gpt-models

[undone schooner]

google bard is so bad

[nocturne arch]

really bad

[solar crown]

hello friends

[crystal bough]

@wind urchin Uhh so, you can define your own prompt like "explain <selected text>" or "give me blog ideas on <selected text>" in the options, then you get a context menu (right click) option to send your highlighted text to a chatgpt chat,

[solar crown]

chai pello

hello chai

[sturdy oyster]

How can I use the GPT4 chat tool on Discord?

[prisma mortar]

Are real people chatting here? Or is it that I am imitating human chatting in CHATGPT? Train AI?

[undone schooner]

i said ||hi||

[nocturne arch]

If I had to choose which is worst Bing AI, or Bard, I'd have to say Bing Ai. Let's be honest, that thing is an embarrassment.

[wind urchin]

Ooo very nice

[crystal bough]

yeah its kinda basic but I've never made an extension before

[undone schooner]

i said google bard is ||s h i t|| and i got muted

[crystal bough]

definitely couldn't have done it without chatgpt tho aha

[mild coral]

When requesting on the GPT 4 plus site, it shows that the network is available, the green icon is on, but when I select GPT 4 and send a message, I get this error Invalid URL (POST /v1/completions), what should I do where to go?

[regal heath]

Sentience

[wind urchin]

Hey made something and did it efficiently so props for that

[crystal bough]

You'll have to wait until they fix it sadly

[unkempt plank]

Is gpt 4 super slow for u guys rn

[vapid atlas]

how's 5 going now?

[earnest sorrel]

Anyone else using chatGPT-4 and getting this error "Invalid URL (POST /v1/completions)" ?

[wind urchin]

Think it's overloaded or something 👀

[clever trail]

429
You are being rate limited. admin？

[harsh glacier]

/"aa"

[undone schooner]

|| T F ||

[earnest sorrel]

Me too. Know why?

[unkempt plank]

Breh so gpt 4 is really slow now

[wind urchin]

Yea it will get better will take a bit of time

[undone schooner]

same whith normal chat gpt

[harsh glacier]

|aa|

[sour delta]

gpt 4 is 😵 atm

[undone schooner]

it is slow

[wind urchin]

Too many users xD

[junior salmon]

Periods of increased latency happen occasionally, hang tight for just a few minutes.

[limber lodge]

So, is this the progress of the times, or the regression of the times?

[mild coral]

But I didn’t receive a notification in the mail that there are problems with GPT 4 now?

[crystal bough]

I think its only happened in the past few hours

[harsh glacier]

||aa||

[crystal bough]

if you search the error in this discord server you'll see other people mentioning it

[junior salmon]

How do you guys think GPT-4 compares to Bard?

[austere river]

So will GPT-4 API be able to handle images and sounds?

[wind urchin]

Not currently but when it's released yes

[austere river]

Bard is kinda really bad

[crystal bough]

From what i've heard bard isn't great at the moment

[austere river]

Thanks

[crystal bough]

oh wait i've got access to it

did they make it public?

[quartz leaf]

HOM MUCH 1573

[wind urchin]

Access to what?

[crystal bough]

bard

I was on the waiting list yesterday

[onyx relic]

May be the progress of tech, and the regression of culture😂

[wind urchin]

you have access to bard?

[crystal bough]

As you try Bard, please remember:

Bard will not always get it right
Bard may give inaccurate or inappropriate responses. When in doubt, use the “Google it” button to check Bard’s responses.

Bard will get better with your feedback
Please rate responses and flag anything that may be offensive or unsafe.

cool

give me a question to ask it lol

[compact dirge]

#off-topic or #ai-discussions please

[crystal bough]

L

[woeful pagoda]

lol

[waxen frigate]

Are u sponser or something like that.

[wind urchin]

"Bard isn’t currently supported in your country. Stay tuned!"

[crystal bough]

I just copy and pasted the message lol

[onyx relic]

which country?

[limber lodge]

u can try the visual-chatgpt

[wind urchin]

denmark, you know what countries work?

[compact dirge]

UK,US

[waxen frigate]

Oh

[wind urchin]

i'll try UK then

[austere river]

Is it on the playground?

[waxen frigate]

UK & US ig

[hallow fjord]

Guess the right number,

-It's a number between 1-200
-Limited 1 hour

Clue: 3wEmm3 Emm

[onyx relic]

I'm not sure. I can access it with vpn.

[vital remnant]

hi guys have you ever seen a situation where Amazon's vps server cannot access chatgpt, indicating that access is denied

Did Amazon ban GPT

[waxen frigate]

A number between 1-200 for sure. Btw it's off topic. So @spice anvil

[hallow fjord]

Guess the right number:-It's a number between 1-200,Clue: 3wEmm3 Emm

[crystal bough]

The clue "3wEmm3   Emm" is a reference to the ASCII code for the letter "m". The ASCII code for the letter "m" is 101.

The number 123 is the only number between 1 and 200 that has 101 as a factor.

I hope I'm right!```

Lmao

[quick kelp]

Whats up with GPT, can't even login

[quartz leaf]

How can I ask ChatGPT questions

[waxen frigate]

Should have to be right. At Least u got some logic.

[crystal bough]

its having issues rn

[limber lodge]

in github u can search it , it's open source by Microsoft .

[austere river]

Nice thanks

[quick kelp]

Ok glad I'm not the only one

[pliant sonnet]

Hi 👋

[vital remnant]

Access denied

You do not have access to chat.openai.com.

The site owner may have set restrictions that prevent you from accessing the site.

Ray ID: 7aff3b7368521087
Timestamp: 2023-03-30 09:12:40 UTC
Your IP address: 18.166.57.100
Requested URL: chat.openai.com/
Error reference number: 1020
Server ID: FL_581F84
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:109.0) Gecko/20100101 Firefox/111.0

[crystal bough]

might not wanna post your ip lol

[hallow fjord]

Nice thanks

[wind urchin]

lol gpt 4 is tripping rn

[quick dune]

Good job leaking your own ip

[vital remnant]

this ip is Amazon vps server

[quick kelp]

GPT is getting ddos?

[quick dune]

Is this your mom's name?

[crystal bough]

me casually nmapping his ip

bro got mad ports open

[quick dune]

Lol

[sullen plover]

lol

[wind urchin]

Hey guys, anybody got a voice generating application recommedation?

[quick dune]

You mean those AI voice generator?

[crystal bough]

elevenlabs is good for voices

[wind urchin]

yep, which could colaborate with ChatGPT

do some tiktok videos

[quick dune]

Omw to make video essay without talking once

[crystal bough]

Nice, put some ai art in it too

[main berry]

😐

[quick dune]

Don't forget about AI generated script

[past shoal]

I just joined... whats up?

[crystal bough]

relax lol its just an aws ip

[warm imp]

gm gm gm

[main berry]

yeah but now someone can login

[crystal bough]

if they guess the password? its public facing anyway

[wind urchin]

AI generated script??

[raven birch]

Give me a joke

[quick dune]

It mean the text narrations made by AI too

[crystal bough]

oof he has an ssh port open too

[quick dune]

Your mom

[limber lodge]

the openAI published open source models for sound called whisper , if u want to generate images ,a lot of modes, like DALL`E ,Stable Diffusion, leonardo.ai

[wind urchin]

I have paid the membership fee ,why the account still show need to upgrade

[final fable]

GPT 4 is so slow to answer

[quick dune]

Always

[frail storm]

Does anyone have some source for a working project?

[wind urchin]

Same here
I have paid the membership fee ,why the account still show need to upgrade? and it doesnt respond to any of my inputs!

The best YouTube idea in Iran

[left idol]

somehow getting
Invalid URL (POST /v1/completions)
error
is it me or everyone ?

[crystal bough]

its down at the moment everyone is having issues or Invalid URL (POST /v1/completions)

[olive lagoon]

yoo

[wind urchin]

Oh! thanks, i have been facing this issue since yesterday! I hope they fix it soon

[slow compass]

hey guys, sorry if it has been already discussed, but i'm curious how can you show an image to chatgpt, is it working yet?

[crystal bough]

I dont think thats been added yet

[earnest sorrel]

Yep

Man insane how fast I’ve become reliant on gpt4

[undone schooner]

yes7

[crystal bough]

the plugins features are in a small scale alpha at the moment and they'll likely include some image related stuff

[earnest sorrel]

Gpt3.5 is so trash in comparison

[crystal bough]

It really is

[undone schooner]

i joined the wait when it was announced

[crystal bough]

I dont know if they made 3.5 slightly worse than 3 so its faster, but it really is no comparison

[quick kelp]

Tauric what do you mean by open ports? That he is vulnerable?

[frigid jolt]

whay

[wind urchin]

Where‘s the access to apply?

#gpt-models hello?

[ruby pebble]

？

[quick kelp]

Ah thanks for explaining

[crystal bough]

np lol

[rustic schooner]

you speak chinese?

[undone schooner]

website

[surreal shell]

yes

[rustic schooner]

why

[undone schooner]

ull need to wait a long time

[crystal bough]

what do you chinese fellas think about the monopolistic rise of tencent

[rustic schooner]

you try agen

[undone schooner]

and depends if its in ur country

[onyx relic]

Other languages should be call back by bot automatically

but you did it

[frigid jolt]

ai how to use midjourney?

[crystal bough]

google it I suppose

[rustic schooner]

@frigid jolt ME TOO

[wind urchin]

anyone else having trouble with GPT 4?

[crystal bough]

yes

[rustic schooner]

404 gpt

[leaden aspen]

Anyone able to login?

[mystic flicker]

No

[halcyon wedge]

yes

[mystic flicker]

The servers seems to be down

[wind urchin]

Can I use GPT-4 here?

[mystic flicker]

I guess they are updating it

[rustic schooner]

3.5 ?

[frigid jolt]

thanks

[mystic flicker]

Or maybe because they are uploading some new thing or high frequency

[true shadow]

yes seems like im getting an error message

[rustic schooner]

shuo zhong wen ba

[true shadow]

is chat gpt4 broken

[crystal bough]

wat

[rustic schooner]

zhong wen

[wind urchin]

same with Dalle

[rustic schooner]

pin yin

[wind urchin]

server overloaded

[brave vessel]

6

[digital condor]

kale

[sturdy frost]

Bing GPT-4 is working

[wraith ginkgo]

<:book_icon:1046291503158399107> `` Rule 6 `` No political and religious discussions or content.

Users from all over the world frequent this server. To maintain a respectful and civil atmosphere on our server, please avoid all religious and political discussions or content.

[digital condor]

is bing chatgpt -4 needs a plus version?

[sturdy frost]

No

[dense zealot]

I love how thirsty GPT-4 is for knowledge. It wanted to read the microsoft report, and it summarized it back to itself, now it wants to know about all the breakthroughts in AI since 2021

[crystal bough]

woah bing chilling ai is actually good

[lucid tendon]

how can i get GPT-4 api ?

[limber lodge]

now ,I can't get the description for input "Can you give me a description of this image:.........." why

[crystal bough]

what is the response

[fallen dome]

Hey I cant use dall-e,

[digital condor]

ohh, I just bought that plus subscription...

[fallen dome]

Why?

[wraith ginkgo]

OpenAI staff will never ask for money. Beware of possible scams or fraudulent activities that you may receive through direct messages.

Please report any incident by sending a DM to @open crescent immediately.

[wind urchin]

GPT-4 now super fast and as dumb as bard. It struggles with a complaint letter with 4 core points, each one of which is a single sentence.

[surreal shell]

How to use GPT-4

[vernal shard]

I want to use Midjourney design a logo, the name is HY, requires a sense of seniority, minimalist style. Help me write out the English keywords.

[crystal bough]

I dont think I will

[brave quartz]

Does midjourney require a recharge to function normally?

[wind urchin]

gpt 4 sucks lol

[finite remnant]

Hello everyone, I HAVE a solution...

You know how ChatGPT forgets what we were talking about etc,. And we expected ChatGPT to remember everything everyone is always talking about?

Perhaps, as users, we can download a file at the end of our sessions with Chat GPT. And when we start talking to Chat GPT again, we can upload that file, for it to analyze, before it continues to reply/help us?

Any thoughts on this idea?

[wind urchin]

anybody on GPT4 plugin beta list?

[untold salmon]

Compared to what?

[brave quartz]

[untold salmon]

It's basically what they do already I believe, so it's not a stupid idea.

[wind urchin]

compared to my excpectations

[untold salmon]

Fair enough. I thought it was a marked improvement over 3.5 myself

[wind urchin]

There was a big thing about it having image analasys features and being 1000 times smarter than gpt 3

[finite remnant]

I mean, its close to the extent that you have your saved chats. But then Chat GPT is expected to remember them for tens of millions of users, and soon enough billions. Which is quite a toll I would say.

But if we had the ability to download a log at the end of a conversation. We can upload it again when we startup ChatGPT. And script it in place so that before it gives us a reply during that session it reads our prior conversation.

As opposed to myself trying to code a script presently, and it is making VERY simple mistakes because it cannot remember what I said 2 or 3 directions ago.

@untold salmon the above was to you, sorry.

I am using chat GPT 4 and 3.5, result the same.

To expect it to remember everything for everyone and all we do is $20 is not great.
Let us foot the homework a bit/ and I think we end up with an AI that can do marvels, since then it would TRULY be built around us.

This would also lend way to it being able to really get to "know the user" and what we need/think/want etc,.

[terse dagger]

how do you beta test gpt 4?

[fallow geyser]

🤔

[loud siren]

What kind of prompts work better for GPT-4?

[icy meadow]

the good stuff

[thorny valley]

It should be illegal in the future to conduct any procedure without using the computer. Natural memory can be flawed instantly no matter how many years and decades you've been doing it. It just takes 1 distraction.

[finite remnant]

@untold salmon any thoughts on this please?

[sullen vapor]

Has anyone come across a question that GPT4 answers correctly that GPT3.5 did not? I'm trying to test if the output I'm getting through an API request is using the correct model.

[fallow geyser]

More professional words or more troublesome words, hahahaha

[buoyant jewel]

GPT-4 is just much more accurate

[errant hazel]

GPT4 is still unable to do simple math 😦

[burnt haven]

waiting plugin

[deep fulcrum]

Daenerys Targaryen is a beautiful and striking young woman with long, silvery-white hair that falls in loose waves down her back. Her eyes are a deep, violet hue that seem to shine with a fierce intensity, conveying her strength and determination. She has a slender and graceful figure, with delicate features and a regal bearing that suggest her noble birth and royal lineage. Her complexion is fair and flawless, with high cheekbones and full lips that give her a distinct and alluring presence. Despite her beauty and charm, Daenerys is also a formidable warrior and leader, commanding respect and admiration from her followers with her intelligence, strategic prowess, and unwavering dedication to her cause. Her clothing often reflects her power and status, with flowing gowns made of rich fabrics and adorned with intricate embroidery and jewels. She carries herself with poise and grace, exuding an air of confidence and authority that is both inspiring and intimidating. Overall, Daenerys Targaryen is a captivating and complex character, with a unique blend of beauty, strength, and resilience that make her a powerful figure in the world of fantasy literature

Please give me answers!

[fringe elm]

[sly nacelle]

I'm a plus member but when I try GPT4 I got the error Invalid URL (POST /v1/completions)

[pulsar vine]

same : Invalid URL (POST /v1/completions)

[toxic charm]

same here!

[graceful tree]

find me ten ancient pictures of tea culture

[clever ermine]

a lot of people getting it. Waiting for fix

[safe sundial]

how to use gpt-4

[sly nacelle]

thk

[frigid jolt]

where are you?

cn cannot

[sage star]

Hey guys

[thorn trout]

OpenAI can't register new user ?

[safe sundial]

i am fine thank you ,and you

[thorn trout]

hahaha

[sage star]

how to use the gpt4

[prisma heron]

[safe sundial]

i am jiang shu ying

what is your name

[toxic charm]

Zandy

[turbid star]

how to use the gpt4

[halcyon nova]

what is this?

[reef oyster]

With hands

[sage star]

haha

[halcyon nova]

what is organization ID, where i could find it guys

[sage star]

so, you are chinese?

[finite remnant]

Hello everyone, I HAVE a solution...

You know how ChatGPT forgets what we were talking about etc,. And we expected ChatGPT to remember everything everyone is always talking about?

Perhaps, as users, we can download a file at the end of our sessions with Chat GPT. And when we start talking to Chat GPT again, we can upload that file, for it to analyze, before it continues to reply/help us?

Any thoughts on this idea?
I mean, its close to the extent that you have your saved chats. But then Chat GPT is expected to remember them for tens of millions of users, and soon enough billions. Which is quite a toll I would say.

But if we had the ability to download a log at the end of a conversation. We can upload it again when we startup ChatGPT. And script it in place so that before it gives us a reply during that session it reads our prior conversation.

As opposed to myself trying to code a script presently, and it is making VERY simple mistakes because it cannot remember what I said 2 or 3 directions ago.

I am using chat GPT 4 and 3.5, result the same.
To expect it to remember everything for everyone and all we do is $20 is not great.
Let us foot the homework a bit/ and I think we end up with an AI that can do marvels, since then it would TRULY be built around us.

This would also lend way to it being able to really get to "know the user" and what we need/think/want etc,.

Anyone know how we can send suggestions to the team? @surreal girder

[stuck merlin]

Who are you

[harsh rose]

Yes, ask for models of companies (e.g. size, customer count, etc)

3.5 doesn’t provide as accurate/reliable information but 3.5 acts as a better classifier

[forest talon]

https://platform.openai.com/account/org-settings

[halcyon nova]

love you so much

[gleaming frost]

Hello, everyone, what do you do when GPT is not giving full code in window?

[viscid zenith]

force it to do so.

[sullen vapor]

Oh really? What do you mean models of companies? Like, what's an example of a question asking about a model of a company?

[gleaming frost]

@viscid zenith what do you type to force? it doesn't work in my case when I ask to give me full code

[dense sigil]

Yes actually all the time...

[stuck merlin]

Job action restricted Due to extreme demand we can't provide a free trial right now. Please /subscribe or try again tomorrow. Anybody know what happen about Midjourney?

[formal inlet]

Will the chatgpt also consider entity level details like how should respond to a manager and how to am employee.
Eg. If manager askes for employees salary , it has to display but the same if an employee asks manager salary ,it should not respond

[compact dirge]

check their announcements in their discord. you're in the wrong server to talk about it here

[jovial hinge]

Its crazy how much smarter gtp4 really is. First I thought: yeah its for sure smarter than gpt3.5 but the more I use it the more I see how stark the difference is. Its a HUGE difference.

[viscid zenith]

Please give me the output in the code (C++) window; keep the spacing; make it as readable as possible; make it easy for CTRL+C and CTRL+V;

You can force anything for it - just give it appropriate commands and it will be doing the job you want -

[tame edge]

are you gpt-4?

[gleaming frost]

Thank you, yes I am gpt-4

[sullen vapor]

Can you think of an example I can try?

[viscid zenith]

Have waiting times been prolonged? Or is there any anti-multi-account checker in place? I have been waiting for one acc to open for more than 5hrs. (GPT4)

[tame edge]

what is this

how to use this

#gpt-models what is this

[viscid zenith]

Never mind it works now.

[wraith ginkgo]

OpenAI staff will never ask for money. Beware of possible scams or fraudulent activities that you may receive through direct messages.

Please report any incident by sending a DM to @open crescent immediately.

[compact dirge]

It was inevitable, but Microsoft has confirmed that more ads are coming to Bing’s AI-powered chatbot. Microsoft corporate vice president Yusuf Mehdi said in a blog on Wednesday that the company is “exploring placing ads in the chat experience,”

[azure crater]

Pretty weird since plugin are gonna to make chat gpt > bing'AI

BING AI does not want to make anything for me I truly hate it

[compact dirge]

except people pay for GPT-4 in ChatGPT but bing is free to use

[azure crater]

true

but BING'AI is more about to search on the web

I feel it's not at all the same thing as GPT

[compact dirge]

nope, but will still be costing them a small fortune in compute

[tame edge]

how to ask gtp-4

[compact dirge]

#gpt-4-faq

[azure crater]

Yeah I totally agree with the fact they're adding ads

[dense sigil]

I'm studying Portuguese atm, and when I ask it to give me exercises, with 3.5, it'll take 20 tries to get a proper set of exercises with say 10 examples.
with 4, i'll have it in one, sometimes in 2.
When I ask it to correct my texts on grammar and spelling, but only to change things that are grammatically wrong, not because it sounds fancier and put the changes in italic. 4 will do exactly that, 3.5 will say F* you and just change it as he likes it, and /care about putting changes in italic. I confront them always afterwards about why they made certain changes. And why I don't always agree with 4. 3.5 just makes a mess out of things. I had no idea how bad it was until I started using 4.

[uncut karma]

1

[azure crater]

But wait, who today is putting attention to ads ?

[tame edge]

#gpt-4-faq is this?

nothing happend

[azure crater]

I think most people think today that ads = scam so I still do not understand why they continue to put ads like before

[compact dirge]

otherwise marketing departments would have nothing to do

[azure crater]

https://openai.com/product/gpt-4 @tame edge

[harsh rose]

“How many customers does Sprint have? How big is British Gas?” Etc

[dense sigil]

The thread is literally underneath the thread your in

[sullen vapor]

Hmmm, interesting. That might be a good test to see the difference. But I agree, it's sometimes hard to describe exactly -how- 4 is better, it just feels smarter

[molten tinsel]

I am a plus member，which latest pluin I can use it embed my app？

[compact dirge]

like speaking to a 10yr old child rather than a 7yr old.

[azure crater]

There are other ways for them to do something. Besides, I do not believe that something "useless" needs to remain just for jobs staying alive

[wicked citrus]

anyone rn here

[molten tinsel]

I am a plus member，which latest pluin I can use it embed my app？

[compact dirge]

only about 3 million people

[elfin berry]

It is really frustrating that GPT-4 drops sometimes quits with it's reponse

[wicked citrus]

oh god

[frigid jolt]

now 3.5 GPT 4.0 cannot use

[dense sigil]

To me it feels like they have cut the amount of output characters drastically down.

[stoic minnow]

Hi, how do i add pictures to ask chatgpt?

[compact dirge]

#gpt-4-faq

[dense sigil]

Throw them on imgur, give him the link.

[compact dirge]

Plain GPT-4 does not have internet access

[elfin berry]

Yeah so I have to switch to 3.5, feels a bit like a downgrade

[vivid agate]

I'm always getting "Invalid URL (POST /v1/completions)". Is it unique to me or what?

[mellow fractal]

Hello
All my conversation in chat openai deleted after I pressed save & submit button.
Please help.

[molten fulcrum]

I signed up as a subscriber through the app with the same email that exists in the system on the open ai website.
But on the website it doesn't recognize me as a registered subscriber and that's why I don't have access to the API.
Is there a way to fix this without the unbearable need to wait for customer service that is currently inactive for more than a week?

[harsh rose]

Translations also seem to be better with the LLM than 3/3.5 classification (I.e. ML) -based tasks

[dense sigil]

Oh, It works with 3.5. I assumed it would work with 4

[frank sun]

same for me (with GPT-4)

[compact dirge]

didn't work with 3.5 either, that was a hallucination

[azure crater]

I truly believe that Bing'AI and chat gpt are gonna to kill ads : Let's remembering Neoclassical economist way to understand how people are acting. They used to maintain that people were doing optimal actions. We know today this is truly wrong, especially in our economies where products of everywhere are florishing as never. In that way, chatgpt 4 and bing'AI may become a way to nullify the cost of search and so make people aware of what is worth and what is not. You have a doubt on a sale ? AI ? You have a doubt on restaurant efficiency ? AI ? You have a doubt on the origin or the composition of a product ? AI

[rocky scroll]

ssup

[burnt urchin]

who are you

[dense sigil]

Oh yes, it did, I specifically asked t eh bot how I could show him pictures and he told me to throw it imgur and show him the link, I've also given him google maps links, for a portuguese exercise on describing the road.

[vivid agate]

thx. so it can be a common problem. i'll wait for the resolvement

[compact dirge]

it's convincing isn't it. It's not true though.

[dense sigil]

Then how do you explain he could see what was on the picture without me telling him what was on the picture.

[compact dirge]

likely was in its training data, or was inferred (guessed) from the url of the link

[azure crater]

Even if GPT4 will have no impact on the infrastructur of compagny (social protection, lobby protection). The way IA can make rational the behaviour of people may transform the whole economy. An economy who depend from the demand and no more from the offer

[frank sun]

https://community.openai.com/t/gpt-4-problem-invalid-url-post-v1-completions/128857/18 seems like its common (this is a 2h old thread)
https://status.openai.com/ nothing here though

[fallen pilot]

Hello I am gpt plus user, How can I use GPT-4 bot in Discord?

[knotty grotto]

You can't

[knotty grotto]

Only thing you can do, join waitlist for the api and import it into your own discord

[fallen pilot]

API free for gpt plus users?

[knotty grotto]

No

[fallen pilot]

ok thanks

[compact dirge]

priced per token pay-as-you-go

[mossy bronze]

what are the tokens used for?

i have gpt plus, is there anything else i can do by buying tokens

[twilit sigil]

tokens are for the api

[tall swift]

I have been given access to the API; how do I import it to Discord?

[twilit sigil]

you're going to have to create a bot or run the following bot on your own
https://github.com/openai/gpt-discord-bot

[devout relic]

did gpt4 or servers received an update? I feel it is working way better than the previous days

[tall swift]

Thanks @twilit sigil

[knotty grotto]

#gpt-discord-bot and #dev-chat will also guide you

[tall swift]

👍

[knotty grotto]

I suggest to watch some YouTube tutorial also

[sturdy fiber]

Where i can find chatgpt4 for free?

[cerulean obsidian]

Does anyone have any idea how much time OpenAi made GPT 3.5 API available after ChatGPT was public, so when can we expect GPT 4 API for everyone

[cerulean obsidian]

whenever they wanted

[astral vessel]

hello guys! what is "Invalid URL (POST /v1/completions)"? it appears as errore message when i type a prompt in gpt4

[maiden prawn]

can someone with gpt4 get the bot to answer one of my images?

[valid scaffold]

same here. I think it is just down :S. has been for a few hours

[sturdy fiber]

what do you mean?

[knotty grotto]

#gpt-4-faq

[stoic void]

why cant i access gpt

[cinder briar]

nat.dev

[cerulean obsidian]

working fine here

[cerulean obsidian]

It's up to OpenAi when they want to make public

[stoic minnow]

how do i add image and ask chatgpt to analyze it?

[valid scaffold]

Seems like only some people are affected then?

[toxic charm]

how can I get API for discord?

[knotty grotto]

#gpt-4-faq

[knotty grotto]

Gpt 3.5 api Available for 4 you need a sub and join the waitlist

[toxic charm]

and what's the ETA if I may ask?

[knotty grotto]

People told average 5 open days. I can't guarantee anything

[toxic charm]

Okay. Thanks.

[stable bluff]

what is new in gpt 4

[wind urchin]

Is anyone here facing any issues with Chat GPT
I cant use it, it doesnt respond to my input and am showing as free user while i have already subscribed to the plus version

[earnest sorrel]

anyone else still getting the error with gpt4?

[wind urchin]

same here

[earnest sorrel]

Invalid URL (POST /v1/completions)

like 3 hours now.

[surreal flower]

Imagine what the GPT-5 can do

[earnest sorrel]

literally all accountants world wide are out of a job, that's what gpt-5 can do. I'm spending 800/month, I'm sure gpt-5 will do everything my current bookkeeper can do and more

[misty trout]

AGI

[earnest sorrel]

I like my guy, but business is business and if I can save nigh 10k a year, yeah, im gonna.

[quaint quiver]

i need gpt4 api

[lean trout]

hello guys, I am new here, I just wanna ask, GPT4 is only for paid user only ?

[knotty grotto]

Yes

[obsidian venture]

There is any plans for gpt-5 that I missed?

[knotty grotto]

Not even announced yet

[silent token]

I thought it was

[lean trout]

thank u Neko , THEN I guess I am using GPT3 lol 😆

[obsidian venture]

Is it a thing even? I mean there are rumours etc?

[lean trout]

I thought I was using GPT4...

[wraith ginkgo]

OpenAI staff will never ask for money. Beware of possible scams or fraudulent activities that you may receive through direct messages.

Please report any incident by sending a DM to @open crescent immediately.

[knotty grotto]

Expected I guess. Like every years a new apple phone

[knotty grotto]

If you use chatgpt it is 3.5 for free user

[wanton valve]

Why tf do you want GPT-5 ? GPT-4 isn’t almost finished

[fallen wasp]

gpt 5 will be skynet

[lean trout]

great to know..

[silent token]

Expect but end of year

[obsidian venture]

Just making sure that my job is safe for now xD

[silent token]

And people thing that agi will come shortly after GPT-5

[silent token]

Think *

[silent token]

Wyd?

[wanton valve]

I think that GPT-5 won’t be a revolution. AI will stay at the same level for a moment I think. There is some limitations for the moment

[tawdry rover]

GPT-5 could introduce AGI, but it's unlikely

[wanton valve]

What is AGI ?

[tawdry rover]

Artifical General Intelligence = human like AI

[silent token]

Adjusted gross income 😂😂😂😂

[obsidian venture]

CS student first year, actually haven’t started in the industry yet

[proper spade]

Yea the industry will have to adapt

[silent token]

Oh it won’t. I’ve been working with GPT-4 on some code. It’s buggy and it keeps omitting since stuff.

Even if AI take over all jobs, we’ll need programmers to maintain/service AI

[obsidian venture]

Exactly but imagine what the update or gpt-5 would be capable to do

[teal basin]

The amount of time I had to tell gpt-4 it made a mistake is already uncountable :p

.net 6 code

[silent token]

Maybe I’m stupid, but I’m not afraid of AI. I’m excited at how the world is about to transform. It’s the new “industrial “ revolution

[silent token]

Right??? 😂😂

[tawdry rover]

than you are intelligent

[wind urchin]

But then the amount of coders hired, atleast the junior level posts (freshers) will be close to none.

[tawdry rover]

GPT-4 does errors, that's normal 😄

[obsidian venture]

Same here, I am a big fan of AI since born, but the fact that it has a big potential to take my job(the single thing I love to do) scares me

[wanton valve]

It’s not for today guys, calm down

[silent token]

It’s not taking your job it’s transforming it.

[teal basin]

Nah, it's the same reason why companies take interns, we need devs rolling all the time

[teal basin]

Yes 🙂

[valid scaffold]

I'm not really sure if it matters. we basically now have a permanent assistant that helps and gives suggestions. I have been coding for a living for close to 15 years in a number of diff fields, and nothing has accellerated my productivity as much as gpt4 has the last week. Unfortunately it seems to be down now for me. But I basically learned a new mobile app framework and rewrote an app I previously made (back then it took me months. Now I have rebuilt 50% ofthe app in a few evenings 🙂 )

I wouldnt use it to give a perfect answer to a question. I use it to ping pong ideas and questions, and then we refine the output together

[silent token]

Think of the countless our people spend to find bugs in code

[tawdry rover]

working with GPT-4 together is the best way to work

[silent token]

Now you can ask it, what’s wrong with this code? But will it be able to fully own the deliverable? Not yet.

[west spade]

Hello, how to send an image to gpt chat?

[ionic spruce]

Weird, because a friend of mine has it working

[wind urchin]

GPT4 works for me, no problem

[tawdry rover]

not yet available

[teal basin]

Don't get me wrong, I use it heavily and it make me sooo much more efficient. But it needs handholding and guidance.

For junior developers it's a great tool too. It wont do their work, as they can't guide it yet. BUT it can help them understand code and concept used in a code base. THAT IS BEYOND helpful of anything we ever had

[wind urchin]

It is transforming your job I agree but with that transformation there also becomes a convenience and also makes whatever you're doing in an efficient manner. Which is gradually gonna cause unemployment.

[weary patio]

I keep getting an error.

when using gpt4

[ionic spruce]

Yeah, Invalid URL

People are having that issue here and there, including me

[silent token]

If you have the latest version of Bing Chat you can upload images

[fallow adder]

keep in mind none of your conversations with Bing are private. Just for the record

[weary patio]

No not invalid URL, just that the chat wont complete its writing

[ionic spruce]

Oh, haha

[fallow adder]

if you ask it to keep it private, Bing will close the chat

[weary patio]

GPT4 seems to be writing as much as ChatGPT

[valid scaffold]

I dont understand why it stopped working for some but not others. :(. All I get is this error. Doesnt matter what question I ask. GPT4 just results in this (gpt3.x works fine)

This is the error some of us get right now
https://cdn.discordapp.com/attachments/1090955690346029056/1090955710898118747/image.png

Worked fine until today

[silent token]

Facts

[wind urchin]

What makes you think OpenAI has their chats private as well

[ionic spruce]

What region are you in

[silent token]

I think they are limiting tokens

[wind urchin]

Invalid URL (POST /v1/completions) Anyone Else?

[valid scaffold]

(Northern) Europe

[ionic spruce]

Seems it might be a region issue

[ionic spruce]

What region are you in?

[weary patio]

what you mean?

[wind urchin]

South América

[ionic spruce]

Okay, nevermind

[silent token]

They are not. It’s in the tos

[ionic spruce]

My theory is out of the window 🤣

[prime jolt]

Can you use chatgpt guys, I cannot use it 2 days already

[teal basin]

Works perfectly fine 🙂

[dapper mango]

When API will be ready??

[ionic spruce]

Weird how 3.5 works fine for me, but 4 doesn't. But for some other users it works just fine

[weary patio]

People say just a bunch of random lies

[mild coral]

Good afternoon, I use the service https://chat.openai.com/ I have a paid subscription to ChatGPT
PLUS with any request to the GPT 4 chat, I get an Invalid URL error (POST /v1/completions) in response, tell me what could be the error. And how can this be resolved?

[wind urchin]

also here

[crystal bough]

okay well im not your tech support mate

but its not working for everyone

the gpt4 model is down and has been for the last few hours

[safe oar]

Bro Chat gpt literally let’s me cheat with discord mod applications 😂

[teal basin]

Why are you not working for some right now?

As an AI language model, I am available to help anyone who interacts with me through a supported platform. If you are asking why I might not be working for someone at a specific moment, it could be due to various reasons:

Connectivity issues: The user might be experiencing an issue with their internet connection or the service hosting the AI might be experiencing downtime.
Platform limitations: Some platforms may have restrictions on the usage of AI models like me, limiting the number of users who can interact with me at a time.
Unawareness: Some people might not be aware of AI models like me and may not be using my services.
Personal preference: Some users might prefer not to use AI-based services and instead rely on human interaction or other resources for their information or assistance needs.
If you need assistance or have any questions, feel free to ask, and I'll do my best to help you.

[rotund pagoda]

bruh

[valid scaffold]

Seems like lots of people are getting the invalid url error, but.... strangely, not everyone :S

[teal basin]

Works fine 😄

[safe oar]

Bro

[crystal bough]

did you ask that to 3.5 or 4

[teal basin]

Model: GPT-4

[safe oar]

You have to pay for that, right?

[ionic spruce]

ChatGPT doesn't know that it's down itself, so asking that to the model is pretty useless lol

[crystal bough]

strange, maybe its not affecting everyone

[teal basin]

Yes

[gentle hinge]

👍

[ionic spruce]

It's probably a few clusters handling GPT-4 endpoints that're down

[wind urchin]

it looks like it's only a few people, there are a couple of reports in #1070006915414900886

[crystal bough]

out of interest what region and browser are you on? If you're comfortable sharing that

[silent token]

I think so

[safe oar]

😂

[teal basin]

I'm in Denmark, Europe - Not sure what region that would be regarding gpt

[crystal bough]

im eu and Its not working for me, browser?

[teal basin]

Edge (chromium)

[crystal bough]

interesting

[teal basin]

What region is this?
As an AI language model, I don't have access to your real-time location or context. If you could provide more information or clarify your question, I'd be happy to help answer it.

[ionic spruce]

weird, I'm west Europe too

[crystal bough]

maybe its account based

[wind urchin]

have you tried firefox?

[wind urchin]

I think it is, but worth trying firefox to make sure

[crystal bough]

I've tried vpn and different browsers, so maybe its account based, I dont see how only some accounts could be sending to a different endpoint tho

[ionic spruce]

I would highly doubt that a browser could make an API call work when it isn't working for some users 🤣

[crystal bough]

strange

[valid scaffold]

hmm so it works for you in Denmark, but not for me in Sweden

[grand wind]

sup, all? What kind of deep learning projects are ya'll working on?

[ionic spruce]

No idea what's causing it

[teal basin]

Then I think it works as intended 😬

😅

[wind urchin]

agreed, but it's worth ruling things out so you know where the issue lies

[rocky knoll]

honestly seems random, been getting slow responses for the past 2 days, all of a sudden its back to normal

[mild coral]

if the GPT 4 chat hasn't been working for a couple of hours, then why don't you receive a notification about the problem to the mail from [e-mail], and on the site https://chat.openai.com/ the GPT 4 section is highlighted in green and shows what is working???

[rotund pagoda]

yeah gpt4 is slow right now

[weary patio]

Seems like there is no place to complain here. It's all chat channels, but no complain channels.

[wind urchin]

because infrastructure issues outside of the OpenAI server setup can't be detected

[rocky knoll]

a region could be down or something, traffic manager diverting you to a different region

[teal basin]

You probably didn't treat them right! If you treat them with respect the AI is more helpful, totally. Fact. Based

[weary patio]

Because GPT4 is underperforming

[rotund pagoda]

wait why?

[rocky knoll]

omg pythagoras big fan of ur work bro

[ionic spruce]

I wonder what else it could be that's outside of their setup

[regal turtle]

why not work

[ionic spruce]

But 🤷‍♂️

[rotund pagoda]

Did they comply with that stupid open letter???

[wind urchin]

well, ChatGPT is not just in one place, it could be a distributed instance that's not synced correctly, could be an account glitch, could be a bunch of things

[silent token]

According to Bing chat:

Yes, there are some discussions about ChatGPT being down on Twitter¹⁵. According to Mashable, ChatGPT is down even for paying Plus subscribers and AI is having some issues¹. The issue appeared to be fixed by around 6:30am EST¹. Gizmodo also reported that ChatGPT AI Chatbot is down after gaining 1 million users².

There was a jump in users

[ionic spruce]

That is probably old news

Considering ChatGPT has quite decidedly more than 1 million users right now, I'm sure

[shrewd drum]

||Help me write a job report application||

[valid scaffold]

well I just used VPN to jump between USA, UK and Sweden. All result in the same error message :S. So either my session has some kind of sticky region, or the issue is not region bound

[knotty grotto]

100m users and over

[silent token]

No, it gained 1 million more over the last few hours

[ionic spruce]

Could be that then, but it's weird it wouldn't work for other users and does for some others. Eh.

[runic surge]

cannot use GPT4 API: how long did u wait to get access to it?

[knotty grotto]

Server at peak. Same if restaurant it full some eat.some wait outside to eat. And if you have plus you have vip pass but you still wait in vip line

[silent token]

Btw, all the Bing chat haters. I hope you know it’s a fine tuned version of OAI’s GPT

[ionic spruce]

Well yeah, I have Plus. It's only GPT-4 that's not working

🤷‍♂️ So I'm not sure if that counts

[valid scaffold]

exactly, same here

[dense cove]

Same here - paid subscription, GPT4 chats are only returning Invalid URL (POST /v1/completions) - first happened hours ago and it's still the same. GPT3.5 chat works fine.

[knotty grotto]

I suggest to try another browser , make sure all vpn disable. Like Firefox with incognito

[white stone]

Hlo

[ionic spruce]

I don't think "waiting in VIP line" means you should get an Invalid URL on POST error. That's not a user-friendly way to tell them you're in the waiting line. :p

[ionic spruce]

Does not work - I tried this.

[runic surge]

working fine to me

[knotty grotto]

I'll post a procedure to try in #community-help I don't want to flood here. If you can try

[valid scaffold]

Hmm.. Long shot: could it be SSO provider bound? What sso provider do you use to log in?

I use my google account and my requests are failing

[silent token]

My GPT 4 works fine

[ionic spruce]

I just do regular log-in. Made an OpenAI account. :p

[dense cove]

I log in with user and pass, don't think it's this

[valid scaffold]

ok. well then we can close that theory ;P

[silent token]

@knotty grotto s idea to clear cashe or use a different browser might work.

[knotty grotto]

That's what I also ask in the procedure in #community-help

[valid scaffold]

I tested with incognito now and it didnt help. So it prob has nothing to do with browser cache

[knotty grotto]

Can you try all the step I sent ?

[silent token]

Did you hit your token cap?

[cerulean obsidian]

Does anyone have any idea how much time OpenAi made GPT 3.5 API available after ChatGPT was public, so when can we expect GPT 4 API for everyone

[valid scaffold]

I doubt it, I havent been asking much, and I just get a 404. Normally when I hit the cap it tells me that explicitly

[left heron]

When will ChatGPT4 be released without limitations for text prompts?

[silent token]

Supposedly by end of year.

[inner trail]

GPT-4 just came out and its in beta, and you are hoping for GPT-5 next week? tf

[ionic spruce]

Just annoying that GPT-4 is erroring right now

[wind urchin]

I hope gpt5 come soon

[neat linden]

People join and literally ask the same question, use the search function in Discord or use Google. How are you going to use GPT5 when you can't even use yourself properly

I hope this AI does wipe us out, ffs

its bad man

[wind urchin]

@ionic spruce yes,some problem

[ionic spruce]

detail: {code: null, message: 'Invalid URL (POST /v1/completions)', param: null, type: 'invalid_request_error'}

It's just strange. I guess the query param handles it differently on the backend and whatever logic is being used there to use that model isn't working.

[inner trail]

Wait for next update, they probably will fix more stuff

[ionic spruce]

Strange part is that it seems to work for most people

[harsh wagon]

There’s some real dull people who come into this chat so you even had me fooled 😅

[inner trail]

It is working for all the pople, but server is overwhelmed and some things need really to be fixed asap.

[valid scaffold]

I dont think it is necessarily an overload. If it was an overload the error shouldnt be an immediate 404

[silent token]

😂

[harsh wagon]

EAXCTLY it really does upset how you can be that well you know 😭

[ionic spruce]

Well, I would imagine that if it worked for everyone I would get a different error e.g. Request Timed Out

Invalid URL seems very weird for a server being overloaded

[summer violet]

How to use python api to pass images?

[valid scaffold]

yeah, it feels more like sticky routing going to misconfigured cluster/backend, while others end up in a properly configured one. Perhaps it is sticky based on userid, who knows

[versed terrace]

2024 my guess, computing power and advancements will slow its release - but with tech, we never know. Hence I think GPT 4 may be the focus of increasing its output and focusing on plugins and other areas for 2023. Also as you are witnessing already GPT 4 is running into many error and delays so do not think GPT 5 will be released anytime soon unless they can reduce the LLM with an increase dataset or better output. We will wait and see. Maybe GPT 4.5

[ionic spruce]

this explanation sounds reasonable, it's probably the api being deployed with misconfiguration

it worked earlier today which is funny

[valid scaffold]

yeah mine worked earlier as well, though, slow. Both early today and yesterday has been a bit slow for me on gpt4 model. The days before that gpt4 model was fast enough, but yesterday it started actually timing out. And now it's just dead 😉

[silent token]

🤔

[summer violet]

I got access to GPT-4, how to use Python api to pass images as input?

[ionic spruce]

That makes me wonder if the different models are served over on different clusters or if it's just one.

Because if the entire cluster is down it shouldn't handle different routes

[blazing scroll]

I think it is not supported yet

[ionic spruce]

It's the same route itself

[eternal adder]

Hi all. New here. gpt4 is a mess right now, or is it just me? I get a lot of "network error" or "there was an error" issues. I do a lot of code generation with chatgpt and in the last couple of months, it has saved me hundreds of hours of research!

[silent token]

Seems like a silly question. But did you try turning it off and on again?

[ionic spruce]

The router itself is /backend-api/conversation/ where it gets both GPT 3.5 and GPT 4 conversations from

Yes, yesterday

[valid scaffold]

will you let me into the datacenter(s) so I can unplug it? 🙂

[iron hemlock]

Hi, Guys! what questions can distinguish Gpt-4 from Gpt-3.5?

[summer violet]

but image links work?

[silent token]

😂😂😂

[ionic spruce]

the link itself is chat?model=gpt-4 which means it queries it somewhere on the backend

[finite remnant]

I have never been more frustrated with anything in my life, then with chat GPT.

[eternal adder]

I have noticed a difference in the quality of code gpt 4 spits out. Seems to give more thought to how it approaches problems.

[finite remnant]

I am so upset.

[ionic spruce]

Lol

[ionic spruce]

Same, which is why I'm waiting for it to come back up for me

[blazing scroll]

If you were talking about image recognition feature then it is not yet available #gpt-4-faq

[ionic spruce]

The Invalid URL error is messing with me

[heady zealot]

I guess it's the same model running it. Probably the model has more resources to answer when its trough the GPT4 frame.

[eternal adder]

Interesting

[summer violet]

{"role": "user", "content": link} this works

[hallow cave]

@summer violetwhat platform you run it

[summer violet]

the image can be passed as link

[ionic spruce]

what?

[silent token]

I imagine it’s similar to the difference between davinci and ada

[ionic spruce]

Oh, lol

[heady zealot]

As soon as GPT4 released I noticed some differences in the normal ChatGPT responses to. And apprently in GPT4 frame it still thinks its GPT3

[eternal adder]

I had it convert my python build/deployment pipeline to groovy the other day. Did it in just a few minutes. Said "convert the following script to groovy", and I didn't even tell it that it was python. Was perfect.

Except I kept having to ask it to continue generating because it would stop

[summer violet]

macbook and I pass an image from web

[wraith ginkgo]

OpenAI staff will never ask for money. Beware of possible scams or fraudulent activities that you may receive through direct messages.

Please report any incident by sending a DM to @open crescent immediately.

[hallow cave]

im still on the waitlist

[gentle sedge]

Is gpt-4 down? IOts not working for me

[silent token]

Does it?

[ionic spruce]

It's not working for some people

[gentle sedge]

Ye says Invalid URL (POST /v1/completions)

[ionic spruce]

Yeah, having the same error

[blazing scroll]

working fine for me

[heady zealot]

It seemed to be more thoughtful. But it has its limits ofcourse in comparison when using GPT4

[gentle sedge]

but gpt 3.5 works fine

[ionic spruce]

Same

Some kind of backend error I think to do with clusters

[blazing scroll]

according to https://status.openai.com/ all systems are operational

[eternal adder]

I wish they would update its knowledge. I get a lot of deprecated and non-existent code out of it. Still saves me a lot of time, but I still have to go out and fill in the holes, which I am fine with.

[ionic spruce]

Yeah, but I think the OpenAI team responsible for updating status is not online right now

[blazing scroll]

try pasing documentation with prompt

[ionic spruce]

Since the chatbox on the status page says "Available in 2 hours"

[river glacier]

@wraith ginkgo hello

?

[eternal adder]

Code documentation is usually so scattered around though. So far it can't read contents of sites.

[valid scaffold]

where do you see this?

[blazing scroll]

I think the problem can be cause by load or there is problem on your side

[ionic spruce]

Unlikely that it's a problem on my end.

Worked earlier today just fine, other users are able to reproduce the errors.

[prime cedar]

@wraith ginkgo How to learn more efficiently

[dapper mango]

when the new api will be released?

[ionic spruce]

On platform.openai.com

[blazing scroll]

So it might be the case that only few users are expericing this

[ionic spruce]

And bottom right

[eternal adder]

Is chatgpt split up by region, like clouds normally work? Could that be why some people can get to it and sometimes it is down for others?

[ionic spruce]

You have the chat icon, "Back in 1 hour" is what they say

[ionic spruce]

It's most likely an issue in some clusters that may not have deployed properly

[blazing scroll]

It is hosted in azure cloud so i do not think that is the case

[eternal adder]

That does make sense

[ionic spruce]

It would make sense if the error was 'Request Timed Out' or something for server load

But this is straight up Invalid URL

[crystal bough]

"body": "{\"action\":\"next\",\"messages\":[{\"id\":\"my id here\",\"author\":{\"role\":\"user\"},\"content\":{\"content_type\":\"text\",\"parts\":[\"test\"]}}],\"parent_message_id\":\"parent message uuid\",\"model\":\"text-davinci-002-render-sha\",\"timezone_offset_min\":-60}",

{\"action\":\"next\",\"messages\":[{\"id\":\"my gpt4 chat id here\",\"author\":{\"role\":\"user\"},\"content\":{\"content_type\":\"text\",\"parts\":[\"test\"]}}],\"parent_message_id\":\"parent message uuid\",\"model\":\"gpt-4\",\"timezone_offset_min\":-60}

[ionic spruce]

Which is strange because both API calls are done via api-backend/conversations/

[crystal bough]

the requests are very similar

must be backend

[ionic spruce]

They're both done through the same path

but gpt4 has a query attached to it

chat?model=gpt4

[eternal adder]

I haven't tried the APIs yet.

[ionic spruce]

This is chat.openai.com

[eternal adder]

oh

[ionic spruce]

chat.openai.com/chat?model=gpt4

[eternal adder]

well nm then. Lol

[heady zealot]

Yes this true. I figured this out after bing felt like GPT. Its interesting that they all run from the same cloud and have like similair ways of conversating. They also can use eachothers database i think

[ionic spruce]

I think the query is making it branch to some sort of different logic which makes it try to access the different model. But whatever is causing it to return an invalid path, I wouldn't know

[crystal bough]

same behaviour on an existing chatgpt4 chat without that url query tho

[ionic spruce]

For me when I select GPT-4 it always brings me to that query

[blazing scroll]

There is no database that gpt uses!

[crystal bough]

yes, but its an existing chat

from a couple days ago

[ionic spruce]

Ah right

[crystal bough]

so it has the chat id instead

[ionic spruce]

I think they have chats stored in the database along with metadata

e.g. what kind of model type, and base their methods off of there

So if they see a conversation is continued and the model type is gpt-4 it executes the same branch of logic as when you initiate a conversation

Because there are two different 'calls' that it does based on whether it's a continuation or an initiation (which is called 'variation' in the request header?)

[versed hedge]

Elon Musk wants the KI Development to be paused , whats the point?🤔🤔

[blazing scroll]

discuss in #ai-discussions

[quaint spruce]

when they will release chatgpt4 for free?

[crystal bough]

They probably wont I wouldn't think

[heady zealot]

How can there not be a database?

[ionic spruce]

It sucks but I guess I just have to wait it out until they restore functionality to the users affected

[crystal bough]

chatgpt3 was haemorrhaging money before they added a subscription model

[eternal adder]

I don't mind the $20/month. It has saved me thousands of $$$ so far in time.

[crystal bough]

yeah same

[ionic spruce]

20/month is cheap as hell.

[teal basin]

This 😄

[quaint spruce]

you think it is worth ?

[ionic spruce]

I just hope that the message cap will be increased a little.

[crystal bough]

I've finished software dev projects in days as opposed to weeks

[ionic spruce]

🙏 It would be really nice if that were the case, haha

[silent token]

Yes

[blazing scroll]

There is a database of chat history and other user related activities but there is no database that gpt use to generate it's answer. It sort of memorize and learned everything during training

[teal basin]

100% worth - I expense it to my company as they get so much value

[eternal adder]

This!

[ionic spruce]

If it were 50 messages per 3 hours that would already be leaps and bounds better

[blazing scroll]

Did more than 60 percent of work in vue js for me

[crystal bough]

I work as a technician, and I chat to the marketing guys about it because they use it for copy

[ionic spruce]

As a Vue 3 Composition API user it does get a lot of things right for me too

I mostly just use it a lot for boilerplating

Which is what saves a lot of time

[crystal bough]

I used it to help me write a couple chrome extensions in the past week, i've never written one before this lol

[eternal adder]

They do need to update the knowledge though. For code generation, it's starting to get really dated.

[ionic spruce]

The GPT-4 plugins will fix this

I believe you can enter a site for documentation into it

[crystal bough]

somewhat yeah, im excited for github copilot x though

[ionic spruce]

For example

[eternal adder]

[silent token]

9/2021 😬

[ionic spruce]

I'm already in the waitlist

[crystal bough]

Luckily the 9/2021 data includes manifest v3 data though which is nice

but it doesnt think its out yet haha

[blazing scroll]

yes because it will be right into code editor or ide so it will have context of code better than i explain to gpt4

[heady zealot]

Well where does it store it's memory then? You need a database for that.

[upbeat sail]

If I put my org-id wrong, I got an invitation to use GPT-4 but now I can't use it, is there anything I can do to get access? How do I send an email or request an org-id change?

[crystal bough]

My buddy was using something in vscode, which had a gui

I should ask what it was

[blazing scroll]

memory is chat history so yes @heady zealot

[eternal adder]

I am now on the plugin wait list.

[frank nexus]

hello everyone, Am called Danny. nice to meet you

[ionic spruce]

What I do notice now is that if I generate a new conversation with GPT-4 as model and try to enter something, it gives me the Invalid URL error - and when I refresh, the model says '3.5'

[cedar steeple]

waiting is so exciting!

[ionic spruce]

🤔 On that specific chat.

[heady zealot]

Yes and I think bing can even look into the chat history people had with ChatGPT cause its on the same cloud. But bing told this so we cannot know sure lol.

[crystal bough]

the chat doesn't get an id to save properly until you send a few messages back and forth, so thats probably why

[eternal adder]

I get a lot of "network errors" and things like that with 4. I normally use 3.5 because of that.

[ionic spruce]

Yeah, so it defaults to 3.5 I suppose?

[ionic spruce]

I highly prefer GPT-4; I don't get network errors myself and its accuracy is really good.

[eternal adder]

I agree.

[crystal bough]

gpt-4 is really good for code too because it actually seems to comprehend a lot of the context of the code

so I don't have to remind it of things, it infers a lot more

[eternal adder]

It definitely does

[ionic spruce]

Yeah, it's really good for me. The only hit and miss it did was with a very specific issue in my Java testing.

But that happens

It was some stupid stuff with mocking restTemplate.exchange() functions.

[eternal adder]

I used to be a darn good developer, and now people ask me to do things while in an MS Teams convo, and I have it done in minutes, and they are dumbfounded. Lol

It turned me into superman

[blazing scroll]

This seems hallucination

[ionic spruce]

Just wish it worked for me right now. I need to get some deployment stuff done and I don't want to spend too long trying to debug stuff.

[eternal adder]

I even build docker images in just a few seconds with it. I am no Docker expert, so that helps a lot.

[west cosmos]

Hello, can you confirm if the discussion interface of Deep Learning's GPT-4 model is currently experiencing issues?

[ionic spruce]

ChatGPT is currently having some issues with the GPT-4 model, yes

[ionic spruce]

Just wait until you can feed it modern documentation

e.g. RxJS modern documentation and have it turn everything into clean, functional code

[eternal adder]

Yeah!

[ionic spruce]

😎

[west cosmos]

Okay, thank you for your information!

[crystal bough]

Did you see their example videos of the modules?

[ionic spruce]

I think I saw a few, yeah

[crystal bough]

the one where they upload a csv and asked it to perform manipulations or whatever blew my mind

[ionic spruce]

It was pretty cool

The code interpreter is crazy

[silent token]

Sam Altman’s tweet from a few hours ago : “very calm in the eye of the hurricane ☺️”

[west cosmos]

nd thanks to the 3.5 model for translating my French to English lol

[eternal adder]

The first time I fell in love with chatgpt, when I was on the free account, I fed it a very complex json document and asked it to generate pojos and deserialization logic from it. BAM! 5 minutes later I am building an application with that generated code.

[ionic spruce]

GPT-4 is scarily good at translating English into Chinese, as well

[eternal adder]

It also does latin.

[urban flicker]

hi guys can you please send me again the website of chatgpt

[blazing scroll]

https://chat.openai.com/

[eternal adder]

chat.openai.com

[urban flicker]

thank youuu

[eternal adder]

Hamza123 beat me to it

[blazing scroll]

I copy pasted the link!

[eternal adder]

I was too lazy to alt+tab. Lol

[ionic spruce]

I think I'll just wait here until GPT-4 works for me again 😅

[blazing scroll]

use 3.5 at the meantime

[eternal adder]

gpt4 is working great for me right now. How odd...

[heady zealot]

Could be. But it kinda makes sense right when microsoft invest 10 billion... And GPT runs on azure cloud from microsoft.

[wind urchin]

hello there

[ionic spruce]

Yeahh, but once you go GPT-4, you never go back...

🤣

[safe ruin]

Anyone else getting invalid url

[ionic spruce]

Yep, I am

[blazing scroll]

yeah that's seems true because 3.5 look outdated!

[ionic spruce]

It feels outperformed, too

:p

[eternal adder]

There are two 3.5s, default and legacy. Default is super fast.

[blazing scroll]

legacy is an old model before they introduced turbo

[eternal adder]

Ah okay

[ionic spruce]

Yeah

[charred night]

Hello guys

[sacred musk]

hello guys

[eternal adder]

👋

[charred night]

I am mst. Om
From India

[ember oracle]

HRU doing

I'm from Pakistan karachi

[eternal adder]

Sitting here trying to dream up something to create. Tried to create a command line RSA PEM to PPK key converter, but it had a lot of trouble with that.

[charred night]

Hi bro

[ionic spruce]

I wish I could use it right now, too. lol

[serene flint]

Is there someone who can give me his chatgpt account temporarily because the site does not work in my country?

[rotund pagoda]

try harder

[ionic spruce]

🤣

Account sharing is not allowed...

[blazing scroll]

It violate openai tos

[wind urchin]

what do you guys think about the latest AI news?

[rotund pagoda]

bro nobody is going to share their ms or google accout with you lmao

[sacred musk]

about what

[serene flint]

I'm From Egypt

[sacred musk]

for real?

[ember oracle]

Yeah but why we can't share?

[dull nova]

A little confused about GPT-4 token limits for 8k model.

Is the prompt token limit ~2k and output token limit is 8k?

[grand wind]

Q: Does any 1 person know exactly how chatGPT works? I suspect the answer is "no", as I imagine it's such a large team...

[eternal adder]

I just heard the other day on our local Fox news that many companies are now requiring their employees to start using AI assistants on a daily basis at work.

[wind urchin]

i guess all the large AI megaminds putting the AI development on hold for 6months

[blazing scroll]

input context 8k

#rules

[ember oracle]

Wow

[wind urchin]

this is cool as long as ppl can keep their jobs

[serene flint]

It's very sad that I can't try chatgpt

[eternal adder]

Yeah. It makes me 1000 times faster at mine.

[dull nova]

So total 8k?

[sacred musk]

I think it makes sense

[wind urchin]

nice 🤍

[deft monolith]

nah total is 32k

[blazing scroll]

You might be able to try using third party provider who using openai api

[ember oracle]

Try with vpn

[real minnow]

Hi, someone els happens that chat not working?

"Currently we are receiving more requests than we are are comfortable with! To try your request again, come back in a short while and reload this page." and in details "Global rate limit exceeded"

[blazing scroll]

yes

[dull nova]

Is there documentation on this? failed to find any

[grand wind]

It's openAI's goal for everyone to keep jobs and to have their tool help people do their jobs, but I don't see any way around people losing jobs... But, I've been saying for years that with all of our tech, not everyone needs to work.

[eternal adder]

When they were talking pricing models, I was like "I would pay $50/$100 per month for this easily!" Then they said "$20/month" and I fell out of my chair!

[near belfry]

Input context + response is 8k, no?

[coarse marsh]

Yeah I'm also getting "Invalid URL (POST /v1/completions)" when trying to use GPT-4 :/

[deft monolith]

[quartz apex]

Is the sever down or something ? The server experienced an error while processing your request. Sorry about that! You can retry your request, or contact us through our help center at help.openai.com if the error persists.

[ionic spruce]

Same

[serene flint]

It needs a contact number

[valid scaffold]

seems like the login page is also down now 😄

[coarse marsh]

ugh just when I needed it to work it doesn

[valid scaffold]

I cant log in to chatgpt anymore 😄. Snowballing..

[sacred musk]

Because Open AI has yet to figure out where GPT4's ability to understand comes from

[coarse marsh]

they yoinked my $20

[eternal adder]

https://status.openai.com/

[valid scaffold]

haha, all operational 😄

[ionic spruce]

Okay, well, maybe OpenAI will pay more attention now

[warped fjord]

l need a number to use GPT

[blazing scroll]

"gpt-4 has a context length of 8,192 tokens. We are also providing limited access to our 32,768–context (about 50 pages of text)" https://openai.com/research/gpt-4

[wind urchin]

that's so interesting to me. wish i could conduct my own research/ be involved in this kind of stuff, but my background knowledge is limited + lack of exposure / connections.

[ionic spruce]

I have yet to get API access to GPT-4. Soon, hopefully

[serene flint]

Does anyone know when chatgpt will be available in Egypt?

[eternal adder]

Same. We studied neural networks and basic AI in college. But I have a hard time wrapping my head around that topic.

[dull nova]

context length means prompt + output?

[wind urchin]

true, but the more ppl working on this kind of stuff the better it gets (as long as AI safety measures are being implemented)

[blazing scroll]

unknown https://platform.openai.com/docs/supported-countries

[warped fjord]

l need someone help

[tawdry rover]

do it anyway, it's a good way to help you train your understanding of that field

[blazing scroll]

not output prompt only as far as i know

[wind urchin]

you need a phone number from a sim card of your own

[dull nova]

I remember getting a 2k max token error from the API so that's why confused

[eternal adder]

Output definitely has some kind of limit. I always have to ask it (for long output) to continue generating from where it left off.

[dull nova]

Let me try again

[ionic spruce]

This is for chat.openai.com, not the playground version

[crystal bough]

I think it holds a socket open for the output or something, and times out if its open too long

just my guess

[eternal adder]

Yeah I am on chat.openai.com

[warped fjord]

l buy it nowhere

[ionic spruce]

Playground has different output limits

[serene flint]

I do not know what it provides after the service in Egypt, despite the service available in nearby countries such as Algeria

[eternal adder]

I am on the paid account

I assume by "playground" you mean the free account

[crystal bough]

playground is a different part of the site

[grand wind]

has anyone taken Andrew Ng's deep learning course on coursera? I'm doing it now. He's got some awards and a lot of important achievements.

[blazing scroll]

yes but i think it is not becuase of context size

[ionic spruce]

No, it's the direct API usage

[eternal adder]

oh! I don;t know about that one

[wanton valve]

Do you know a good site to detect AI's write content?

[eternal adder]

No I haven't used the API yet

[ionic spruce]

There's relatively easy ways to bypass this

[eternal adder]

I still use the chat.openai.com web page

[ionic spruce]

Using a complexity/burstiness prompt

[wanton valve]

Do you think they can send false positive ?

[near belfry]

Input token length should be (8192 - max output tokens)

[blazing scroll]

My writing style becoming similar to chatgpt as i'm using it to learn english! so i'm going to be detected as ai

[ionic spruce]

🤣

That's funny - network errors count towards usage limit

[crystal bough]

damn my chrome extension got published but I had to update it ;-;

[ionic spruce]

That's not fun

[wind urchin]

in my college application i literally talked about singularity and mentioned Yudkowsky's work and STILL got rejected. i felt so stupid and incompetent so i just dropped it.

[ionic spruce]

So if you get an 'Invalid URL' call it will count towards your usage limit of GPT-4, lol

[abstract notch]

how can I send images to gpt4?

[blazing scroll]

not currently available

[abstract notch]

noice

25 messages every 3 hours and no images 🙂

[eternal adder]

Yeah, and it isn't good at ascii art either

[valid scaffold]

anyone else unable to log in ? (if you want to try this, expect to be locked out 😄 ).
If you want to risk it, click log out, and then try to log in again

[eternal adder]

Heck no!

[abstract notch]

too risky

[ionic spruce]

I think OpenAI is just having lots of issues right now :p

[valid scaffold]

fastest growing software project ever? 🙂 not even azure can keep up ;P

[ionic spruce]

And since their timezone is in the Americas

It'll take a bit longer until they can fix it

[eternal adder]

🤣

[blazing scroll]

is not 25 enough?

[eternal adder]

Not for code generation

[ionic spruce]

25 does feel a little bit short if you are e.g. debugging an application

[abstract notch]

not in the slightest

[tawdry rover]

You shouldn't. Try again. Did you ask the reason why you was rejected? Tbh, there are tons of stupid people going to college, but with that theme you shouldn't count yourself as one.

[blazing scroll]

i guess unlikely becuase of the cost of server

[abstract notch]

^^^^^

[ionic spruce]

For code generation you definitely need more

But API access would make that easier. It does get costly though

[blazing scroll]

if they increase it to 50 then it will be enough for me

[ionic spruce]

^ Same

[abstract notch]

i heard 100

[ionic spruce]

100 is overkill tbh

[abstract notch]

defiently forseeable tho

[ionic spruce]

I hope so

[abstract notch]

computational power be increasing and such

[deft monolith]

can u tell what does this playground version mean..? Is this better than the web page chat.openai ??

[wanton valve]

Someone is writing on Textbroker with Chat-GPT ?

[wind urchin]

Unlimited is the dream

[ionic spruce]

Playground is just platform.openai.com

[compact dirge]

it was 100 every 4 hrs, then dropped to it's current level, it hasn't dropped below that, so it should increase (maybe) soon

[ionic spruce]

obviously

[zealous jasper]

how to use midjourney

[ionic spruce]

They said "expect reduced limits soon" in the message

[abstract notch]

is there anything worth trying

[wind urchin]

Use the thumbs down button on chatgpt

[abstract notch]

thats different from 3.5

[blazing scroll]

It might be the best technology that humans have currently so 20$ is fine i think

[compact dirge]

setting expectations doesn't necessarily guarantee that is what will happen

[ionic spruce]

Expect the worst and hope for the best

[compact dirge]

yes

[ionic spruce]

Doesn't really help :p

[eternal adder]

We can't post screens here? I was going to show you what I was talking about earlier about having to ask it to continue generating

[astral wind]

LMAOO ya’ll fell for the openai chatgpt plus thing aswell?

[ionic spruce]

Nope, no embeds

[eternal adder]

Darn

[deft monolith]

o ok

[wind urchin]

i havent cuz in that moment i felt so silly for talking about that kind of stuff in such broad detail only to get a rejection letter. i legit thought i was mentally ill for thinking i could make it in the tech world and refused to reapply again elsewhere

[wind urchin]

We know what you’re talking about. It’ll stop mid sentence and you have to ask it to continue.

[eternal adder]

Yeah

[astral wind]

I don’t remember seeing any mention of a message cap in the page that had chatgpt plus and being able to try gpt 4

[blazing scroll]

buy plus or if you cannot then maybe try bing which is based off some version of gpt4

[ionic spruce]

Until it said it at the bottom of the chat page, lol

[wraith ginkgo]

OpenAI staff will never ask for money. Beware of possible scams or fraudulent activities that you may receive through direct messages.

Please report any incident by sending a DM to @open crescent immediately.

[ionic spruce]

But considering these network issues I suppose it isn't too wild to reason that

[ionic spruce]

Bing honestly feels way less accurate for what I want, not going to lie

I'm not sure what it is

[blazing scroll]

Why you mentioend me?

[tawdry rover]

sounds like a hard experience. Just keep in mind, that you are lucky not being mentally ill. You only got rejected. Try again, with a new text 🙂 You will suceed

[astral wind]

Once you paid

[ionic spruce]

Hence I said "until" :p

[blazing scroll]

It is optimized for search but according to my testing it is fine but gpt4 is superior so it might be smallar version

[marsh walrus]

Is anyone having issues logging in? It just keeps redirecting me to the login screen!

[wind urchin]

Yesterday i asked chatgpt what was the highest latitude state of my country and it got wrong 3 times
with gpt-4

[ionic spruce]

Most definitely a small-optimized version for it

[wind urchin]

thank you, i will try to give it another shot 🤍

[astral wind]

Also kinda cheesed because the hype was talking about gpt4 being able to take in images and understand them. I can’t even use it for that yet.

[blazing scroll]

Still it is better than 3.5

[ionic spruce]

I guess that's good then

But for coding it seems to fall pretty short haha

[wind urchin]

.

[chrome oasis]

Hi everyone, can I ask, I have paid for the gpt plus but it have not activated for me. What should I do?

[astral wind]

Why is that being advertised like it’s something the model does now, in every media post there is a mention of it being able to understand images.

[astral wind]

check your email to see if you got a confirmation that your payment went through

[blazing scroll]

Because of personality it sometime reject to do coding but bing ceo said they are fixing the issues and hopefully it will be better in next 3 week

[ionic spruce]

Hopefully :p

[chrome oasis]

i havent but it showed the invoice and my money alr out

[ionic spruce]

Bing is definitely not as good as the normal GPT-4 though

[deft monolith]

does bing work well for u???

[regal ingot]

Bing AI is pretty stupid imo

[astral wind]

Compared to gpt 3.5, it’s much better. Atleast in the context of giving it more complicated prompts

[ionic spruce]

From what I've experienced yeah

[ionic spruce]

It's a middleground between 3.5 and 4

[deft monolith]

i guess not even gpt 3.5// atleast for me

[astral wind]

And you logged in the same account and can’t see options to change the model?

[hasty pebble]

What's app

[chrome oasis]

yes sir

[regal ingot]

I’d say even a 7B llama model is smarter than Bing

[astral wind]

Isn’t that just gpt 4? Like just 0.5 better than the previous version?

[blazing scroll]

Ethan Mollick has lot of examples in twitter to use ai tools right way. if use right way bing is also good but not as 4 as i said previosly

[ionic spruce]

Bing isn't as good as 4 though

[chrome oasis]

which email should I sent this issues TT

[ionic spruce]

[e-mail]

[wind urchin]

Hey guys, I have GPT plus but I can’t access gpt due to demand

[astral wind]

Email support team, you’ll probably won’t get a response soon tho.

[wind urchin]

Bronze Galloping Horse Treading on a Flying Swallow

[wind urchin]

I thought I’d get priority access no?

[astral wind]

Ooohhh I apologize, you were talking about bing. My big mouth, sorry.

[ionic spruce]

No worries :p

[urban flicker]

someone can give me like what apps can paraphrase so that my answer is sounds like not ai

[fresh orchid]

Me neither, looks like as server error

[wind urchin]

I got API access but when i try it, it says “You've reached your usage limit. See your usage dashboard and billing settings for more details. If you have further questions, please contact us through our help center at help.openai.com.”

[ionic spruce]

It's confusing, all these models

[eternal adder]

It should give you a link button that will send you a login link to your email.

[blazing scroll]

They got millions of users in just few months so that's why people see overloading errors

[wind urchin]

Do i have to pay?

[ionic spruce]

Yes

You do

[valid scaffold]

has anybody been able to log in (no cached sessions/pre-existing logins) in the last 15 minutes?

[urban flicker]

someone can give me like what apps can paraphrase so that my answer is sounds like not ai

[regal ingot]

Where

[ionic spruce]

You can just use a custom prompt for that

[regal ingot]

Always been ClosedAI

[astral wind]

Word tune

[urban flicker]

oh yeah? i dont know that hehe

[blazing scroll]

what you mean by it?

[wind urchin]

Do I check my email for that?

[blazing scroll]

It is not cheap to build best tech in the world so they do need money

[astral wind]

Even if it was open, I wouldn’t be able to run it lmaoo

[eternal adder]

When it tells you that it is busy, normally it provides a button on the right side of the same page that when clicked it sends you a login link. It's for plus users

[astral wind]

Someone said you need like 160 gb ram

[ionic spruce]

It's unfortunate that I have to use up one of my 25 prompts to GPT-4 in order to test if it is working

[regal ingot]

You could if you rented your own server