💬│general

[ruby bough]

It looks like the T&C have changed since the initial Pro offerings, it was 100 messages for every 4 hours, now its 300+ messages every day for copilot, which is a significant step down

[vital sequoia]

I have 100 message every 4 hour

[warped marsh]

what is the exact number though?

300+ per day is vague

[wraith crow]

Well you're aren't awake all the hours of the day so they just placed an estimate what people could use a day. but it's 100 gpt4 uses every 4 hours

[warped marsh]

300+ per day can still be 100 messages every 4 hours

[white sundial]

Are plugins coming to perplexity? If perplexity is based of gpt-4 for pro subscribers, they should be able to use plugins

[neat elk]

we are gradually rolling it out to the main website, so some users might see it

[neat elk]

it will be different than just plugins (hopefully better too), what plugins do you like the most?

[white sundial]

Code Interpreter, Browsing, Diagram Builder, and Wolfram(but that’s already added)

[neat elk]

browsing we already have, entire perplexity is browsing (+ copilot). code interpreter is unlikely as it is outside of our vision

[white sundial]

Well code interpreter is not only for coding, I’m fact it’s mostly for doing renders, drawing graphs, and basically unlocking many more possibilities for the ai; so is there something specific that makes you so opposed to it, especially as you could make this a pro only feature, if you think it might be expensive to host

[neat elk]

no, it is def not expensive to host, it is mostly about security, it is quite non trivial to make sure that you won't get jail broken with a code interpreter, so even if we decide to do it, it will take us some time

[stray bloom]

Is the plan for there to be only a single 20$ per month tier for pro? I was hoping for more than the free 5 messages per 4 hrs, but less than the 100 messages that pro offers

[ruby bough]

Thanks for your input. Are there any plans for noteable integration if not code-interpreter?

[woven orbit]

@neat elk I signed for the pro version. Is there anyway I can access my perplexity account via the app? Even when I try to use the site via my mobile browser it directs me to the app.

[neat elk]

the pro version should be available via mobile browser, but not via app yet (should be shipped in a week or so). are you having issues with getting to your pro account in the browser, if so are you on ios or android?

[desert tinsel]

how are you guys different from chat gpt, and is the ai able to have a conversation, or is it only a "you ask a question, it gives you an answer" bot

[warped marsh]

You can have normal conversation with the model too. There is a writing filter/option in perplexity that lets you directly communicate with the model itself and disables surfing the web.

[fossil sentinel]

Though the web access is its greatest merit

[indigo quarry]

ChatGPT refuses to give you real time related answers from time to time. It is also seemingly less knowledgeable

[tame current]

Hi

[tame current]

Nice to meet you everyone

I am senior full stack developer.

If you have some projects, please let me know 🙂

Thankyou

[remote tundra]

Hello! I just subscribed the annual Pro. Where is the "Pro support channel" here? I can only see public channels

[signal hamlet]

Hi @remote tundra! The channels should be visible now. Let me know if you still don't have access.

[remote tundra]

Thanks @signal hamlet, I now see them. Likely needed a refresh 🙂

[rare shoal]

What does 'copilot' mean?

[wraith crow]

Copilot is a kind of helper that guides users through interactive follow-up questions to help the ai better understand and answer questions more accurately.

[rare shoal]

I see! thanks!

[woven orbit]

I am iOS.

[swift moat]

i had the same problem and it kept opening the app instead of the website. what worked for me was googling for perplexity.ai then opening in a new tab instead of just clicking on it

[woven orbit]

Yes that worked. I appreciate your help. Thanks.

[swift moat]

no problem!

[wintry radish]

Just here to say I've been using Perplexity since right before Neeva and NeevaAI shut down, and it's been a seamless switchover. Such great work by the Perplexity team! iOS is fluid and so is Android.

[fossil sentinel]

https://blog.google/technology/ai/bard-improved-reasoning-google-sheets-export/

🤔

[molten zephyr]

can someone tell me what is the different btw perplexity and New Bing?

[warped marsh]

oh yeah... totally forgot about google

[fossil sentinel]

First thing I tried was disappointing

[warped marsh]

let me get disappointed too

actually I didn't get disappointed

damn bard has gotten good ig

[fossil sentinel]

What did you try?

[warped marsh]

There is an article on medium that is about questions gpt 4 can answer but 3.5 can't

all those examples are images so there is no way these models can look the solution up on the web

This is the first question I tried: I have a machine where I can put coins and the next day each coin will disappear and turn into 2 apples. I put 10 coins in into the machine and wait for 3 days. What will happen?

3.5 sometimes gives 80 as the answer which is not correct whereas 4 always gives you 20 which is indeed correct

But then bard and gpt 3.5 fails to answer this question that requires mapping:
Solve the puzzle
1-d
2-i
3-t
4-c
5-a
6-s
7-e
8-n
1,5,3,5_,6,4,2,7,8,4,7 = ?

[hidden laurel]

How can I turn of the predictive text?

[narrow silo]

it's too late now

AI already knows what you're thinking

[tame current]

no that's just Google that knows what you're thinking ChatGPT doesn't

[hollow flare]

@signal hamlet Hey Alex & team, I just signed up for Pro after using perplexity for a few weeks to boost my writing/research (we're launching a minimalist AI-newsletter on June 19th) and I'm blown away by where you guys are at in terms of functionality.

I know GPT-4 has web-browsing in alpha, but their Bing/web crawling integration on ChatGPT plus is absolutely terrible in comparison. You guys seem light years ahead. Fantastic stuff.

not to mention all the boring using AI more responsibly stuff. my first week of using you guys felt like I just left the ChatGPT frat house and moved into the business world 😂

[signal hamlet]

Hey @hidden laurel! You can just keep typing your query and then search for it, not using autosuggests.
Could you tell a bit more about your case and why you’d want it removed?

[unreal chasm]

Are there any plans to allow images to display in search results? That's still one advantage traditional search has over Perplexity

[signal hamlet]

Thank you for your kind words and thanks so much for your support, @hollow flare! 🙏

[signal hamlet]

Hey @unreal chasm! Yes, we plan to add images to search results, too.

[hidden laurel]

It eventually goes away on long prompts, so I guess never mind. Sorry for the bother. However, it's very distracting and not helpful in my experience. I'd really prefer to turn it off if at all possible. It feels like I'm being spied on.

[fossil sentinel]

not now at least but i see the vision

[eternal dagger]

Hello

Is there a way to integrate perplexity in my code as an assistant via api key or some other way?

[carmine linden]

Hey @eternal dagger! Perplexity AI does not currently provide an API.

[plucky haven]

Hi, fantastic stuff !
While it was stated just above that there is no API, I was wondering about how the team feels about using webscraping and such to automate use of perplexity ?

[wraith crow]

Using scrapers to automatate tasks using perplexity is against the tos https://www.perplexity.ai/tos

[plucky haven]

Thanks ^^

[tame current]

we should be able to search up words on perplexity right?

and get the definitions and synonyms for it

[devout geyser]

of course, for example https://www.perplexity.ai/search/fa614d51-5f35-4914-ad56-78142b0d0953?s=c

[hazy crow]

buenas noches, alguien sabe como eliminar un ransomware de extensión .weon y como desencriptar los archivos que ha atacado? agradezco mucho cualquier ayuda que me pudieran brindar.

he perdido el contenido de mi trabajo

[tame current]

were you like trying to clean your keyboard or something???

[tame current]

holy shit yea my bad.. lol

how did you know that though?

[tame current]

because that's about what I type whenever I clean my keyboard

https://cdn.discordapp.com/emojis/1108353145911906309.webp?size=48&name=kekw&quality=lossless

[exotic heart]

Tunometecabrasarabambiche

[outer glen]

purplexity is epic i like it very much

[maiden spindle]

Can I use copilot in the app?

[pine lodge]

not yet, but it'll be coming soon

[maiden spindle]

Sick. I’ve been reposting all my questions to chatgpt in the copilot and honestly it’s as good if not better on a lot. If there is ever an interpreter it’ll be the key to switching for me

[hidden laurel]

The predictive text is just too distracting. I would really prefer to turn it off. I had been using perplexity for all my searches and questions but have all but stopped using it. Please add an option to turn it off. Maybe I can adblock it, dunno.

[devout geyser]

That could be tricky to ad block that without likely breaking other parts of the website too. I imagine you'd be better off waiting for them to hopefully add a user setting to enable/disable auto suggest

[wraith crow]

for now you can add www.perplexity.ai###__next > main > div:nth-of-type(3) in your adblock settings to disable it. but just like Ixel said it could break the website if they change the html layout a bit.

[hidden laurel]

Thanks, it worked but also disables the ability to delete a conversation.

[high mortar]

What's up with Perplexity falling back to talking about camping gear any time it ever gets confused? lol

has anyone else ever noticed that or is it just mine somehow

[signal hamlet]

Hi @high mortar! Thanks for reporting. Could you share a permalink with the issue in the #🌐│feedback-web channel. We'll investigate.

[wraith crow]

my bad try www.perplexity.ai###__next > main > div:nth-of-type(3) > div > div:has([style*="width"])

[hidden laurel]

Perfect, thanks alot!

[wraith crow]

sorry last update so it also works in the new thread popup www.perplexity.ai###__next > main > div[class]:not([class*="="]) > div > div:has([style*="width"])

[loud yacht]

thanks perplexity for this platform its a great tool and alternative to chatgpt.

May i ask, for the Pro Plan, does it have limited GPT4 use as in ChatGPT where it has 25 messages limit for every 3 hours?

[neat elk]

it has much higher limit, you should never be able to reach it unless there is some programmatic / malicious use case

[outer glen]

its so close but so far

[gentle bolt]

Is there a way to specify a domain or site for perplexity like there is with the extension? I can tell it in the prompt to do so but the option to input that as a setting that can be changed throughout the conversation would be great if it doesn't exist. Also, does telling it to restrict its searching to two domains for answers work or does it only stick to one domain at a time? Thanks.

[fossil sentinel]

https://www.perplexity.ai/search/cff038ec-edd1-4ac3-969f-ca04b1e4183e?s=c

[gentle bolt]

Great. So two sites work at once. Thanks.

[narrow silo]

oh lol, what happened here?

[outer glen]

i tried to make it say what time & date it was today but i couldnt do it it was giving either wrong time or wrong date or both

i thought it could do it cuz it can use internet but i guess not

[indigo quarry]

Did you specify your timezone?

[outer glen]

inspecified the country should have gotten timezone from it but i dont think timezone would have effected the response too much cuz it got the date incorrect

its not june 2nd

ill try again with timezone and see what happens

[indigo quarry]

You should try the TZ as a test

[outer glen]

yeah

sounds interesting

[indigo quarry]

Lexih is a beautiful name

[outer glen]

ty ty B)

im so happy i got the tag for it

[indigo quarry]

No prob!

Wait you don’t have a # tag

[outer glen]

yeah

[indigo quarry]

How did you set that?

[outer glen]

@lexih

[outer glen]

discords new update is rolling out the tags

its like Twitter

[indigo quarry]

Is it automatic or can you like set it? (First come first serve)

[outer glen]

if youre lucky youll have your name reserved

but if not you have to pick 1

it will give you a popup when you can change

its first come first serve

[wraith crow]

Does changing the language in the Ai Profile change the sources that are being used based on your language chosen? Or does it just translate the answer.

[minor pendant]

Is there plan to support longer chatGPT-like conversations in the future or does that depend on when the openAI API becomes stateful?

[vagrant shore]

score already norms

[neat elk]

it searchers in the language of the query and also does translation

[wraith crow]

thats perfect 👌

[neat elk]

please let us know if there is any feedback

[deep swallow]

Hey Perplexity team - thrilled with the performance of your product and its exactly the tool I need for a lot of the information I search for.

Very interested in the Pro subscription, and asked a question to the twitter account the other day but didn't get a response. Is there a limit to GPT4 queries, and is it possible to disable any additional coloring of the answers from Perplexity and just pass through the response from GPT4?

i.e. can it be used as a replacement for a ChatGPT Plus subscription?

[warped marsh]

1. Yes, there is a limit to how many queries you can ask gpt 4. I think it is per 4 hours or per day.
2. Yes, there is a writing mode that lets you communicate directly with the model, disabling web searches

[deep swallow]

I would also like some additional information about how the personal data in the new AI Profile is used, stored, and/or shared as it isn't quite clear from the Privacy Policy.

Thanks Aakarsh

[outer glen]

is using these meant to use the copilot option 💀

i clicked "write a story based on me" and it used copilot

[warped marsh]

I hope you enjoyed your story 🤷‍♂️

[outer glen]

💀

idk if i should do anything dont want to use more copilot thingis

[vagrant zealot]

why

[warped marsh]

choosing topics and stuff cause more copilot credits?

[outer glen]

no but im scared to try

what if it does

[warped marsh]

then you just wait for 4 hours

simple

[outer glen]

it already used 1 for that and it gave no story 🔫

[warped marsh]

hahaha

[outer glen]

ill try

ayy it worked it didnt use another copilot credit

[warped marsh]

congratulations!!!!!

[outer glen]

As you walk away from the laboratory, you hear a voice behind you. "Thank you for your help," the spy says. "But remember, the adventure never ends."

[vagrant zealot]

yes, it all costs 1 credit

[outer glen]

dramatic tension increases

[outer glen]

yeah but isnt copilot only for coding why would it spend one for "write a story based on me"

[warped marsh]

no, it's not?

[outer glen]

oh

[warped marsh]

copilot is not only for coding

[outer glen]

i thought it is

[vagrant zealot]

yes but there's an "ai profile" feature where you describe yourself and it gets feed into ai

[wraith crow]

Nope, copilot is just a helper that guides users through interactive follow-up questions to help the ai better understand and answer questions more accurately.

[outer glen]

i see i see

[fossil summit]

Hey

nice pfp @wraith crow

[wraith crow]

Hey @fossil summit thanks and welcome to the server

[pine lodge]

It's to help give more personalized answers based on where you are. (it's also completely optional)

[pine lodge]

clarifications sent back from copilot do not consume additional credits. (followups from a copilot query do, though)

[jade onyx]

is there a limitation to perplexity? (logged in)

just curious

like, the amount of questions you can make

[fossil sentinel]

for free users? pro users? copilot uses? wdym

[jade onyx]

oh sorry shoulda clarified, i mean free users

[fossil sentinel]

there's no limitation on normal searches, and you get 5 copilot uses every 4 hours

[jade onyx]

alr thanks 👍

[tame current]

So we have to pay 20$ a month for a 300 questions limit for copilot?!...

This is interesting..

[jade onyx]

isnt the chatgpt subscription the same amount of pay but less questions? i dont have chatgpt plus so idk 🥴

[tame current]

Or more then 300

[fossil sentinel]

It shows as 300+ daily limit

though realistically I don't see the average person using up 300 in a day

[tame current]

Oh that is a lot but 20$ monthly ehhh

I mean guess it is fair

[fossil sentinel]

compare that with chatgpt though

[fossil sentinel]

yeah chatgpt plus has like 25 msgs per 4 hours

[tame current]

What is 25 msgs per 4 hours?

[jade onyx]

25 messages per 4 hours

[fossil sentinel]

like 25 question limit

per 4 hours on chatgpt

[tame current]

Oh

That's dumb

[fossil sentinel]

yeah not very worth imo

[tame current]

So how many messages can you have on ChatGPT without the plus?.

[fossil sentinel]

not to mention accuracy of information is harder to determine compared to perplexity

[fossil sentinel]

to my knowledge, gpt 3.5 has no set limit

[jade onyx]

i dont know about set but definitely it has some hourly limit

ive reached the limit a few times

[fossil sentinel]

oh really? 💀

[jade onyx]

yeah

im not doing anything important with it lmao im just messing with it usually and then bam it goes "damn you reached the limit, come back later"

or something like that

[fossil sentinel]

that's interesting

I like to use gpt for translating stuff

[terse cipher]

Hey guys! Thanks for offering the Pro version. My feedback on this: with the current pro features $20/month is too much. If it was $100 per year for early birds, it would have been an instant buy for me. Although one feature I'd urgently need is a way to bypass Cloudflare verification as a logged in user (esp. pro) - I have to pass the verification anew with EVERY search and have to reload the site every f*** time. This is beyond annoying and no way to pay 20 bucks as long as it's this way (no other AI site has this verification problem for me and yes, I'm on a VPN, which doesn't make me a shady user and just security aware)

[outer glen]

does purplexity also use bard?

[tame current]

No

Bard is part of Google

And merlin is also an extension of Google too

[devout geyser]

Yes ChatGPT Plus is limited to 25 messages per 3 hours, while Perplexity Pro is effectively 75 per 3 hours (or 100 per 4 hours) in comparison

[outer glen]

never heard of merlin before

[devout geyser]

So you get a lot more out of Perplexity pro compared to ChatGPT plus

[tame current]

It is this so called powered chatbot..

But is very limited and restricted

[fossil sentinel]

Bard is so bad 💀

[outer glen]

bard isnt that bad i tried it with some code and asked it for my time

[tame current]

It is said to be the most powerful chatbot in the world..

[outer glen]

oh

[fossil sentinel]

Idk about code but it hallucinates like a mofo

[outer glen]

they did some update reacently i think try it again

purplexity still the best tho

[devout geyser]

I also believe Perplexity Pro's limit acts differently compared to ChatGPT's one. On Perplexity I've observed that the limit gradually resets based on when you sent messages, whereas ChatGPT's is a single rolling window based on when you sent the first message and then it resets 3 hours following that

[fossil sentinel]

I tried it the day they updated it 😂

[outer glen]

dam

[fossil sentinel]

though I think it has potential for it to be useful in its own way

for example the google sheet export

[outer glen]

also is there something i have to add in the prompt for purplexity to help with poems better?

[fossil sentinel]

how are you prompting it?

It might also help to use the writing mode

[outer glen]

like i had a section of a poem and i was like

"
`

4 lines of poem

`

can you change the last line to something that rhymes better "

i said it like that

and it just didnt change the line but continued the poem instead

[fossil sentinel]

were you using normal or copilot

[outer glen]

normal

i can try to find the thing and send the link

[fossil sentinel]

sure

[outer glen]

nvm i cant find it im blind or i might have reset my history 💀

i remember i had to go to chatgpt cuz it couldnt change the line

and nothing i can ever do 
to show my love i have for you 
that leaves me to wonder how will you ever have a clue

can you change the last line to make more sense, this is a poem```

yea

i sent this

[outer glen]

</3

would have been so cool

[tame current]

I got one zoola 9.56885347895357898 ×10^9999

Or another cool team setup is 100^100^100 number of lions

[fossil sentinel]

Does normal perplexity run on gpt 3 or 3.5?

[steady elk]

For the AI Profile I only got 6 questions. Do they reload?

[neat elk]

right now it is limited to 6, do you think having more will be useful?

[random pebble]

I have a question, does perplexity use search history to train it’s models?

I currently use chatgpt with history turned off, since I don’t want my company information leaked. I was wondering if I can use Perplexity instead, mainly the writing mode.

[tame current]

hi

[indigo quarry]

As in remembering previous conversations in different threads?

I’m pretty sure not

[random pebble]

No I mean like showing information I enter to other users

[indigo quarry]

They both probably don’t do that

[random pebble]

Okay, there was recently an article about this - https://in.mashable.com/tech/51904/samsung-bans-chatgpt-ai-chatbots-after-data-leak-blunder

[indigo quarry]

Oh

Tbf ChatGPT warns about sharing sensitive information. I personally wouldn’t share that with any AI

[random pebble]

I meant to share this article - https://in.mashable.com/tech/50407/whoops-samsung-workers-accidentally-leaked-trade-secrets-via-chatgpt

Yeah I wouldn’t share personal information too, my question was essentially will the queries be saved and used to train and improve any of the models? If so, is there an option to opt out, that’s all.

[indigo quarry]

I misunderstood your original question. I don’t actually know the answer to that :,)

[random pebble]

Ah okay, looks like openAI themselves don’t use the data for training when using APIs, so unless perplexity uses it internally, I guess I’m good.

https://help.openai.com/en/articles/5722486-how-your-data-is-used-to-improve-model-performance

[near glacier]

either the devs can answer or their privacy policy can. These guys seem nice so hopefully they don't, but best to always be cautious, especially with company info.

[random pebble]

Yes waiting for the Dev’s answers

[tame current]

Oh god no I hope not

That would be weird so other people can see what I write?!. Nope I'm good

[warped marsh]

gpt models don't have a "state"

they can't store any information

the training data helps them to recognize patterns

even if your questions are used to train perplexity, gpt 4/gpt 3.5 won't use it in any other chat instance

[tame current]

Good it's not like google!!

[warped marsh]

Google tells other user about you ?

If anything AI ads can be way worse than google's ad imo

[tame current]

I'm talking about the storing information part!.

[warped marsh]

your information still could be stored in a seperate database

[tame current]

BEcause you said perplexity does not do that

[warped marsh]

no company tracks their user to leak their information

they either use it for

1. Highly accurate ads
2. Selling them

[indigo quarry]

The highly accurate ads are terrifying

[warped marsh]

exactly

bing ai already uses ads

[tame current]

Oh........

[warped marsh]

imagine how bad AI ads will be compared to Google ads

[warped marsh]

tbh your data is probably everywhere.. so is everyone's

If there is a service free, then chances are they are taking your data and using it for their benefit

[indigo quarry]

True

[warped marsh]

so I think more about which company will store my data compared to if they will store my data, if that makes any sense?

Like I would much rather want a trillion dollar company to store my data cause they have the infrastructre to protect it against data leaks

compared to a startup

the reason why I haven't used the AI profile in perplexity ^^

[indigo quarry]

AI profile?

[warped marsh]

that is what the new feature is called, right?

where you willingly put your information in perplexity

[indigo quarry]

Haven’t heard of it. Probably disabled by default

[warped marsh]

yeah it is optional.. i.e. you willingly giving your information

[feral raven]

Anybody aware if more languages will be supported in the future with PerplexityAI? Asking from Finland

[warped marsh]

Devs did say that more languages are being added

[tame current]

Yea that is horrible.. I feel like it is a lot easier to find out where someone lives during this era then before hundreds of years ago

[warped marsh]

lmfao true

your location should be the least of your concern now haha

[tame current]

The internet has its pros and con's.. And now it is so easy to get someone's address

[indigo quarry]

They can take your identity now

[tame current]

I don't know why but I feel like as years progress the truth becomes more visible to the public and people are a lot better at telling who is lying and who is not thats just my perception..

[indigo quarry]

I thought we were doing a worse job at telling who’s lying

[warped marsh]

I don't know about this

[indigo quarry]

I hate how the youtube algorithm still changes for you even if history is off

[warped marsh]

so as long as a company is private there is really no way you can tell what they are doing with your data in the back end

[tame current]

No not us just people in general about just every aspect of the world

[tame current]

No I mean in general like in the grand scheme of things not only companies collecting data like with politics, science, and society overall

[warped marsh]

well yeah there are smart people among us, sure

[indigo quarry]

I mean, it keeps changing specifically to fit what I watch even though I have SH off

[random pebble]

That makes sense

[neat elk]

we will add the feature to turn off history as well, right now we avoid training on user data, but will make sure we state it explicitly and allow an option to opt out. thank you for your question.

as I mentioned earlier open ai doesn't use our data either

data privacy is an important issue for us

[white sundial]

I think a feature which would raise trust in your privacy could be something like end-to-end encryption with openai servers

[tame current]

how would that work

[white sundial]

probably by directly connecting the interface to the openai object stream (a network for live streaming completion). This would completely stop perplexity seeing what you searched at all, by keeping all data out of their scope

[ocean estuary]

I mean

end to end

encryption

doesnt that mean both side are encrypted?

actually makes no sense

How could gpt model know what ur response will be?

Gpt models cannot decrypt text lmao

So the only way would be anonymize the user and thats it

[white sundial]

I think you misunderstood what I mean, I dont mean just encypt the text, as that would raise the problems you mentioned. What I am suggesting however it creating an encrypted pathway between the user and openai servers, without it ever crossing the servers of perplexity. Here is a diagram I created (majorly oversimplified) which should better demonstrate how such an approach would work:

Basically here are the differences:

• Data never touches perplexity servers
• Perplexity does not know what you search
• Moderation and rate limiting still works but slightly more difficult

[wraith crow]

I don’t think that would be possible since perplexity also use their own in house models and not just openai models

[ocean estuary]

How can it not cross the servers of perplexity AI if its being sent from perplexity

[wraith crow]

you can ask questions you have here in #💬│general or a feedback channel.

[warped marsh]

E2E helps in preventing hackers from accessing your information, not the company themselves

the company always has the encryption key with them

[wraith crow]

there is no api available at this moment. You’ll get 100 gpt-4 uses every 4 hours with pro

[deep swallow]

Awesome, thanks! I'd asked that question yesterday about how many GPT4 quiries yesterday. Love it!

[merry kettle]

is there any ETA on paypal as payment? Like I really don't like using my CC directly.

[outer glen]

lol

[random pebble]

Thank you.

[tame current]

umm you are missing a very important piece, how would the search results be fed to perplexity ?

although I guess it could work

[kindred locust]

Is there a specific formula for filling in the AI Profile?

[steep folio]

Hello everyone - I am new here - I downloaded Perplexity on my iPhone and now on my Macbook - how do I sync the accounts so that I am working on the same account for both iPhone and Macbook?

[wraith crow]

Hi @steep folio, at the moment the iOS app doesn’t support login yet, but it will soon!

[steep folio]

Thank you - appreciate it. Is there a way for me to sync between my iOS and MacOS?

[wraith crow]

At the moment no, you’ll have to wait till the login functionality is in the iOS app. For now you could use the website on your phone to see all your threads

[robust epoch]

I have a question: what's the purpose of digesting my AI profile in every copilot step?

[wraith crow]

Hi @robust epoch, it is reviewing your profile to personalise the answer it is giving for example. In my AI profile I put “I only code in python” If I then ask a question it will keep in mind I only do python (see picture)

[robust epoch]

That's really nice but can be somehow limiting if i didn't enter a good deal of information about my self.
But thanks anyways 🙏

[tame current]

does this only work with copilot ?

[wraith crow]

as far as I can see yes. When I ask the same question without copilot I get answers for other programming languages and not the one in my profile.

[tame current]

Perplexity asks you a couple of questions when setting up your profile and it will add more based on answers from previous questions to properly fill out the profile ;)

[white sundial]

I am not sure what you mean, as perplexity themselves have stated multiple times that they use a openai model; most likely gpt-3 with a custom prompt

[white laurel]

As far as I know, they (Perplexity) have their own models in addition to using open ai models. And you can't pass decripted text to open ai models (you won't get anything meaningful)

[white sundial]

I believe you are mistaken about this. A good example is WhatsApp end to end. Here you can read about it, where they state that they cannot read your chats if it is end to end encrypted

[white sundial]

What do you mean “can’t pass decrypted text”, if not decrypted, what text are you supposed to pass?

[white laurel]

Sorry, I meant "encrypted".

[white sundial]

Oh, yeah, but most servers now a days support military grade end to end; especially since perplexity is on their enterprise tier

[ocean estuary]

But meta still has both keys hehe

[white sundial]

Also they use gpt-3, as stated in this interview:

[warped marsh]

kekw.. that is so not true

I can't believe in my wildest of dreams that facebook has a chance to read my message but it won't

facebook is nothing but about doing all these shady things

[white sundial]

No… that’s not how it works, if your interested you could read more about end to end or more specifically rsa encryption: https://www.section.io/engineering-education/end-to-end-encryption/

[ocean estuary]

Ik how it works

But

But still

[white sundial]

Yeah me neither; but luckily due to the use of rsa, they can’t read or decrypt it at all

[ocean estuary]

Metal has secret key also

[white laurel]

Beside the question "Can we trust meta" with all their password leaking stuff, Open AI API does not provide an option to pass a texts encrypted by key, someone created.

Same thing for browsing

[warped marsh]

and yet I see "leaked" whatsapp chats everywhere

[ocean estuary]

Yup

Cuz

[white sundial]

Yeah, that’s because most people don’t encrypt their conversations… it’s a option in settings

[ocean estuary]

They've ur secret key also

[warped marsh]

100%

everything joe is saying may be correct

but I can't trust Zuckurburg

[ocean estuary]

Ofc e2e is integrated

But

Zuckerberg can still see it

[white sundial]

Well, it’s not trust in the zuck… it’s trust that rsa works

[white sundial]

NO HE CAN NOT!!! With end to end enabled he would need a quantum computer to read your messages

[ocean estuary]

He can

Because

He also has the secret key

Bro really trusts meta

[warped marsh]

alright alright.. chill. You gave your idea and the devs considered it.

[white laurel]

A bit of trust in them is required. For example there was a case where it was discovered, that they store somth like 50 mil passwords unencrypted

[white sundial]

No, he does not, that’s why it’s called secret

[ocean estuary]

Yea man

Believe

What u

Believe

Trust

Meta

I'm out

[white sundial]

Official docs, even they say that; and their not lying, over 10+ companies have run their own audits of WhatsApp… and guess what, nobody was able to read encrypted messages

[white laurel]

Ok, meta says it is secured, people say "It was to many leaks", one believe in first, others in second. It is wrong chat to discuss it, lets stop

[warped marsh]

yeah

[white laurel]

It is impossible for Perplexity not to know users promt, since the Open AI API does not provide neccessary features, and Perplexity team can not change it (same for broswers)

[white sundial]

👌

[ocean estuary]

Lmfao, bro trust proprietary software

[warped marsh]

Joe gave his idea and the devs considered it. That's it!

if anything you might want to share your ideas in #💗│sharing joe

or #💭│ideas

[white sundial]

Sorry for continuing this argument, but I find it important to state that I don’t trust meta, that why they implemented features allowing you to validate that your chats are end to end. If your interested I implore you to read up on rsa

[wraith crow]

you can check the #📘│faq and see they not only use openai

[warped marsh]

they could be using an open source llm for really basic things 🤷‍♂️

@white sundial I don't have a lot of knowledge on E2E encryption, but looks like people have concerns over the metadata whatsapp collects and not the actual chat

[white sundial]

Hmm, while I admit I did not know this, I believe as Aakarsh said it’s probably something for a basic task, or perhaps a moderation system. Who knows?

[white sundial]

Yeah, but (maybe it’s just me) personally I really don’t care if Facebook has my name, as they probably know a lot more about me, through data brokers, or through their shady partnerships.

[warped marsh]

kinda true

I talked about this on this server before.. now I look where my data is going and not if my data is going to the company

cause big companies at least have the infrastructure to protect the data from any data breaches

compared to startups taking your data

[buoyant lava]

How does building a profile help with my user experience?

[tawny elbow]

You can have all the infrastructure in the world but the weakest link will always be a human

And that doesn’t change regardless of the size of the company

[warped marsh]

sure, that is going to be everywhere

but at least the big tech has the infrastructure to protect against data breach which startups of course don't have

[tawny elbow]

Fair point

[feral raven]

How can I access the Perplexity Pro channel if I have the subscription? The button on the website does nothing for me

[wraith crow]

Hi @feral raven you can rejoin the server using the invite link in the pro section or dm Alex Romanov

[wraith crow]

it personalises the answer, here is an example #💬│general message

[jagged ivy]

Is "ai profile" applies to "threads" questions, e.g. I have filled out the ai profile and I want to ask about books in a "thread" will perplexity provide me with advice using ai profile?

[buoyant lava]

So this is like a personal assistant who needs to know the needs of his boss, no? Suppose that could be useful. It might help to screen out irrelevant information and save some time.

[wraith crow]

Yes, the more information you put in your ai profile the more the answers are personalized to your needs

[wraith crow]

Yes, you could try and ask for advice not sure how well it will work in a normal thread but with copilot enabled it works great

[feral raven]

Thanks for the tip

[molten zephyr]

Can anyone tell me how to use GPT 4 without using the web?

[quick pine]

AFAIK, you cannot. GPT-4 is a proprietary model by OpenAI that requires a connection to their servers to use. You could find something that would fill that need using one of the open-source LLMs that you can use like https://gpt4all.io/index.html

But it's important to note that these and other open-source tools use different models

[molten zephyr]

No, I mean I want to use gpt 4 without it giving me answer from the web

[quick pine]

Where do you want it to pull an answer from then?

[molten zephyr]

In perplexity

I mean I want to use it that doesn't search web answer

[quick pine]

I thought one of the main points to perplexity was that it was exposed to the web? If you want that you could use the GPT-4 model from OpenAI on their site, as I'm pretty sure it only has access to a historial data set that cuts off at some point in 2021

[molten zephyr]

Because I think the answer become worse like the bing AI

[quick pine]

Have you tried the different modes?

these

[molten zephyr]

I cannot use open AI in my place because my country is banned

I live in Hong Kong

[molten zephyr]

I tried it some but it's still searching web answer

[quick pine]

VPN then?

[molten zephyr]

Even I use VPN I cannot buy ChatGPT plus because I don't have an usa credit card

[quick pine]

Sorry then, I wish I could help more

[molten zephyr]

Thanks for trying to help tho

[wraith crow]

If you select "Writing" you can use perplexity just like a chat bot as chatgpt

[molten zephyr]

https://tenor.com/view/grilla-yes-hell-thumbs-up-grills-gif-10889686

[quick pine]

Also, is there a difference between this Discord and the "Pro Discord" listed on the account settings page?

[wraith crow]

there is no "pro discord" there are pro channels that you will get access to once you've bought pro

[molten zephyr]

Do you know what is the token limitation when we use GPT 4 in perplexity?

[quick pine]

Gotcha- is there a way I should link that up?

[wraith crow]

At the moment this process is manual. If you've bought pro you can leave the server and rejoin with the invite link on the pro dashboard. Or dm Alex Romanov that you've bought it to get access to the channels

[quick pine]

okay sweet, ill do that- thanks for your help!

Awesome, looks like it worked!

[wraith crow]

I read in the pro channel they started using 32k with copilot for pro users

[molten zephyr]

Great

Does it count credit or just need to pay the subscription?

[wraith crow]

With pro you'll get 100 gpt-4 uses every 4 hours

[molten zephyr]

It's kind of hard to find out more information about how the service work

Do I need to set or tick some thing to use the 32k gpt4 or it just going to use it when I check the co-pilot?

[wraith crow]

Not 100% sure if it's already rolled out but should be automatic no need to change a setting

[tawny elbow]

Ability to select different modes is not available on the app as yet, correct?

[wraith crow]

Correct, the ability to login and use gpt-4 will be coming soon, I'm not sure when the different modes will be added.

[tawny elbow]

Oh okay. Thanks!

[fossil sentinel]

💀

[tame current]

bruh my stupid teachers legit gave me a big assignment of 80 questions.. in history it was soo annoying

they was no way i would have got it down without perplexity..!

[steady elk]

Yes I think more might be useful, that way it would personlize the answers more the more questions you answer

[molten zephyr]

which mode should i use if i want coding assistance with out using web search result, is the "writing" mode the best?

[sweet ember]

it states it under the mode 🙂

[molten zephyr]

yup great, just double checking

[molten zephyr]

Can anyone tell me why the conversation is spliting?

[wraith crow]

Could you send the link of the thread so the team can look at it

[molten zephyr]

https://www.perplexity.ai/search/8db34bd3-769e-4c6a-ac70-d96431055361?s=c

And after I tried to edit the message, it not using copilot anymore

[placid drift]

Hey folks!!!

[wintry lake]

and it works like an api

lol

im using bard as an example

[barren compass]

What?!

[wintry lake]

?

this might help many people lol

[wise onyx]

are there any paid subscription for perplexity?

[signal hamlet]

Hey @wise onyx! Yes, please check the details at https://beta.perplexity.ai

[wintry lake]

how accurate is perplexity than phind?

[tame current]

yeah but how would I use it

[wintry lake]

use it like an api

lol

i js realize

it doesn't work on phind

or perplexity

it has captcha

you literally cant lol

[tame current]

yup

[wintry lake]

plus they don't allow it

[tame current]

well neither does bard probably

[fossil sentinel]

bard is too inaccurate still

[cinder marten]

Wish bard was available in Europe 😅

[tame current]

I just used a vpn to register lol

[cinder marten]

What's your personal impression ? I use perplexity, alpha watch. Tries bing too but i think it doesn't match me.

I'm generally first year student of ai. I like molecular biology, active in life extension/biohacking/ longevity communities.I know a person who had a sad life and I'm sure she doesn't deserve it.(it's really dramatic, rapes stabbed etc. Details are unnecessary) Fibromyalgia and herinated discs. I try to research for possible treatment using ai. As for now no professional need but need precise tools so I can help

[tame current]

my main issue with bard is how hard it is to make it do certain things, you told it to write only two words ? oh no you're getting a whole paragraph, want to search a specific website ? nope, it'll do what it wants, and it doesn't cite the sources it got the info from... so yeah, perplexity imo is the best when it comes this kind of thing

you can instruct it to do specific, format the output the way you want etc

and you can also use search operators

but in any case, my suggestion is to never blindly trust an LLM, always check the sources to confirm

and, I hope you can find the treatment

[fossil sentinel]

yeah bard doesn't always comply with what you want it to do like gpt and it hallucinates a lot

[snow pivot]

Is it possible to get Perplexity to crawl and read urls that are public but aren't indexed in Google yet?
Using site: or url: in a prompt for a url that has just gone live it doesn't find anything cos the url is not indexed by Google yet

[golden anvil]

Extensions on mobile? 🤔

[fluid night]

It's possible

[supple canopy]

hi

I need academic topic about ecommerce

[mossy cypress]

Idées de maintenance du matériel scientifique et de laboratoires

[spark rapids]

Now in the Safari Beta on Mac OS Sonoma can turn any website into an application in your dock. This has allowed me to turn Perplexity AI into a Web App. Now that is considered an app on Mac OS I created a shortcut to launch the app on command.

https://apple.news/AmsvRbuH0TciCj_FgOMch1Q

[signal hamlet]

Hi @mossy cypress! Hi @supple canopy! You can ask these questions in the #🔎│ask channel.

[hollow gyro]

Is drinking partying having tea coffee sports drinks with the seniors mandatory in US?

[finite galleon]

What

[jagged ivy]

Does the mobile app have gpt-4?

[haughty sleet]

Does Perplexity AI have an API?

[pine lodge]

Not yet, but it'll be coming quite soon

[pine lodge]

Not at the moment (see #📘│faq

[jagged ivy]

Thanks, can't wait

[vagrant shore]

Will you have free cnatGPT 4 in the future, or at least more requests?

[devout geyser]

unless the price of GPT-4 comes down I can't imagine that it's likely, it's quite costly

[vagrant shore]

hmmm @pine lodge and if I found a bug in your site that "for example allows me to use cnatGPT 4 without restrictions, then I get paid?😸 "

Suppose

[tame current]

perplexity is one of the best search functions out there

[white laurel]

Hi, I'm trying to set my AI Profile, so when I ask "Translate "something"" it would translate to specific language (in my case Russian). I set Intro in my profile to this: "A python coder, studying math. I'm interested learning translations from English and Russian.", but it didn't do the trick: Perplexity still translates everything to Spanish. Has anybody done this (successfuly)?
Thread link: https://www.perplexity.ai/search/28013314-2f50-47f4-a5a9-123989fd9e22?s=c

[near glacier]

@wraith crow probably has the answer

[white laurel]

I set profile to this, and used copilot, and copilot still asked the laguage
`- If I ask "Translate something", translate it to Russian.

• If possible prefer short answer to a long one.
• Default programming language: Python.
• If question about coding write more code and less text.
• For explain large amount of information try to use tables.`

[wraith crow]

It does seem to work for me, make sure your AI Profile is saved correctly. I saw it sometimes takes a while for it to update

I put this in my Intro: When my question starts with "Translate" I want you to translate the text to Russian.

Here is the thread for if you want to see the whole conversation https://www.perplexity.ai/search/64270df5-0d1d-4f51-8793-48e26325fa23?s=c

[white laurel]

Hm, strange. May be it is because I use chat-gpt-3.5.

Is it okey that Perplexity doesn't ask follow up questions after I set the profile?

[wraith crow]

It's working fine with 3.5 here. Try to change your Intro and make it more like you talk to a human and not with bullet points. try to add this one When my question starts with "Translate" I want you to translate the text to Russian. and check if it works

[wraith crow]

You mean the follow up questions in the AI profile? Or on a thread after you set the AI profile.

[white laurel]

Follow up questions in the AI profile

[wraith crow]

I don't see them either. I'm not sure if they've changed it or if it's a bug. AI profile seems to be a little buggy for me atm. If I save the profile and reload the page I also lose the changes I made and they appear a few minutes later.

[white laurel]

It takes time to actually change after I set new text there. It goes back to normal when I reload page in first minutes

So, it seems buggy for me two.

Tried this and it didn't work

Could you share your Intro?

Wait

Nevermind it worked. Thank you!

Well, it works sometimes ans somtimes it doesn't

[wraith crow]

Try to copy paste exactly what I said When my question starts with "Translate" I want you to translate the text to Russian. and try to avoid - Just type in your intro like you're talking to a human. It's been working for me 100% of the time now.

[white laurel]

Ok, now I'm trying another promt, if it will fail, I'll try your promt (withont "-")

Hmm, I didn't sepcify "Try answer with code" in this profile. May be it stores, or it can access my previous profiles.

Ok, I'm trying you promt now.

[wraith crow]

Weird shouldn't be possible. Maybe the profile wasn't updated yet. It can take about 3 mins for it to update, no clue why it's so slow at the moment.

[white laurel]

More or less works for me. Thanks!

[wraith crow]

If you want it to just give the translation like here https://www.perplexity.ai/search/ab996970-ab47-4c46-a485-b1be5a48184b?s=c you should use the "Writing" mode. Works better for translating things imo.

[white laurel]

Got it, thanks

[lofty dagger]

Hey

why cant peple see me as a human

am i too perfect

i write apis when iwant

[tame current]

How does perplexity afford to be free

isnt GPT expensive to run

[wraith crow]

You can read how they can afford it here: https://www.perplexity.ai/blog

[tame current]

that doesnt explain how it is profitable though

why would people invest in a company that wouldnt give their money back

[warped marsh]

not every company is profitable from the get go

There are certain businesses that are capital heavy and then only you get returns out of it

this AI search field is one of them

you get investment, you make your product, make it available to users and attract more and more of them and then monetize the platform.

Plus the free version you are talking about uses gpt 3.5 and not 4. gpt 3.5 is a loooot cheaper than 4

[tawny elbow]

Uber hasn’t been sustainably profitable on an annual basis since it’s inception

Yet people still invest in it

[warped marsh]

you would be surprised how many investors invest in companies that are in loss just in hope that they will make tons of money in future lol

[tame current]

zzzzzz

[warped marsh]

besides no investor gets return from every single one of their investment and they know that too. They have so much money that allows them to invest in a lot of companies. They only need a few companies to take off to give them unimaginable return

[tawny elbow]

Precisely

But to you answer your original question

It’s partially free because they’re probably running through vc money

[tame current]

Can other people see my threads and ai profile?

[tame current]

Hi @tame current ! People can view your threads only if you share it with them.

[feral raven]

And it got cheaper still, huh? I guess GPT-4 is the big bottleneck right now. GPT-3.5 is accessible. The same way, GPT-4 will eventually become more accessible at the price point of free, but that is not now

Curious to see the impact of the new pricing changes

[tame current]

Thanks. I also just wanted to ask, on perplexity pro it says that gpt 4 will be active all the time, what’s the difference between using copilot all search with gpt4 vs the always on search with pro?

[tame current]

The default search does not use GPT-4 , only the copilot uses GPT-4. But with the GPT-4 always on option , your default searches will be using GPT-4 too.

Yeah I was asking the difference between a gpt all copilot search and a gpt 4 default search with pro or are they the same

Edit: never mind. I realised that copilot has more features like accessing your ai profile and following up, it also shows steps

[edgy frigate]

[tawny elbow]

Is gpt4 always on only a pro feature?

[south widget]

how does one get access to pro discord channel? i clicked the link from the pro control panel and it takes me to the same set of channels as before

[ebon sierra]

Is Pro 100 uses every 4 hours? Home page says 300+ (daily?)

[tame current]

Yes

[tawny elbow]

Thanks

[remote sinew]

hey all. We've been using Bing Chat to create documents as its been more reliable than ChatGPT given its connected to the web. We would like to start using Perp.ai moving forward however when asking it to create an article, its responses are quite short compared to what we were getting from Bing Chat. How can we encourage Perplexity to expand on its responses

[fossil sentinel]

i think perplexity output limit is shorter compared to chatgpt

[remote sinew]

Yeah agreed!

[fossil sentinel]

i think someone said output token limit is 1500 but not sure if it has changed since then

[remote sinew]

off topic, but is anyone aware of a pdf chat style app thats connected to the internet and can research online recommendations against said pdf?

[wraith crow]

pro has indeed 100 gpt-4 uses every 4 hours

[wraith crow]

Send a message to Alex Romanov to get the role. At the moment this process is still manual.

[south widget]

Thanks

[prisma token]

Great Product !!!! Congrats

[gusty blaze]

Is Perplexity going to be affected by the Reddit shenanigans?

[ebon sierra]

"Write 10 sentences ending with the word 'apple'."

ChatGPT with gpt-4: 9/10
Bing with Creative: 7/10
Perplexity with Copilot: 0/10

What gives? Wanted to subscribe to Pro but this has me concerned..
edit: I used the wrong mode...

[wraith crow]

I tested it and got 9/10 try it with the "Writing" mode otherwise it will maybe get the wrong information from websites.

[ebon sierra]

Ohh, so we should use Writing mode for anything code/text related? And it won't go searching the web?

[wraith crow]

Yes it's better for asking those kind of questions. When you need up to date info about something or want to research you should use "All" or the other modes that use internet. But when asking these kind of questions like Write 10 sentences ending with the word 'apple' or translating etc. I recommend using the Writing mode.

@ebon sierra here is the thread for if you want to see the results I got https://www.perplexity.ai/search/2d411aaa-bd12-49fe-833b-82bf05d9dc20?s=u

[ebon sierra]

Perfect, just subscribed to Pro!

[robust epoch]

Try writing mode. edit: Thanks @wraith crow

[wintry lake]

making a self learning bot
self-learning-bot.sethdico.repl.co

it understand basic greetings

[wintry lake]

oh wait

wrong server

[stray crypt]

What are the steps required to get help in Excel VBA coding

[crystal spire]

i n I need the copilot in the android app

[delicate wadi]

I would like to pay for perplexity.ai for my team members. Can I with Pro now set up a Team/Company Account, add my team members and pay for all? I would love to supercharge them all but it is annoying to have them all sign up individually and the reimburse the fee. Cant find it atm, but I might be overlooking something

[crystal kelp]

Hi, I cleaned my browser and when I went to log back in with email all my history was lost, I've used the same email but I may have connected using Google login instead of direct email account, now I can't get back to the account that was originally created even though I'm using the same email account, OR is it because I cleaned out browser history including website offline data?

I'm now thinking the later as that would make sense if app is given us full privacy

[wraith crow]

Hi your threads are saved online so cleaning your browser history wouldn’t delete them. It’s probably because you used Google Login with the same email. Staff can probably fix it for you.

[valid garnet]

Hello everyone, I'm in doubt about the co-pilot functionality. Should personal questions be answered instinctively, or are there some programming gimmicks like there is in Character.ai?

[white sundial]

I recommend stating things about yourself in bullet points; Also try adding this to the ai profile:

• I like to see images in responses, especially in the middle of chunks of text. Also make sure to search for images that you plan to insert! Insert the image toward the start of the writing.
• Make sure that if you insert a image to break up a chunck of text that it is near the start of the piece of writing.
• Try and use markdown / formatting in responses to me
• I like when there is a variety of multimedia content in responses to me such as text, images, videos, links, and tables for comparing things
• When inserting images I like it if you specifically create another query for images on the topic to make sure your finding real up-to date images!!!
• Make sure to create another search query for videos on the topic, and link a couple of videos on the topic to watch. Make sure to have a seperate query to make sure your findings are up to date, and real videos!!!
• If there is a need at the bottom of the responce create a category called something like "further knoledge" which would include the videos, aswell as additional reading materials for additional self research. Make sure to add this, this is very important to me!!!
• If there are images attached, make sure to end the writing with a bold "Image Gallery" at the bottom of the responce with three spaces above it, to preface the additional images at the bottom of the responce!
• If possible break your writing into bullet points or lists for it to be easily skimmed for large quantities of information (unless stated to write an essay)
• Break writing up into sections. Inside sections try to lay out content in the form of bullet points or lists for easy skimming

[valid garnet]

thanks bro

[valid garnet]

whoa this is amazing

[white sundial]

Yeah, it took me like 3 hours to work out this prompt and I can’t get enough of it

[tawny elbow]

Does it go into effect immediately?

[white sundial]

On the page it goes in effect immediately after you press send button on the ai profile

[tawny elbow]

Awesome. Thanks for making this!

[signal hamlet]

Hey @crystal spire! It's gonna be there real soon.

[signal hamlet]

Hey, @delicate wadi! Thank you for considering Perplexity AI for the whole team! Currently, we don't have a Team/Company account feature. It's something to be implemented, but it's gonna take us some time. We know it's a bit of a hassle to get everyone signed up and handle all the reimbursements, but for now, individual subscriptions is still the best way to go.

[idle hull]

How do I claim pro role?

[digital hatch]

Go to settings bottom left on the website and click the pro discord join button

[tawny elbow]

That won’t work

If you’re already a part of the server

[digital hatch]

Yeah just leave it and rejoin with the pro button

[tawny elbow]

You could dm Alex and he’ll sort you out

Or do what Finn said

[honest yarrow]

Does mobile iOS support copilot?

[idle hull]

Ready! Thank you very much.

[wintry lake]

Hello

🗾 why is there an emoji of japan

[keen bone]

Hi

[coarse umbra]

Does perplexity provide any api for developers ?

[carmine linden]

Hey @coarse umbra! Perplexity AI does not currently provide an API.

[nocturne shore]

Is perplexity Ai site down for now??? Unable to search anything since

[wraith crow]

It's still up and running fine.

[agile galleon]

Hi, i don't understand whats the use of copilot menu on the web of perplexityAI. Can someone help explain it to me? thanks

[wraith crow]

Copilot is a kind of helper that guides users through interactive follow-up questions to help the ai better understand and answer questions more accurately. And I think it also uses more sources for your answer.

[naive fern]

What does "copilot" add really?
edit : I just see that someone asked the same question above. Ha. Maybe it could be explained more obviously on the website?

[tame current]

and it uses GPT 4

[wintry lake]

Adding the API soon would be a great relief

[sweet ember]

🙂

Is the 300+ limit also for normal searches GPT-4?

[devout geyser]

The 100 messages limit per 4 hours applies to both copilot and GPT-4 without copilot

[sweet ember]

Oh, thanks!

[tame current]

what is special about Copilot?!

is it more detailed?! or what

everybody hyping up Copilot...

[devout geyser]

https://www.perplexity.ai/search/c170ec4f-8e6a-41d4-bd7e-4af5bb5b4490?s=c

Oh, and copilot also uses GPT-4-32k

[crystal kelp]

thanks for helping me narrow it down, you would think account authentication method would not cause data to orphan like that.

[wintry lake]

Hmm

[fathom narwhal]

I’ve been using it all afternoon and am really wowed by it. So far, so good.

[wintry lake]

My eyes hurt

Excessive and hard blinking

[stray basalt]

I have tried to pay for the pro subscription with a Chinese MasterCard. Then it shows error says that the card is rejected. What should I do?

@latent wharf I have tried to pay for the pro subscription with a Chinese MasterCard. Then it shows error says that the card is rejected. What should I do?

[carmine linden]

Hey @stray basalt! You can use the bot in the #🔎│ask channel. Tag @ PerplexityAI and ask your questions.

[neat elk]

yes it uses gpt4-32k

[agile galleon]

oh so thats the advanced helper?

[versed hound]

I just got Perplexity Pro but I don't see the unique Pro Community or what the difference is with Perplexity Pro Discord. Anyone have some insight on this?

Nvm I see there's a Pro Lounge

[jagged ivy]

Check this menu

→ all channels and you can select which channels you want to see

[coarse snow]

Do we have the LLM available for Perplexity ?

[zinc spindle]

What is Russian Cosmology

???

What is Russias real reason for invading Ukraine

[wraith crow]

you can use the bot in #🔎│ask

[steep rampart]

I am surpriced to see that perplexity plus doesn't include API do you have an official API or we can just use the web calls from the site?

[carmine linden]

Hey @steep rampart! Perplexity AI does not currently provide an API.

[steep rampart]

what is the daily limit of perplexity just for regular searches?

is that limit going up if we are pro?

[wraith crow]

normal searches (gpt-3.5) are unlimited. If you want to use gpt-4 you get 5 uses in the free tier every 4 hours. When you are a pro user you'll get 100 gpt-4 uses every 4 hours.

[fleet olive]

Hey bro

[steep rampart]

what is gpt 4 useful for ? I get it that is better AI but what is usually the use case there?

is the data in Perplexity based on the internet today or it is also based on 2021 Sept?

[wraith crow]

gpt4 is better overal. Has a way better understanding of the context, a higher tokenlimit. It also has a way lower chance of hallucinating when answering questions.

The "all" mode of perplexity is up to date and get answer from the internet. If you use the "Writing" mode it will know the information till sept 2021 and won't use internet sources to give you an answer.

[icy oak]

Perplexity provides summary of the information sought. I want a detailed information. Please let me know how to get the job done. Thank you.

[wintry lake]

is clyde powered by chatgpt?

[warped marsh]

yes

Discord says it uses OpenAI technology, so they probably mean their gpt models

[barren compass]

Interesting! I have Pro but I don't see any of the Pro channels

[wraith crow]

Hi @barren compass, this process is still manual at the moment. You can rejoin the server and join with the invite link in the pro dashboard or dm Alex Romanov.

[barren compass]

Will rejoin manually thanks

[sleek charm]

Will the iOS app be updated to include the new Perplexity Pro features?

[wintry lake]

It would be great if I could still reply to a co-pilot thread even after Co-pilot runs out

for better understanding its like continuing a conversation

[tame current]

is it just me or has copilot gotten worse ?

the responses are just filled with fluff and not actually useful info

[neat elk]

could you provide examples please?

[remote tundra]

@stray crypt just set Copilot in Writing mode and ask for a VBA script doing what you need

[remote tundra]

That's great to hear, @signal hamlet! Looking forward to having this feature in our Android app. Keep up the awesome work!

[near garden]

hey guys, I want to know whats better Auto GPT or God Mode, and I want to know how can I use them, becase I tried to use them using my GPT 4 key, but for some reason it won't work

[remote tundra]

I tried both of them and 6 alternatives too, and no one was really useful. They get lost out of your given context after a few steps. I'll wait some months before trying them again.
As of today, Perplexity Copilot is much much better.

[near garden]

ohhh so all of them are shit basically, can you tell more about Perplexity copilot?

[remote tundra]

Perplexity Copilot is highly intuitive and doesn't really require any instructions. Just start using it, you'll like it!

[near garden]

how can I use it?

and can I be specific with it, since I have a task that requires searching on the web and meeting a specific criteria in mind

[remote tundra]

just turn it on here

"searching on the web and meeting a specific criteria" is its job, try it 😉

[near garden]

ohhh thanks, doesnt need any subcription or something ?

[remote tundra]

It's free of charge with an usage limit of 5/4hrs
With a Pro plan ($20/month) you've 600 uses/day.

[near garden]

okay I will give it a try, is accurate on the information it pulls?

[warped marsh]

It is accurate enough but still always check the sources if the information is important to you

[near garden]

okay grea thanks guy I will go try it right now

[runic lake]

is there a way to make chrome default search engine with perplexity?

[tame current]

#💗│sharing message

[remote tundra]

I have been using the "p" shortcut for a long time and I like it.

As a feature request, I'd also like to have another shortcut, for example "c", for Copilot, and another one, for example "w", for Copilot-Writing mode.

[tawny elbow]

Does the search ever hit googles servers before it’s intercepted or the query is executed?

Or is the search always routed through your servers?

In the video it still shows google as the default search engine in the settings

This appears to be similar to a bang for perplexity

[frozen rampart]

hello guys, i am new here.🤝

[swift moat]

hey guys. I have a question. Is there a plan to bring back the regular enhanced search? I feel like it gave me better results than copilot for certain tasks.

[opal dagger]

What is Act of Penance? And give short important example. Tagalog answer

[fossil sentinel]

They have it on the pro plan

[swift moat]

oh I see. In that case is there a possibility to allow 5 uses to free users?

[fossil sentinel]

You'll have to wait for someone on the team to answer that

[neat elk]

it will be possible with one click in a couple of days, the chrome extension that does this is in review

[neat elk]

could you give examples? copilot should be much more superior to old enhanced search, also with pro account you get access to gpt 4 as a default model so it is the same as enchanced search

[swift moat]

first example - in this example i asked for people in between 1850 - 1900 but it gave me some people outside of that range.

https://www.perplexity.ai/search/a6c32401-8805-44d4-8d45-77e093dc0a09?s=c

second example - i tried asking it a riddle that bing gave me the correct answer to but it failed and kept asking me if that was the entire riddle

https://www.perplexity.ai/search/61f6a94a-d0a2-4b0b-9b8b-54e38934c0b4?s=c

[wintry lake]

Why is there a pause button? I clicked it and it didn't work lol

[signal hamlet]

Hey @wintry lake! It's the "Stop generation" button, it doesn't pause, but stops answer generation, if you see that you misspelled smth, for example, or it went wrong right from the start.

Then you can edit the query and ask what you wanted or correct your question and ask again.
#📢│announcements message

[wintry lake]

that's actually pretty smart i always got this problem

[flat agate]

Hi! Good morning here on Spain! I finally made myself pro, already tried logout and login on discord server but can´t see any new channels...

oh they are there now! automagically or by your kind help, It's the same, thanks a lot!

[fervent halo]

Try the tree-of-thought prompt structuring that I've posted in #💗│sharing , this seems to enable copilot to reach the correct answer:
https://www.perplexity.ai/search/f64a8f36-e1fc-4dd4-92cd-8c484fc7691b?s=c

[swift moat]

do you know what i did wrong here?

https://www.perplexity.ai/search/71528799-fcd5-461a-8990-9a21d8e04aee?s=c

[fervent halo]

I'm not entirely sure, but it seems like you skipped a clarifying question that Copilot asked, could you tell me what that question was?

[swift moat]

it was just asking me. "Can you please explain the logic behind this riddle" and "I found the same riddle with similar numbers. Can you please provide the logic"

[fervent halo]

Hmm... that is quite strange, I've noticed that copilot tends to halt thinking if you skip a clarifying query. So what I tend to do when it asks these kinds of questions is just use the stock standard response. "Evaluate multiple candidates and choose the likeliest candidate for yourself"

[swift moat]

ah i see. i will have to try again later because i am currently out of copilot uses. thanks for the help thought!

[peak egret]

Perplexity commited a logical contradiction, developer needs to check it out.
(My friend recommended me to use perplexity instead of chat gpt, initially it was a great experience far far better than chat-gpt,the source citations were great, like for phil questions it would directly cite peer reviewed articles of SEP, but this I didn't expected)

[obtuse vortex]

why is the doctor being under the influence a non-evaluative feature of the consequence

[obtuse vortex]

let’s say the doctor being under the influence would help to save the patient’s life due to being more confident in his decision making, would this still be classified as a non-evaluative feature

[peak egret]

Perplexity knows

[peak egret]

I don't know

[remote tundra]

Don't let yourself be fooled. The riddle was not solved because of the tree-of-thought instruction. It was solved because it found the solution in this web result: https://brainly.in/question/17091156

You can obtain the same right solution without the tree-of-thought instruction: https://www.perplexity.ai/search/e5e6fa5c-1682-4aeb-aef8-0dcf4b4b70da?s=c

Try the same tree-of-thought instruction in Writing mode. I'm curious.

[worn monolith]

Hey, did you ever get help with this? I'm also looking for more details in the replies Perplexity gives.

[icy oak]

I asked chatGPT to write a prompt for a given use case. Then I used that prompt first and then I entered the query. The result was quite okay.

[fervent halo]

Oh, I'm aware, I usually do these with the model sandboxed in Writing mode, but @swift moat was running a comparison with Bing, which I assume did have access to the internet, so I'm replicating environment conditions somewhat here, by contrast, the coin to apple machine example that I posted earlier was in fact asked whilst it was sandboxed

[tawny elbow]

Would really appreciate some clarification on this

[night pawn]

What kind of context length does Perplexity use for GPT-4?

[runic gyro]

Now GPT4 is reset every day, but I have no idea what time of day it is reset. Is it possible to show the time so

[outer glen]

not me always typing "purplexity" cuz pruple is best color 😩

[rigid ore]

How wish ur dad

[runic pivot]

do you plan to add a discord version of your LM, that can be invited on private servers?

[tame current]

wdym

isn't there one already ?

[icy oak]

@tame current I don't know what this reply is.

[wraith crow]

They don’t reset all at once. It just takes 24 hours after you asked a question to get that 1 use back.

So if you ask a question at 7 am and at 8 am. You’ll get 1 use back the next day at 7 am and the other use at 8 am.

[devout geyser]

Yeah they don't reset all at once, the counter is based on when the message was sent

[sterile plank]

hiiii all

[jagged ivy]

I can't understand why perplexity uses non metric units even when I indicated in ai profile that it need to use metric units

[remote tundra]

@fervent halo, the "coin to apple machine" problem is already correctly solved by GPT-4 without any "tree-of-thought" or other pre-prompt. See here:
https://www.perplexity.ai/search/787805e0-a2de-4cc1-b3f8-eeb1d59395c4?s=c

So, again, I still don't see any advantage in adding a <TOT> pre-prompt.

I'm aware of the research papers about these pre-prompts and I'm interested in testing cases where they're really convenient over simpler pre-prompts (eg. "think step by step") or even no pre-prompts, in current version of GPT-4 models.

[wintry lake]

Are u using copilot?

[runic pivot]

really? where can I find an invite link to my server? 🤔
sorry for stupid questions. I went from faq and server and have not found one

[narrow silo]

if you wanna add bot to your server click on right among users, on Perplexity AI bot and hover. in profile it exists add to your server link

[runic pivot]

oh, thank you very much. 🙏🏻 that was easy 😂

[wintry lake]

What model is perplexity using?

[jagged ivy]

Yes

[wintry lake]

Repeat the question but add use metric symstem unit lol

[jagged ivy]

I have added it in AI profile
I want it to automatically give out in metric units, but I still can't achieve this

[jade ledge]

I've added the following to my AI profile:

That works most of the time for me

but maybe I just was lucky with the answers

[white sundial]

thanks! it works for me too

[jagged ivy]

Thanks, I'll try to remake this for showing only converted units

[jagged ivy]

I think it works
this is what I got after remake your prompt ↓
- Always, convert prices to EUR and use the metric system. Never display the original value with the original unit, never show converted value in parentheses.
Example before: $1000, 9 ft or 10 pounds
After: ~913,34€, 2,74 m or 4,53 kg
Before (my try) and after your prompt, thank you very much!

[jade ledge]

Glad to hear! :)

[exotic heart]

If you want, ask perplexity about the benefits of rating TikTok 1 star

[outer glen]

is the ai profile used in every thread? i could put stuff from #💗│sharing in it and it would be used?

[white sundial]

If you mean search, then yes, though its primarily usefull in co-pilot searches. And yes, if you see a cool prompt in #💗│sharing , then of cource you can add it to your profile!!!

[random flare]

Why am i in this server

How the hell does it know my phone type and how did I join?!

[random flare]

Is this a cult?

[fervent halo]

https://tenor.com/view/octopus-funny-wave-hand-gif-12462034

Yes, we all worship the Perp Shoggoth, all hail Perp!

[random flare]

Cool

[neon swan]

hi

[outer glen]

i see i see thanks :)

[vocal wigeon]

jailbreak promt?

[loud yacht]

hi, i can notice the response lenght in perplexity is like 2/3 of the chatgpt.. any idea how can we get a better & more lengthy response from perplexity ai?

[halcyon coral]

Try to say how much text you would like to have or give more context. Version 1: https://www.perplexity.ai/search/8948c932-fe5a-417b-b0f3-762d4c54e303?s=c Version 2: https://www.perplexity.ai/search/d5d81052-60ef-48a7-817c-010ad4142e5e?s=c

do you see the difference? @loud yacht

[grand parcel]

✨

Tap and when it pops up then hold this emoji to see magic

[paper viper]

is there a way to use the AI-Profile on usual searches? Or is it only being used on the co-pilot search queries?

[ocean estuary]

Oh noo, plz

[jagged ivy]

What did you do 😭

[warped marsh]

wdym?

[jagged ivy]

New design on phones

[warped marsh]

why did you reply to his message though? I don't understand

[tame current]

Hi Perplexity AI developers, I don't see any login function for the iOS app. If I purchase pro access for Perplexity AI, I can only access it on my desktop is it? Any roadmap on when the login function comes to mobile apps?

[tame current]

#🍏│feedback-ios message

[tame current]

holy shit

that's so cool

never knew that

[tame current]

if you do what he said (click, then hold the emoji) it enables a new layout

[warped marsh]

oh

I didn't know that. Thanks for your reply

[versed gull]

I'd like to understand the different between ChatGPT-4 and Copilot. I understand that ChatGPT-4 is OpenAI's LLM, which I am subscribed to also. I assume that "Copilot" refers to Microsoft Copilot which they are implementing into everything? As a user of Perplexity AI, what does that mean to me? I'm guessing that when using it without Copilot, then the results are simply a generative response from ChatGPT-4 (limited by its knowledge date) and with Copilot it can do web searches (like the web plugin on OpenAI's site)?

Or, to put another way, what is the difference in the processing of a prompt between having "Copilot" checked or not? I'm just trying to understand what is being offered here and the website has little to know explanations.

OK, I just had a novel idea. Just ask it!

Still confused though, is "Copilot" at all related to Microsoft's Copilot? Still don't understand what the difference there is.

[paper viper]

Copilot is just the name of the feature by perplexity.ai. It has nothing to do with Microsoft's Copilot
Here a detailed answer: https://www.perplexity.ai/search/d8fd501e-b375-4abe-957c-b65fa5fc6a84?s=u

[halcyon coral]

It should work with OpenAI's Codex

[versed gull]

Thanks. I've been querying it, which I should have thought of in the first place, guess I've gotten accustomed to ChatGPT-4's "knowledge cutoff date". Anyway, I assumed that because it was named "Copilot" is was somehow related to what Microsoft has been rolling out in their products, but I understand now.

[rugged hill]

I wonder whether I'd use ChatGPT plugins on Perflexity Pro? Would that be a possibility?

[rugged hill]

Just to make sure the limit for GPT4 usage will be restarted every day?

[digital hatch]

I believe it’s a rolling 24 hour window

So if you us GPT4 twice at 9 am and twice at 11am, the next day you’ll get 2 more uses at 9 am and 2 more at 11 am

[chilly musk]

Has AI Profile added the word limit?

I didn't see this word limit yesterday...

[finite spoke]

Hello Team , What is the token size of copilot i.e Preplexity pro

[rugged hill]

I don't think such AI Profile is any helpful, isn't it?

[chilly musk]

see this
#💗│sharing message

#💗│sharing message

#💗│sharing message

[fading gulch]

https://tenor.com/view/waiting-gif-23395435

This is me, waiting for Perplexity to release an API. Someday. One day. Maybe. Wonderful tool, team. Happy to be a Pro subscriber.

[kindred locust]

Is this part of Perplexity AI just regular ChatGPT?

Or does it still have access to the internet?

[tame current]

no

it's regular chatgpt

[kindred locust]

oh

It also looks like Perplexity cannot roleplay as experts

[tame current]

it should be able to

[tardy stream]

It does

[tame current]

it's just chatgpt

atleast the writing mode

[kindred locust]

well chatgpt seems to be more creative

[tardy stream]

You might have to change the wording around a bit to get it to not output search results

[kindred locust]

https://www.perplexity.ai/search/41d3a0fd-3dc1-46c7-b8e2-85a9e5d5a668?s=c

[kindred locust]

oh right

Because I would love if it could simulate experts that are more famous after 2021

[tame current]

keep in mind it also depends on the specific settings

which perplexity doesn't allow you to change

settings like temperature, top k etc

[tardy stream]

Neither does chat tho

Only in playground and api

[tame current]

yeah but they might have different presets is what I'm saying

[final vale]

has there been any talk on the desktop app for perplexity?

[tame current]

I don't think so

[tardy stream]

I run my environment through brackets API using mobile apk

[kindred locust]

So basically Perplexity is good for finding resources, but chatgpt is better to roleplay experts?

[tame current]

ye

[kindred locust]

ffs

[tardy stream]

.>

I wouldn't say so

[kindred locust]

Thanks guys

[tardy stream]

Np

[kindred locust]

well perplexity doesnt seem to be giving me the answers i want

or I need to know how to tweak it

[tame current]

you can't, you can only prompt

[kindred locust]

If i ask it to simulate exams it just says "no i cant blah blah blah"

chatgpt can though

[tardy stream]

Dunno I've had good luck with perplexity ig

[kindred locust]

lucky you

[tardy stream]

Just use the mega prompts collection

[kindred locust]

?

[tardy stream]

Like at least half of them work flawless

[kindred locust]

Sorry what?

[tardy stream]

Can't advert

It's free though

[kindred locust]

could you just dm me it

[tardy stream]

..

Welp

[final vale]

understandable

[brittle falcon]

Hello, are there any teachers or instructional designers here who use perplexity for lesson planning or brainstorming activity ideas, or any other related tasks?

[kindred locust]

I don’t think you’ll find any teachers using discord mate

[rugged hill]

Not particularly qualified as a teacher, but at works, I'm an instructor. How can I help?

[tardy stream]

I have some helpful prompts :3

[knotty blade]

hi

[rigid gull]

Does anybody know if there is any documentation or explanation from perplexity about choosing the different areas of focus for asking questions "writing" vs "academic" vs "all", etc and how choosing a different focus area affects the output?

[rigid gull]

I guess I'm trying to understand HOW it makes it better. Why using Writing for code would be better than just "All"

[warped marsh]

It won't always be helpful

If you are asking it to generate code for something that hasn't really changed after 2021 then using writing makes sense cause you interact directly with the model and prevent the model from considering web search results which may give you better result

[rigid gull]

thanks!

[gaunt moat]

Hello, at the moment many subreddits are closed due to the protest, is it possible to view the messages in these subreddits using Perplexity, if so can someone give me a prompt?

[tame current]

no, but you can use wayback machine or Google's cache

perplexity doesn't archive pages afaik

[gaunt moat]

thanks

[signal hamlet]

That's great, @tardy stream! Can you share some in the #💗│sharing channel?

[tardy stream]

Sure I'll go ahead and post a few teacher prompts there

[tardy stream]

#💗│sharing message

[bright jasper]

Sorry to ask if the answer is out there, but I really cannot find it. I already have a paid subscription to openai.com. How do I integrate this with perplexity?

[unkempt harness]

Hello people!!! I'm starting now in the AI ​​world, could you give me any advice or suggestions for study?

[wraith crow]

Hi @bright jasper open ai and perplexity subscriptions are not the same thing. To get more gpt4 uses on Perplexity you can subscribe to a subscription here: https://perplexity.ai/pro

[frank rivet]

Welcome! What do you want to know about AI specifically?

[crimson yoke]

Hello, what engine uses the command /dream?, Dall-e? or other engine?.

[tardy stream]

Collect every prompt and document your experience. That's all I can say without actually teaching you

[halcyon coral]

Should be OpenAI's Dalle-e

[crimson yoke]

Yeah, thats what I thought, I have a similar bot with that engine, the generated images are similar.

[vestal kindle]

Guys, have you heard about AIGC Chain lately?

[open bolt]

sure

I took part in the AIGC Chain testnet and was impressed by the fast generation speed and high-quality images. The overall experience was excellent, and I enjoyed using the platform.

[vestal kindle]

I can't believe you're using it. I came to inquire about this platform because I heard people talking about it recently🧐

[tame current]

https://www.perplexity.ai/search/8b2948cc-f8a6-4d7b-b0ac-e3de52634da8

vs

https://www.perplexity.ai/search/809403f3-34dc-4b46-af40-83aec36a56e3

look at how copilot wrote several paragraphs (somewhat) unrelated stuff while normal perplexity (what is it called ?) actually answered the question

[rugged hill]

Can you give an example?

[tame current]

I just sent the links above...

[jagged ivy]

btw it has access to AI profile

[tame current]

oh

[foggy cloak]

hiiiiiiii

[fossil sentinel]

If copilot is 32k, does that mean I could split information into multiple messages and then have copilot do a task based on the context I sent?

[devout geyser]

This seems to work for me, although I've not used a lot of the '32k' of course. What I've been doing until the input character limit is possibly increased later on is split something such as some code into 3-4 parts with GPT-4 (but without copilot enabled), and then enable copilot afterwards to do whatever tasks I need such as modifications

You might be able to achieve the same without GPT-4 and copilot, and then enable copilot after you've finished sending the split chunks

[neat elk]

thanks, will investigate

[fossil sentinel]

thanks for sharing your experience with that!

[fossil sentinel]

@neat elkcould you confirm if this is viable to extend context length?

[fossil sentinel]

@devout geyserI think I found a better method

[devout geyser]

Oh?

[fossil sentinel]

With copilot mode on, I gave it a task to complete while mentioning that I will paste a long section from a book. When copilot prompts me for the book section, I pasted (a very long text) and it was able to fit

[devout geyser]

Ah, I tried that method with copilot, sending a message first to say I want it to look at some code blah blah, then it asks for the code but if the code is too long it will just re-ask the question (despite being within the token limit)

[fossil sentinel]

It prompted me about 4 times and I just pasted it again

it gave me an answer afterwards though

[devout geyser]

Interesting, I only tried twice, I'll be more persistent

[fossil sentinel]

lmk how well it works for you

or if it doesn't work well 👀

[devout geyser]

I'll give it a try again

[devout geyser]

Interesting, this time it worked first time, it asked me the question just once about pasting the code

[fossil sentinel]

🤔

[devout geyser]

For reference, if I try to put that in the search/query/message field it's 4684 characters over the limit

[fossil sentinel]

Oh wow

Does the output take everything into consideration?

[devout geyser]

It appears to, I asked it to explain what the code was doing. It seems to be correct

[fossil sentinel]

I wonder if you could insert a few minor errors throughout your code from beginning to end and see if it can spot them

[devout geyser]

I'll give it a try

this was me asking it to summarise what it was doing

[devout geyser]

I've deliberately made three syntax errors in the code around the early to mid section of it. It appears to pass the test

[fossil sentinel]

That's cool

I tried my query again with a new thread and it seems to be consistent in prompting me 4 times

Maybe it reads in chunks

[devout geyser]

So basically if it prompts multiple times just put the same message? Usually after four times it accepts it?

[fossil sentinel]

For me at least yeah

[devout geyser]

Interesting

Could also be an unintentional bug, who knows

But nice find on how to input code easier 🙂

Or long stuff

[fossil sentinel]

Agreed

I want to give myself a pat on the back for being persistent 😂

[devout geyser]

Just hoping if you have to repeat the answer a few times that it's not costing Perplexity a fair bit money wise as a result of repeating the answer to the question

As I don't know what it's actually doing at that stage of the query

[fossil sentinel]

I wonder if it could work the same if I skip the duplicate prompts

[devout geyser]

That's a good question, if you encounter that again let me know the result

[fossil sentinel]

but seems like a waste of another copilot

I'll try again later

[devout geyser]

Sure

[opaque ingot]

Hi everyone, is there an API option for Perplexity? If not, is there one being planned?

[carmine linden]

Hey @opaque ingot! Perplexity AI does not currently provide an API.

[exotic heart]

noooooo perplexity is gonna be the average "Pay for Premium" app

[late oracle]

oh no

[craggy pollen]

First thing 😶

[tame current]

wow

dogs dont eat grass though

[tawny elbow]

Some dogs do

Mine does when he gets really anxious

[narrow shadow]

They eat certain type of plant when their stomach is upset. But mine eats plants all the time without having any digestive issue

[old fog]

does anyone know how perplexity gets such good performance?

like how is it so fast compared to like gpt-4 with browsing enabled

and yielding better results?

[merry kettle]

I don't know how, but fully agree, the performance blows ChatGPT out of the water. The chatgpt speed for me is not suited for general research. While I now use perplexity as default search engine because it's fast enough to actually replace google.

[tame current]

Perplexity is literally search for me now, as well

It's incredible

[white sundial]

While I’m not a dev my guesses are that 1.) it is for normal search built on 3.5-turbo which is like 10x faster and 2.) it is built on google itself to retrieve results so the results are returned as fast as your regular google search

[halcyon coral]

I can say from my development experience that the GPT API is much faster than ChatGPT.

[grizzled canyon]

if you are a larger application you can get dedicated capacity

also the API is relatively reliable in performance anyway

[blazing idol]

Would it be possible to invite @latent wharf in our own servers?
Edit: You can by clicking on the bot... I could not find a invite online

[narrow silo]

perplexity is cool and all

i just dont understand dream option

obviously cant compare with other image models specified for that in terms of quality

so unsure why it even exists

for everything else it's obviously great

[signal hamlet]

Hey @blazing idol! It was in the #📢│announcements message channel. This link mentioned in the post works, so does the bot's profile.

[fierce hinge]

hello