 2.0, not 2.5 pro.
Oh. Okay. Wait.
do u have gemini advanced?
So funny thing I killed my MacBook
I'll come up with something in a few hours.
And I got advanced since it comes with 2TB cloud storage
u can get 1 month for free
Instead of using iCloud for backup
apple is not even in race🤣
That's why I chose Google instead of paying iCloud just for storage
which country u from, how much it cost there
Canada, $26.99/month
Nice
canada is little cheap, its costly in india here
Just get grok if it's discounted there
argh, never mind... 😭 😭 😭 I just created a bug report about it. https://ptb.discord.com/channels/1047197230748151888/1360672938235396416 @distant aurora
Isn't Gemini advanced 20 usd in india?
he is talking about cad which it is showing currently around 1600 and here it is 1900 around
Oh
https://www.perplexity.ai/rest/user/settings
Here in the statistics it shows Gemini 2 Flash, not Gemini 2.5 Pro, could it be that it is outdated?
I have the gemeni app here on android and you can use Gemeni 2.5 pro.
I don't think it shows the deep search switch when you select 2.5 pro.
Egg
Fried
Damn. Nice egg.
a pretty one indeed, nice find!
Hyd?
pplx down?
Working for me.
hey guys!!! im back with another question, what ai would you say is best for school work? im writing a big assignment about greenland and i need help. chat?
claude 3.7?
Are you replying to me or her?
you
You want me to check if Claude 3.7 is working on perplexity?
yes
Gemini 2.5
Really?!
Go back and check in library, open the chat and see if the result is there.
empty
.
refresh should fix this
☝️
I also sometimes get the same error, I just restart the app and response is there
Nah, it's fked
same xd
The Easter bunny was here and left an egg for you to find!
🏆 First collected by bakedpotato28!
- bakedpotato28
- william_j._billy_butcher
- jenyx01
- geri_888
- venkybeast
and 2 more people...
Collection window has ended.
I did close and re open app.
claude search 3min, and error..
Then is pro search good?
Yeah
Probably some sort of glitch going on.
197 sources.
Lol
What do you really get from the easter eggs after all?
I don't plan on keeping it after I don't need cloud storage
3 months of pro
that's sweet
Gemini app sucks compared to AI studio so half the value is gone
Single prompt and you'll be botting that bot easy
Death and taxes might either be a bot or an egg hunting God
Fr
Anyways, questions for deepsearch:
What's the waqf drama in India about
Compare the discrimination compared to Hindu counterparts
@cobalt vale put both in same chat.
Who tf is edu.contin
Bot
I need this bot.
Let’s all vibe code one
One after the other?
Yes
I have never vibe coded before.
Not on the level of making applications.
It’s not bad, just a bit overwhelming cause you have so many options to choose from
Or both together?
Run the first report then the second?
Which model is good for vibe marketing?
Vibe marketing is a thing?
Most likely perplexity and grok 3
Maybe? I have no idea
Is it running or waiting?
Do you need to edit the plan?
🥚
Bro is legit lurking for eggs only.
First place?
Honestly though
We don't have much active members.
You might easily make it in top 10 considering it's only 15 days.
I've seen this server more active than openAI server
Easter egg alert 🥚
Probably cuz open ai doesn't do as much marketing or give aways.
They changed it from 4 days to 15
Very mad
Whatttt???????
Lol
pfp? what is that?
https://g.co/gemini/share/2fd2caf3a106 @distant aurora
Hold up
You have at least 12 eggs
How to check?
Just search username in this channel
Now I'm impostor
lolol
I need template to copy for the bot message.
Just doesn't show when names get cut off after 5
This guy saying cap to hide his own position💀
if I want to use the AI models as a general model instead of search, do I use pro search and switch off the web search toggle?
I don't know how to check.
also what are the daily limits for reasoning models? thanks
lolololol
Well now you can see
Doesn't really matter though, you got it
Reduced to 300
okay thanks
Yes, although it won't be an identical configuration to other chat providers
Oh I see
When? They've always said 300 on Pro sub description
I think so?
First
Not big enough.
Second
lolol
🐣 Easter Egg Alert! 🥚
What's that? It's an Easter egg ready for collection!
How do I put the text in that format.
space
@distant aurora https://g.co/gemini/share/06bfc46bf87c
hie
Hop to it! An Easter egg is waiting to be collected!
🏆 First collected by mares1317!
- mares1317
- parthsloved
- william_j._billy_butcher
- death.and.taxes
- bakedpotato28
and 1 more person...
Collection window has ended.
Nice
Okay I'm number 1 now.
🐣 Easter Egg Alert! 🥚
What's that? It's an Easter egg ready for collection!
**🏆 First collected by you
lolol
why me 🥚
I'll remove that.
🐣 Easter Egg Alert! 🥚
What's that? It's an Easter egg ready for collection!
🏆 First collected by ⏳
⏱️ Collection Window
This egg can be collected for 15 more seconds!
what is bro trying to do
Prank
he do a little trolling
Eggies
🐣 Easter Egg Alert! 🥚
What's that? It's an Easter egg ready for collection!
🏆 First collected by ⏳
⏱️ Collection Window
This egg can be collected for 15 more seconds!
It works fine for me, I've had no problems at all. The internet connection you're using probably has a problem or something.
@cobalt vale @crystal carbon
Ilysm king
Huh
Pinged you for collection.
Wow thanks
Such a nice guy
Yep

I guess previous one was more convincing than the last one.
I think so
We just have to make sure we at least get top ten for our free 3 months
Bruh that’s almost $100
If I'm among the top 10, I'll give it to someone in need. I have my Pro subscription.
Need is subjective.
How is it almost 100?
60
We can make an egg hunter club group chat at this point
Timezones for what?
But how would that help
Well yes but 15 seconds is a pretty short window
Lol that's probably a bad idea
Will Comet be able to automate the collection of eggs?
Probably not
Maybe Gemini 2.0 Flash Lite is cheap enough
🥚 Easter Egg Hunt Time! 🐰
Hop to it! An Easter egg is waiting to be collected!
⏰ Collection Window
Collection window will last for 15 more seconds!
LOL nice idea to test
@crystal carbon
lololololol
I appreciate it
I've missed one in the same minute
What do we think will be the required count by April 25th
100 eggs?
More
200 eggs
Lets use AI to find the pattern they show up in
Around 130ish atleast.
I was gonna try this. But idk how to export channel chat.
Would they really not sufficiently randomize it?
Algorithms are patterns.
Though current ai probably ain't powerful enough to read them.
And we won't be able to export a large enough sample.
🥚 Easter Egg Hunt Time! 🐰
Hop to it! An Easter egg is waiting to be collected!
⏰ Collection Window
Collection window will last for 15 more seconds!
Also they probably have chat activity influencing the bot
@cobalt vale @valid charm @crystal carbon
wtf
