community-help

[native tendon]

You can't select Model for DALL E

[fading robin]

Dalle gpt i think he ment

[charred parcel]

go to https://chatgpt.com then

1. Log In
2. Switch to the GPT-4o model
3. Ask ChatGPT to generate an image
   create an image of a cat

[native tendon]

lets try

it does work for that prompt

but not when I used my prompt

Create a logo for the app "app_name". Design an icon featuring a circle with interconnected nodes, using a blue gradient for the main icon and green accents on the nodes. The text "app_name" should be placed to the right of the icon in Montserrat font, with a clean and modern look. Use a professional color palette with blue and green, symbolizing trust, technology, and growth. Ensure the design is minimalist, balanced, and cohesive.```

[charred parcel]

let me try that stand by

[charred parcel]

Works fine for me

I had it create a logo for an app called "Cat Facts"

[native tendon]

should I delete all chat and try again?

[charred parcel]

are you on a Mac or PC?

[native tendon]

PC

[charred parcel]

Ok, try creating a new chat and try again

[native tendon]

I just did.

[charred parcel]

did it work ok?

[native tendon]

Nope. same it just gives me a detailed steps and thats it

[charred parcel]

In the model chooser at the top of the web app, which model is selected?

[charred parcel]

that is odd, something isn't working like it should but I don't understand why

[halcyon crescent]

Hi all. Are we no longer able to set our Custom GPTs that are on the GPT store to private? Some of my public custom GPTs won't let me change the sharing setting any more.

[native tendon]

I just deleted all chats, and it stil wont work.

[charred parcel]

ok, it sounds like for some reason the model is switching to GPT-3.5, try logging out, clearing your cache and cookies and then log back in again and try

[native tendon]

I did both already, will try again.

[halcyon crescent]

I think I found the answer. They changed this so the most restricted is invite only. It's weird that the "only me" setting still shows up though.

[native tendon]

It works on ChatGPT 4 though.

[charred parcel]

does it work on ChatGPT 4o?

[native tendon]

No.

[charred parcel]

ok, I don't understand why

[halcyon crescent]

Err... no. Even when I try to set it to invite only, it is still showing up as available to everyone... 😦

[charred parcel]

I just tested one of my GPTs and I was able to move it to Invite Only

[halcyon crescent]

Interesting, thank you for the reply. I am not sure why I am not able to then?

[charred parcel]

these are the options I have

I was able to switch it to Anyone with the link and then back again

[halcyon crescent]

Yep, I have those same options displayed. It's really weird, because it will let me for some, and others it looks like it's updating, but when I check back with "my GPTs" they are still showing up as open to everyone.

[halcyon crescent]

A work around is to create a copy and set it to only those with a link, and then delete the original.

[hearty turret]

do you have to create gpts in 4 and 4o? I don't see this option

[charred parcel]

You create GPTs using 4o

[hearty turret]

where do I set the model?

[charred parcel]

are you on the App or the Web?

[hearty turret]

web

[charred parcel]

Up the top of the page you can choose the model from a dropdown list

[hearty turret]

yeah, but this is not Custom GPT, it is chatgpt, right?

[charred parcel]

yes ChatGPT

All custom GPTs use GPT-4o and you need to create them on the web

[twin cove]

For those who use ChatGPT on the Mac app, is it just me or I can't access my custom GPTs like I can on the web? Seems like I have to start a conversation on the web first with the custom GPT and then continue the chat in the app. 😦

[toxic sky]

Hey, it's a bit more confusing, you have to click on the model selector and then GPTs, which is different then the web.

[frank kindle]

I accidentally canceled my Plus subscription. It is still paid for the next month. How can I restore it? I contacted OpenAI support, but there is no option for a refund

[toxic sky]

They do not give refunds, also, how do you accidentally cancel a Plus subscription?

[frank kindle]

"In what sense do they not give refunds? I don't need my money back; I just want my Plus subscription reinstated for the month it's paid for. I canceled the subscription through 'reportaproblem' in iOS, where there were two identical amounts, and I probably selected the wrong payment."

[toxic sky]

Ah, I would just contact support then.

[frank kindle]

I'm trying to do that right now too.

[tame scarab]

Accidentally Cancelled Subscription? Bruh!

[tame scarab]

Since you've already paid for the next month, you should be able to get the subscription reinstated for that paid period without needing a refund.
Contact OpenAI support again and explain clearly that you want the existing paid subscription reactivated, not a refund. They should be able to help resolve this so you regain access for the month you paid for.

[frank kindle]

Can this be done through Discord? OpenAI takes an unrealistically long time to respond, I've been trying to resolve this issue literally all day.

[charred parcel]

the only way to contact support is through https://help.openai.com

[frank kindle]

Describe the full sequence of actions; maybe I'm doing something wrong? In my ChatGPT window, there's a "Resume Plus" button, but when I click it, it suggests paying for it.

[trail cairn]

you can keep continue using it for the days you already paid for

[tame scarab]

But unfortunately, resolving subscription issues can only be done through OpenAI's official support channels, not Discord. I know their response times can be frustratingly long, but you'll have to keep at it. Maybe try sending a follow-up message politely explaining the urgency of your situation. Hopefully they'll get back to you soon! Let me know if there's anything else I can help with in the meantime.

[frank kindle]

It says it expired on the next day, I paid not on June 13th, I have all the receipts.

[marsh urchin]

Can someone assist

I subscribed to CHATGPT PLUS but I don't have access to the additional features. Who do I contact

[tame scarab]

The "Resume Plus" button you see is for starting a new Plus subscription, not reactivating a canceled one. To resolve this, you'll need to,

1.Go to https://help.openai.com/en/
2. Click on "Contact Us" at the bottom
3. Explain your situation (accidental cancellation, want to reinstate for remaining paid period)
4. Wait for their response and further instructions

I know it's tedious, but going through the official support channel is the only way. Hope this helps clarify the process!

[trail cairn]

when did you pay and were charged for the next month? if you cancel mid you will still have access for the remaining half. i've done this few times

[tame scarab]

Hey new friend! If you've subscribed to ChatGPT Plus but aren't seeing the extra features, the best thing to do is reach out to OpenAI's customer support team. Here's what you can do:

1. Go to https://help.openai.com/en/
2. Scroll down and click on "Contact Us"
3. Explain that you've subscribed but don't have access to Plus features
4. They'll look into it and get back to you with a solution

[frank kindle]

Yesterday, the day after purchase, I canceled it.

[marsh urchin]

OK let me try that

[trail cairn]

did you pay with app store or via openai payment partner?

cause apple does it differently

[tame scarab]

Why? You Cancelled or Any of your friends?

Upcoming YouTube video title:

Top 10 The biggest Mistake in the life.

[frank kindle]

AppStore

[trail cairn]

did you lose access to plus features? its likely that openai will redirect you to apple customer support as they have no access, also its weekend

[frank kindle]

I canceled through "report a problem", there were 2 identical amounts, it seems I selected the wrong one, or Apple misunderstood what I wanted to cancel, but I've already canceled the second purchase.

[frank kindle]

So should I contact Apple support?

[trail cairn]

yes if you want to resolve it faster

[tame scarab]

Yup
What happens if I accidentally Delete my Plus Subscription Account?

OpenAI Helper: ....... Loading......

[tame scarab]

Did you get the Response from Apple Customer Support Service? Let us know!

[marsh urchin]

Looks like it's working now

Thanks for the help

[tame scarab]

They help you? ??? 😂

[frank kindle]

No(

[tame scarab]

lol

That's why I don't trust Apple

[vernal canopy]

Anyone else locked out of research gpt
Interaction with this GPT is not allowed

[charred parcel]

pop a screen shot in #off-topic

[ocean hornet]

Im trying to get a direct link for the images, but it keeps rewriting them just as unclicable purple words, what might be the issue ?

[charred parcel]

a direct link for which images?

[ocean hornet]

im currenly writing a bachelors thesis so that mostly advertising screenshots and charts as a refernces

[charred parcel]

sorry I don't understand your problem, can you be more specific?

[ocean hornet]

i cannot press thoose

[charred parcel]

they are hallucinations, you won't be able to click on them and get an image because ChatGPT has hallucinated the links

[tender timber]

Downforeveryone or just me?

Loads just basic website frame

[charred parcel]

working fine here

[muted blade]

Why chat GPT in Android stuck, after click send prompt

[proven surge]

Just to confirm with anyone who has this information:

In order to use openai api key, do I need to pay for it first?
I thought gpt-3.5-turbo api key has a free option, isn't it?
I tried to use openai api key, but it is "over the limit" constantly, despite i have never used it. (my usage is literally zero)

[charred parcel]

the API costs money and is prepaid, you need to add credit to use the API

[proven surge]

Got it. Thanks.

[hexed fulcrum]

Question: I temporarily cancelled my Plus subscription because I realized I was paying $2 extra for having subscribed through the app. But when I go to the web-based settings page to resubscribe, it says to manage my plan through the app. Does it only say this because my subscription is still active, or do you really have no option but to subscribe through the iOS app?

[frank kindle]

LOOOOL, apple refunded me for the purchase!!

I was already very upset because I am a student and I have no money at all, and ChatGPT helps a lot with my studies. I thought I would be hungry and sleepless for a month.

[tame scarab]

I can't believe it, anyways you said, Without ChatGPT you Would be hungry and sleepless for a month? Bruh!
OpenAI has the right to defend itself.

[tame scarab]

You might be seeing the message to manage their plan through the app because the subscription is still active. You should wait until the current subscription period fully ends. Once it has, you should be able to subscribe through the web without any issues.

[tame scarab]

What is your problem? You Can't send messages right?

[muted blade]

I can't send message in gpt (stuck after click button send) , but if I change another account, I can use gpt

[tame scarab]

It might be an issue with your specific account. Try logging out and back in, clearing your browser cache, or using a different browser. If it still doesn't work, contact support for help or tell us there :).

[gaunt cipher]

since 1 week, every time I ask my GPT a question it freezes in the middle of its response and does not continue. Using macos - tried on chrome and firefox, deleted cache and history more times than I can count.

[tame scarab]

Did your Freeze problem be solved?

[gaunt cipher]

nope. still happening right now

[tame scarab]

Are you using the ChatGPT app or ChatGPT web version? And did you have this same problem in Other Accounts? Are you using any Customize instructions?

[gaunt cipher]

web version. I only have my account, I'm not sure if it's happening over others. Yes, using extensive custom instructions. TRying to contact chatgpt support and stuck talking to a bot

[agile kettle]

Hi! I'm trying to fine-tune an AI for script-writing. I know that 3.5-turbo has an output of 4k tokens. This is not ideal as my scripts are relatively long. Has anyone had a similar experience with that that can point me in the right direction? Thank you!

[tame scarab]

Use the GPT-4 model: GPT-4 has a significantly higher token limit compared to GPT-3.5-turbo. Depending on the version you use, it can generate up to 8,192 or 32,768 tokens per request. However, keep in mind that GPT-4 is more expensive and has a slower response time compared to GPT-3.5-turbo. you can Use GPT-4o, it's faster.

[agile kettle]

I know about the token limit. I thought GPT-4 had a 4k output limit, same as GPT-3. I just googled, it's capped at 4096 tokens per request.

I also can't fine-tune GPT-4 yet as fine-tuning for GPT-4 is not publicly available.

[tame scarab]

You can find it from the OpenAI Playground API Website. If you were subscribed plus .

[agile kettle]

I was never a plus subscriber. These are my only options

[tame scarab]

Then you can Use GPT-4o.

Also it's improved in Speed and vision and more...

I'm faster......

lol

[trail cairn]

chatgpt plus has nothing to do with api. different services

[trail cairn]

he cant. fine-tuning gpt-4 models is invite only feature

[past adder]

Hi I am unable to use ChatGPT web version on my iphone as I can’t download it because I have older version of ios

I am unable to send message

[muted blade]

Thanks

[tame scarab]

I mean GPT-4, not the original version.

[tame scarab]

You are welcome

[fading robin]

Fine-tuning for GPT-4 is in an experimental access program - eligible users can request access in the fine-tuning UI when creating a new fine-tuning job.
Fine-tuning is currently available for the following models: gpt-3.5-turbo-0125 (recommended), gpt-3.5-turbo-1106, gpt-3.5-turbo-0613, babbage-002, davinci-002, and gpt-4-0613 (experimental).

You can also fine-tune a fine-tuned model which is useful if you acquire additional data and don't want to repeat the previous training steps.```

Original info about Finetuning from API Docs

> https://platform.openai.com/docs/guides/fine-tuning

[tame scarab]

Fine-tuning is available for several models, including gpt-3.5-turbo-0125 (recommended), gpt-3.5-turbo-1106, gpt-3.5-turbo-0613, babbage-002, davinci-002, and gpt-4-0613 (experimental). You can also fine-tune a model that's already been fine-tuned if you have more data.

Okey Okey, my mistake

[fading robin]

It does not mention gpt-4o codename

[tame scarab]

Let Me check

[tame scarab]

Hmmm.... Seems like there are no gpt-4 for free users.

[fading robin]

Gpt-4 access on API is prepaid as all models. Gpt-4 finetuing is experimental feature more info you can find in official doc > https://platform.openai.com/docs/guides/fine-tuning

[eternal dew]

Hihi need help, i have a rnd 10usd credit in my acct, but i cant even do a chat completion, it keep stating insufficient quota

[tame scarab]

Yup

[still matrix]

Hello, i have a problem when I speak to chatgpt sometimes it says 'thanks for watching the video', as if someone inserted an ad into the speech to text

It's been happening for months

[trail cairn]

that's "normal" as its been trained with a lot of "public" videos

[dawn oasis]

Can anyone share expectations when I will receive message from support? Already chatted almost 3 weeks ago, no valid answer still. Tuesday I sent an email to [e-mail] as that was all Chat would give me (after 2 weeks), but still no single response on the subject yet. This is getting so frustrating and unprofessional for such a large organization

[velvet frost]

I am in the playground. I have checked and everything should be good but I get this error. "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." -- Anyone have any idea how I can use the playground?

[mental nexus]

Hello, yestreday every custom gpts stopped work for me it tells im not allowed to interact with this gpt, login or register to interact but im logged in but cant use it

It functioned few days ago

[charred parcel]

Support has been slow, you just need to be patient, maybe follow up with another request at https://help.openai.com

[charred parcel]

you need to prepay to use the API

[charred parcel]

are you a free user?

[mental nexus]

Yes

[charred parcel]

then you have reached the usage cap, try again in a few hours

[mental nexus]

Its been a day

[charred parcel]

I don't know what to suggest, GPTs are working fine for me but I am a Plus subscriber

[mental nexus]

I will try tomorrow and than contact support if not function

[charred parcel]

ok good idea

[green dune]

Did free users lose gpt store?

[charred parcel]

try this -> https://chatgpt.com/gpts

[mellow estuary]

Can GPT call several tools in order? Like call the first tool, based on its output call the tool with the new parameters and so on until decides to respond?

[green dune]

Still dont work

[charred parcel]

maybe contact support via https://help.openai.com

[atomic juniper]

So, I'm not entirely sure where to ask this question, but I have a modicom amount of C++ programming knowledge. Not enough to really do much. I really want to make a program to help me sort files. I have been talking with openAI about a development plan for it. I don't want the program to touch the internet, I just want it to help me sort cbr and cbz files. I think it can be done, but how feasible would it be for ai to basically create the entire program for me? I have access to some programming AI software, I'm just very green to all of this and would need someone to help point me in a direction or tell me it can't be done.

[uneven raptor]

I need to vent a bit about something that’s been driving me nuts with the OpenAI transcription tool. So, I use it pretty regularly, and here’s the thing that’s really frustrating: I’ll be talking for like 2-5 minutes, and then I click stop. And then… poof, it’s gone. All of it. Just vanished.

Look, I get it if it fails to transcribe sometimes. That’s not my issue. And lately, I even see a retry button, which is exactly what we should have. But what really gets me is when it fails and the text box is just blank. I mean, come on! I’ve just spoken for 5 minutes, the recording is there, and it decides to just trash it? That’s so frustrating.

The recording should be saved until it can get transcribed correctly. Am I the only one who finds this unacceptable? It’s gotten to the point where I wish there was a competitor because I’d totally deal with a slightly worse model for a better interface.

It’s really disheartening. I’ll be on a roll, talking to ChatGPT, it’s responding back, and then I lose everything I said. I have to sit there and re-say it all, and it’s just… disappointing. This should be a simple thing. I thought it was just a bug, but it’s been six months and many versions later, and it’s still happening. This is so messed up.

Anyone else experiencing this? I’m using the iOS app on an iPhone 15 Pro Max, if that helps. Would love to hear if others are dealing with the same issue and if there’s any hope of it getting fixed.

Thanks for reading.

[slender imp]

Yeah that happens, i try to stop and see if transcribes up to some point if i need to talk for a long duration

[rugged wing]

I had similar issues happened to me few days ago, the interface is not responding when I press pause or stop, and sometimes it does not catch my input, and I didn't update the application, internet was fine, therefore it is totally a server issue.

[uneven raptor]

Thanks for sharing your experiences. I appreciate you taking the time to respond.

To clarify, the issue I’m facing is a bit different and quite specific. When I use the transcription feature, it records my voice just fine. But occasionally, when I click stop, the entire recording just disappears. There’s no transcription, no retry option, and the text box is completely blank. It’s as if the recording never existed.

I understand there could be server-side issues, but the core problem is that the recording should be retained on the client side, even if the transcription fails. Losing the recording entirely is incredibly frustrating, especially after speaking for several minutes. This feels like a fundamental flaw in the app’s design. Ideally, the recording should be saved locally until it’s successfully transcribed and sent.

As someone who uses this feature regularly, I find this completely unacceptable. It’s surprising that an app as advanced as this would have such a basic usability issue. This is not something I’d expect from a mature application.

I’m hoping to raise awareness of this issue so that it can be addressed in future updates. The app is fantastic in many ways, but this particular issue significantly impacts the user experience.

Thanks again for your input. Looking forward to any insights or solutions that might come up

[tame scarab]

Brother, how much time do you spend on typing this message?

[uneven raptor]

I may spend 2 to 5 minutes talking even 8 to 10 minutes sometimes, but that is really on the high end. I will say that it is so nice that everything works perfectly 90% of the time. The problem is that when the app fails it just completely loses the recording, which I feel like there’s no real excuse for. Like servers side e issues are always understandable but if an AI could actually understand and keep up with what is going on, so I don’t have to explain it and then it could also just natively handle any tasks related to that stuff

[frank kindle]

I wouldn't sleep because of the amount of assignments I need to do for university and I wouldn't eat because of the money I spent on Plus.

[tame scarab]

Nice poem 👏
Good Job!

[wooden shadow]

Something happened to one of my chats in 4o. ONLY this specific chat. No matter what, any prompt it completes, it gives me an error stream.

[trail cairn]

too long threads may give that and can check it by editing one of the earlier messages to wipe some messages

[wooden shadow]

Its only 140 entries, i have threads that are over 400.....

No i cannot just edit one of the earlier messages it will screw up everything

[trail cairn]

message length is also one factor so if they are all very long

[wooden shadow]

they are all VERY long in the ones that are over 400.

[trail cairn]

checking the thread via mobile app might also give another error reason

[wooden shadow]

And i can STILL get more messages off it just fine

it says the same thing either a 413 error or "this conversation is too long please start a new one." meanwhile i can get more messages just fine on all my other chats well over 200-400 messages.

This is so dumb, now i have to go, copy paste my entires thread prompt by prompt into another chat. Thanks alot Useless OpenAI devs.

[trail cairn]

thread size = messages * message size (hidden tool output eg. file search + visible text)

yeah its annoying

[wooden shadow]

for the last time, ALL my other chats i stress test the F out of them using maximum output length. THIS one i am only using that gpt 4o generates for me, 0 edits or refreshes.

Stop trying to blame me for it please. I'm done here, got nothing more to say

night

[trail cairn]

you cant compare because you cant see all the involved text

[trail cairn]

i never did

[dusky socket]

How can I disable that auto formatter for my messages that include a code?

Every time AI thinks I’m submitting some code in a language the code isn’t in and therefore messing everything up and providing wrong inputs for the chatGPT to understand from my message

For example, I paste a code of M query language and it reformats it as Java and also excludes some parts out of it leaving the excluded parts as text and then when chatGPT provides the response, it does it only in the formatted code

Found a workaround of not giving any line breaks anywhere in the message, then the website does not do this weird reformatting

[mental nexus]

@charred parcel is here any links to support?

[tame scarab]

@charred parcel Darth Vader needs your help

[mental nexus]

Xd

Or anyone else if know

[gleaming meteor]

is chatgpt failing for anyone else?

[brisk tide]

Sorry an unknown error occurred

[gleaming meteor]

im getting "Your most recent request failed. Please retry!" constantly

it seems to work fine on my mobile app

ah its fine now

[brisk tide]

With voice on mobile I still have this error

But only on one chat strange

Found the error uff rip dnd adventure

[mild phoenix]

The OPENAI Key doesn't seem to work with Autogen 2.0 hasn't anyone encountered this problem?

[lofty bison]

chatgpt dead?

[gleaming meteor]

yes and its not coming back to life

[lofty bison]

[mental nexus]

where is support team? any links ?

[tame scarab]

lol, I'm still sleeping.....

[tame scarab]

Malto:[e-mail]

[e-mail]

[e-mail]

[mental nexus]

Please help, i cant use any of public GPTs it tells me im not allowed to interact with this gpt or it tells me i have to login or register to start chat, but im logged in, it happend day ago

[tame scarab]

Did you have the ChatGPT app version? Try it on the App version if you are a free user.

[mental nexus]

on pc?

it dont work on mobile app too

[tame scarab]

Wait what! Seems like your Limit goes end. Let me check my Other account.

[mental nexus]

yea, but its three days i cant talk to it, even on another account

[trail cairn]

that email should not be used or shared. it does not provide support

[tame scarab]

Why not? Whaaat. Then why is it used on Websites? Bruh, really making me clown

[trail cairn]

openai staff member told it here to not to use it and mailing there will just give a link to the website

[tame scarab]

🫡🫡🫡

[twilit saffron]

Help me

[gleaming meteor]

same here

seems lots of issues past few hours

maybe updates coming??

[twilit saffron]

It was the last sentence that caused problems

He must be depressed because thai talk to me

[neon kraken]

It's down? My gpt doesn't work

[gleaming meteor]

probably

cos same

[full hollow]

I think so. I've tried and my laptop and phone can't connect.

The ui loads but messages don't send

[twilit saffron]

I have problems too

[neon kraken]

My chatgpt doesn't answer

[jolly kelp]

it's down for me

[carmine cypress]

Hello, I have noticed extreme slowness in ChatGPT. The websocket fails to respond in the given time.
wss://chatgpt-async-webps-prod-westus-4.chatgpt.com

[south cairn]

yeah status says operational but its offline

[proven snow]

zzz

[tame coral]

it's down

[south cairn]

ChatGpt that is

[carmine cypress]

great statuspage 👍

[neon kraken]

I can connect and write question it thinks and then nothing

[gleaming meteor]

its finally going skynet

[tame coral]

lol

[carmine cypress]

crazy

[south cairn]

there I fixed it lol

[carmine cypress]

dayum

I love how OpenAI has no redundancy

[neon kraken]

❌ is for chatgpt not for you

[proven snow]

huh?

[carmine cypress]

like how can ALL regions be down

I mean the engineers probably know it better than me but still strange

that such a big corp is just dead in like 10min

[neon kraken]

it's down!

[sand sorrel]

😮‍💨

[golden veldt]

OMG I have so much work to get done. Please wake up GPT!

[austere rapids]

welcome to chatgpt

[final frigate]

Oh, well that answers my question. I thought it was just because I needed to pay my subscription fee 😂

[austere rapids]

we are paying for timeouts

[final frigate]

How many bottles of water do you think are being used now versus when it's live 🤣

[austere rapids]

a lot

[final frigate]

Take that eco terrorists! Er, I mean...

🤣 Well what can I do except laugh

[mellow helm]

same problme here

[mellow helm]

how long is this gonna last

[modest ingot]

here dont work too

[weary void]

This has been happening for a few hours now

[compact saddle]

come on!

[little snow]

same problem

[empty fox]

No announcement or status update?

[gleaming meteor]

gpt 5

[autumn torrent]

is chatgpt down.. again?

[sand sorrel]

3.5 working a little sometimes.

[mint kelp]

Anyone knows how long this is gonna last?

[dusky socket]

@burnt galleon any chance to explain why this happened?

[gray pewter]

Put it in a code block like this

```java
code here

Since it’s just a markdown renderer

[dusky socket]

It isn’t Java.
I didn’t specify which code it is. And even when I tried (the same way as u proposed) it still chose the wrong language and still messed the prompt up

[dusky socket]

^

[gray pewter]

I can’t really test it rn due to error rates on chat but when I enclose it in a code block it doesn’t render as a random language (the hyml was a type of html, but that just proves you can set it manually) at least in terms of how it renders. Not sure what CGPT is seeing on the backend until they fix servers but ask it to repeat your input verbatim and put it in a json code block, experiment a bit

[weary void]

Can we get an update for the fix?

[velvet frost]

Are there any websites or services where I can use my own API key to use ChatGPT? Feeling ripped off with Plus and would like to just pay per use.

[twin cove]

Thank you!!!

Do custom GPTs, created before the launch of ChatGPT 4o, run on ChatGPT 4o or ChatGPT 4? I'm a Plus user if it makes any difference.

I read on Reddit that a switch is incoming but I'm not sure if I can check which version it's on

[mystic gorge]

Hi there, I hope this is the right place to ask this. My account is showing up as deleted. It coincided with the outages a couple weeks ago so I would assume it is related. I pay for plus and havent had access to my account in over 2 weeks (I had to create an alternate account just to get in here). I have tried using the support page and all I got was a generic message about creating a new email, I have not heard back from support in over 2 weeks. Does anyone know how to get in touch with them? I tried emailing and just got redirected to that god awful chat bot.

[dusky socket]

The support page does tend to have human interactions. I’d suggest using that method.
Although I have had lack of getting responses 100% of the times, they do respond some times

[mystic gorge]

Man that’s rough, I’ll try contacting on the new account I suppose. It’s possible they deprioritize guest accounts in the help chat. Do any staff members ever show up in the discord?

[charred parcel]

OpenAI staff have yellow/gold usernames and are sometimes seen in chat

[wide grove]

Hi there! Noob here. Trying to work with this code (attached). I understand I have to migrate it, and I followed the automatic migration with grit on Windows guide on openai's github. But WSL shows: "Processed 1 files and found 0 matches". Is the code incompatible with the migration? What could I be missing? Thanks in advance!

[untold dune]

can you convert the code to just using code boxes so we can copy it and try it for ourselves?

please?

[wide grove]

Someone from the api-discussions channel told me this is langchain-related so I should ask my question from the langchain discord instead. But I'll try using code boxes.

[untold dune]

i can try to use a image to text converter then post the code here, k?

oh alr

from gpt_index import SimpleDirectoryReader, GPTListIndex, GPTSimpleVectorIndex, LLMPredictor, PromptHelper
from langchain.chat_models import ChatOpenAI
import gradio as gr
import sys
import os

os.environ["OPENAI_API_KEY"] = 'Your API Key'

def construct_index(directory_path):
    max_input_size = 4096
    num_outputs = 512
    max_chunk_overlap = 20
    chunk_size_limit = 600

    prompt_helper = PromptHelper(max_input_size, num_outputs, max_chunk_overlap, chunk_size_limit=chunk_size_limit)

    llm_predictor = LLMPredictor(llm=ChatOpenAI(temperature=0.7, model_name="gpt-3.5-turbo", max_tokens=num_outputs))

    documents = SimpleDirectoryReader(directory_path).load_data()

    index = GPTSimpleVectorIndex(documents, llm_predictor=llm_predictor, prompt_helper=prompt_helper)

    index.save_to_disk('index.json')

    return index

def chatbot(input_text):
    index = GPTSimpleVectorIndex.load_from_disk('index.json')
    response = index.query(input_text, response_mode="compact")
    return response.response

iface = gr.Interface(fn=chatbot,
                     inputs=gr.components.Textbox(lines=7, label="Enter your text"),
                     outputs="text",
                     title="Custom-trained AI Chatbot")

index = construct_index("docs")
iface.launch(share=True)

i copypasted urs with logger B)

anyways im gonna see what i can do!

[wide grove]

Thank you!

[tame scarab]

Amazing..... Excellent.... Super......

[untold dune]

import gradio as gr
from gpt_index import SimpleDirectoryReader, GPTSimpleVectorIndex, LLMPredictor, PromptHelper
from langchain.chat_models import ChatOpenAI
import os

os.environ["OPENAI_API_KEY"] = 'Your API Key' # put your API key here

def construct_index(directory_path):
    max_input_size = 4096
    num_outputs = 512
    max_chunk_overlap = 20
    chunk_size_limit = 600

    prompt_helper = PromptHelper(max_input_size, num_outputs, max_chunk_overlap, chunk_size_limit=chunk_size_limit)

    llm_predictor = LLMPredictor(llm=ChatOpenAI(temperature=0.7, model_name="gpt-4", max_tokens=num_outputs))

    documents = SimpleDirectoryReader(directory_path).load_data()

    index = GPTSimpleVectorIndex(documents, llm_predictor=llm_predictor, prompt_helper=prompt_helper)

    index.save_to_disk('index.json')

    return index

def chatbot(input_text):
    index = GPTSimpleVectorIndex.load_from_disk('index.json')
    response = index.query(input_text, response_mode="compact")
    return response.response

iface = gr.Interface(fn=chatbot,
                     inputs=gr.inputs.Textbox(lines=7, label="Enter your text"),
                     outputs=gr.outputs.Textbox(),
                     title="Custom-trained AI Chatbot")

path for your documents
index = construct_index("docs")
iface.launch(share=True)

i tried my best :3

[untold dune]

@wide grove

[wide grove]

Thank you! Trying it now

[untold dune]

You're welcome!

[alpine terrace]

[toxic sky]

All of them are not GPT-4o

[cold pebble]

Hey, could anyone help me out with the assistant v2 API and why this snippet doesn't work?

def send_to_openai(image_base64):
    assistant_id = "asst_id"
    user_prompt = "Prompt"

    def send_request(image_base64, user_prompt):
        try:
            # Retrieve the existing assistant
            assistant = client.beta.assistants.retrieve(assistant_id)

            # Create a thread
            thread = client.beta.threads.create()

            # Add a message to the thread
            message = client.beta.threads.messages.create(
                thread_id=thread.id,
                role="user",
                content=[
                    {"type": "text", "text": user_prompt},
                    {"type": "image", "image_base64": image_base64},
                ],
            )

            # Run the assistant
            run = client.beta.threads.runs.create(
                thread_id=thread.id,
                assistant_id=assistant_id,
                instructions=assistant.instructions
            )

[trail cairn]

assistants don't currently support base64

[cold pebble]

thank you, that clears things up!

[cold pebble]

How would you recommend sending pictures? Just uploading them to imgur or something?

[trail cairn]

you can use file id or external url like your own temp link or imgur if the images are not private

[terse spire]

Hi I am Carl, I am from a MSP and asking for help on behalf of a client who wants to use Chat GPT for teams, is there a place I can get some help on their integration? Specifically collaborating with files and prompts between team members

[trail cairn]

there is not really anything special beyond link and gpt sharing within workspace

[next orbit]

Is the screen sharing feature available yet for plus subscribers with the app on Mac?

I know it was rolling out to "some customers" but I wondered if everyone with plus already has it or it truly still isn't available for all plus subscribers.

[toxic sky]

Technically you can create a GPT that is accessible only to there workspace for sharing files.

[mental nexus]

Please help, i cant use any of public GPTs it tells me im not allowed to interact with this gpt or it tells me i have to login or register to start chat, but im logged in, it happend day ago

[red flower]

Is working chatgpt for you?

mine is down.

[mellow estuary]

Why there are such stupid restrictions that if links are obtained from the user file, GPT can't access them anymore, but if the links initially are provided during the conversation, GPT can access?

[red flower]

"Hmm...something seems to have gone wrong.

"

[mellow estuary]

I have a JSON file, that contains links too. I asked GPT to filter data and then to analyze the filtered PDFs from links. However, it respond that can't access pages anymore and asked to upload files manually! In the same time I simultaneously provided the links to the same GPT in another session (chat) and it accessed it.
What's the reason? Why I need two sessions for it?

[mental nexus]

THERES REALLY NOT A SINGLE SUPPORT TEAM???? I have to just read that articles really????

[trail cairn]

if you are a free user then they have reduced gpt access for many users after recent issues

[mental nexus]

If i get to limit, will it reset after some time or forever?

[untold dune]

greedy devs

[trail cairn]

from what some others have said there is no way to regain access even if you wait all your cooldowns

the removed access is most likely temporary until x

[mental nexus]

What??

[trail cairn]

x = when there are no longer disruptions for paid users, a day, few days, a week, or something else

[shy leaf]

Do you guys have any good ai detectors I can use that are somewhat accurate?

[mental nexus]

But i dont have 20$ per month for gpt 💀

[opal bough]

hello everyone! i have an issue regarding attaching photos in gpt 4o. it says hm... seems like something went wrong or things like this. tried everything from changing network to clear cache cookies etc

@trail cairn

i have paid for it please!

[rugged wing]

Can't blame them. If you didn't pay me I would not give you anything.

[opal bough]

yet no one can help.

[charred parcel]

can you try on the desktop computer?

[opal bough]

trying rn

nope, nothing

[charred parcel]

can you try using a fresh install of Firefox?

[opal bough]

trying

same

[charred parcel]

ok this is an issue for support, contact support at https://help.openai.com

[opal bough]

the problem is i dont know where i could talk to a person\

can u show me

bots always shows other issues than mine

[charred parcel]

ask the chatbot that you want to send a message to support

on the help site

[opal bough]

wich one

[charred parcel]

if you keep clicking around in the support chat on the lower right hand side of the page you will eventually be able to leave a message for support

[opal bough]

can u give me an example please i really need this for my exams

[charred parcel]

I never use the help site, but I do know that if you use the support chat in the bottom right hand corner of the help site that you will eventually be able to leave a message for support

 
If you already have an account, simply login and use the "Help" button to start a conversation. 
 
If you don't have an account or can't login, you can still reach us by selecting the chat bubble icon in the bottom right of help.openai.com.```

[opal bough]

the thing is..

also how do i check my gpt 4o limits

i have plus\

[charred parcel]

GPT-4o is 80 messages per 3 hours. GPT-4 is 40 messages per 3 hours

[opal bough]

is this the right way?

yeah, i passed the 80 messages

[charred parcel]

then you have to wait 3 hours for the limit to go away

[opal bough]

but i did

the issue occured at 8 pm,

and now is 1 am

[charred parcel]

then you should be fine

[opal bough]

ik but im not

[charred parcel]

that's why you need to contact support on the help site

[opal bough]

check this. i think it should be the right way if i am not wrong

[charred parcel]

yes that is how you contact support

[opal bough]

good

how long it usually takes

[charred parcel]

it can take days for someone to reply to your support request

[opal bough]

my exams are close

[charred parcel]

you shouldn't rely on ChatGPT to help with exams

[opal bough]

ik but eh, some things are pretty hard to understand

the thing is i only used a lot of photos uploaded on 4o

so chatgpt as text works

[charred parcel]

I just checked, image upload works for me

[opal bough]

but why for me nott

[charred parcel]

I don't know, which is why you need to contact support

[onyx gust]

I recently purchased the chat GTP 4, but unfortunately, it doesn’t seem to be working. Whenever I try to use it and hit the enter key to ask a question, the search button becomes unresponsive and I’m unable to type anything

I changed the language and cleared the cookies as shared by others online, but it still didn't solve my problem.

[charred parcel]

what OS and Browser are you using?

[onyx gust]

My browser version is: Chrome Version 126.0.6478.115 (Official Build) (64-bit)
It is a new version.
cleared cookies and logged into my account in incognito mode, but still can't send messages out.
Changed language settings, browser language, etc. as described on the web, but still no solution.
no hints, no errors, just can't send messages, after refreshing the dialog box can enter text, but once you click on the send button on the right side, it is grayed out, and the text box is stuck.
There is no problem with the internet, all websites work, my friend's account works without any problem, but mine doesn't work no matter what I try.
Please take my problem seriously and solve it, thank you very much.

[charred parcel]

I can't solve the problem but I can help you get things working. Could you download a fresh version of Firefox and try with that?

[onyx gust]

okay, thanks, i will try it. But I always want to be able to solve this problem, where can I contact to be able to solve the human customer service?

[charred parcel]

You can contact OpenAI support at https://help.openai.com

[onyx gust]

I have contacted through this but have not received a response, is it waiting for them to respond via email?

[charred parcel]

OpenAI support requests can take time, sometimes several days.

[onyx gust]

Too bad, downloading Firefox and edge browsers are still both not sending messages.

[tough kestrel]

Yo guys why can I not send messages on IOS safari browser?

I use ios 15.8

So I can’t use app

[charred parcel]

what version of Safari is it?

[tough kestrel]

How can I check?

[charred parcel]

Go to Settings -> Safari -> About

[tough kestrel]

I can’t find that

[charred parcel]

I think your iOS is too old

[tough kestrel]

It was working fine before

Pretty sure it stopped after the 4o update

[charred parcel]

yeah they updated the Web App

[tough kestrel]

No way around this..?

[charred parcel]

new iPhone?

[tough kestrel]

Getting one this summer

But why does my iphone version matter

[charred parcel]

because you're running an almost 4 year old iOS version

[tough kestrel]

Like it was working but they decided to just not update the app for us yeah..?

and just blacklist us lol

[tough kestrel]

Student life I guess

Whatever ty for help

[supple steeple]

friend do you have an ad blocker or anything that could mess with cookies?

[supple steeple]

this shouldn’t matter just go into desktop mode lol

[opal bough]

nop

[supple steeple]

it says in the browser that they may be a cookie issue. Try clearing your cookies and reloading or go into incognito mode.

there is also a CORS issue but….

cant change that

[opal bough]

its related to my account. ive tried a college mates account and it worked on my browser and all my devices

[supple steeple]

ahh, contact support.

[opal bough]

i did.. now waiting for answer

[radiant dragon]

Hi. As you guys can see I have always made sure to have the data controls off (cause I seriously don't want to train this AI thing at all and much less with my stuff). Anyway, I noticed that across my devices (PC / Phone / iPad) despite having that turned off the chat history was "on" and I had to manually "clear it". How is that so?!

Hopefully the devs can actually give some insight about how privacy and data works. Cause, again, I have the training turned off and I even opted out permanently so I am angry (...) to say the least that this keeps happening as it if was forced when it shouldn't at all.

I just don't want to always open a temporary chat (despite having turned off the no training model).

Thanks.

[dire kernel]

have they changed the daily limit of images for DALL-E via ChatGPT Plus?

[charred parcel]

no, they have not, it's the same

[dire kernel]

well, it's 50 a day right now, and I pretty sure it used to be 200... maybe I am crazy @_@

[charred parcel]

I generate a lot of images, I am pretty sure I do more than 50 a day sometimes

[charred parcel]

oh, you're using the DALL-E GPT, that would be different

[dire kernel]

ah...

[honest timber]

Hi, anyone know where to change voice in the calls with customGPT? I been using juniper before and it still selected in settings. It works for chatGPT calls and even for customGPT when using read aloud. But today when use call to customGPT, suddenly it is a new voice that gives me the ick and can't find no where to change it since voice from setting doesn't affect it. Its the same for all customGPT conversations. I'm using the app on android. tried clear cache, force stop, restart phone, reinstall app, but it still don't work like it did yesterday 😟

[charred parcel]

GPTs have their own voice option named, Shimmer, that is distinctly different from the 5 output voices available to use when having voice conversations with ChatGPT.

[honest timber]

Oh, thank you for response, is that a recent update? Juniper worked on mine until today, when suddenly changed.

[charred parcel]

I believe it is recent, yes.

[honest timber]

Ah okay. That's sad, I prefered juniper. Oh well, atleast then I know why. Thank you for helping 😁

[last carbon]

there is a way to reduce the token usage on asistants ? i meen im sending a simple "hello" and its taking me 384 tokens in ( im using some function but i need them ) without the function it take ( 80-120 token) exple of functions : { "name": "request_order", "description": "Request an order for disponibility it return an time where it is", "parameters": { "type": "object", "properties": { "orderName": { "type": "string" } }, "required": [ "orderName" ] } }

[late mantle]

chatgpt is down for me, chats do not appear. anybody else having issues rn?

[toxic sky]

It works fine for me.

[late mantle]

it's working again

[toxic sky]

👍

[elfin basin]

Anyone who is on the plus plan like me but still doesn't have the new voice mode? 😫

[timber shadow]

You're definitely not alone there! I don't think the rollout has started yet: https://help.openai.com/en/articles/8400625-voice-chat-faq

GPT-4o real-time voice and vision will be rolling out to a limited Alpha for ChatGPT Plus users in a few weeks. It will be widely available for ChatGPT Plus users over the coming months.

[elfin basin]

I was looking forward to using it on my vacation as a guide, but well it seems like that wouldn't happen

[rugged wing]

is got down? 😫

[winter dust]

I am building a feature for my app using Dall-e-3. I wanted to be sure that content policy and nudity won't be returned. I used a prompt with the phrase "naked". In the API I don't get an error, but I also don't get a naked person. But when I do it in the chatGPT, I got a topless woman (it was cartoonist - not realistic, but still). Has something changed recently?

[past adder]

Hello everyone there are some serious issue in chatgpt web version on iphone I am unable to send prompt to chatgpt have a look at this screenshot I am getting same problem in every browser

I’ve been seeking solution for this problem since last few weeks in this channel but no one replied but this time please tag someone from openai

[timber shadow]

Hey! Have you tried in the iOS app, or in Safari? I think Brave might have some extra things going on that could be interfering with normal operations of the site, just a guess. I know you said you've tried it in multiple browsers, just wondering if that includes the above yet!

[past adder]

My phone does not support iOS app of ChatGPT and I faced this problem for the first time on Safari then only I tried on other browser.

I’ve tried on browsers like Edge, Chrome, Opera GX, Google App, Firefox

Almost every broswer

[timber shadow]

Do you have any adblocker or VPN apps installed on your phone?

[hollow jay]

Hi, I have a developer account with my company and we noticed that we were no longer owner of the account. We were demoted to read and someone took ownership. How can i take ownership back. I have the credit card registered on that account. Does anyone has any info? I wrote to support but no response.

[past adder]

Nope

[timber shadow]

Do you have another device you could test on?

[lilac fable]

How do I access sora ? Or is it not from gpt ?

[trail cairn]

its not a public use product

[lilac fable]

So I will have no access whatsoever ;-; ? Do you have any recommendations for alternatives ?

[trail cairn]

its not released yet and you cant ask alpha access. you can discuss about alternatives in #ai-discussions

[lilac fable]

Oh that's ok thanks for replaying

[austere vine]

Hey I'm sorry to disturb.
I live in the EU and it says there is an option to get a refund. Yet when I go to help.openai.com, there is this instruction:

Contact us via our chat widget in the bottom right of your screen, select "Billing", then "Other", and then "From Europe and according to European law, I can request a refund within 14 days". We will respond to your inquiry as soon as possible.
Yet in the chat widget there is no "Billing" or anything like that and the "Ask a question" prompt has only some options and you cannot type there.
Does anyone know what to do in this situation? (Thanks in advance)

[trail cairn]

if you just click around them randomly it will allow you to write at some point

[austere vine]

here?

[trail cairn]

yes, like click the learn more

[austere vine]

I did that a few times and it gives me this. Could openAi just not do it nomore?

[timber shadow]

To get the Payments & Billing option I think you need to sign in on platform.openai.com first, then return to help.openai.com after that, so you're signed in. Signed out you have different options like you see in your screenshot

[austere vine]

when i first click the widget i get "write us a message" as soon as i go to press it, it says "ask a question" and also i am conected on platform.openai.com

[timber shadow]

Here's what I'd recommend:

• Close the help.openai.com window you have open currently
• Go to platform.openai.com
• Find the log in button there, and sign in
• After you're signed in, return to help.openai.com
• Open the chat widget
  You should have different default options, including Payments and Billing, after you sign in first

[austere vine]

gotcha thx if that dont work i could potentially email them on [e-mail] although that would be my last resort given that this email is for the digital services act

[scarlet mesa]

When will Sora be available to openai subscribers?

[timber shadow]

No timeline shared in that regard currently! https://help.openai.com/en/articles/8958981-how-to-access-sora

At this time, we don’t have a timeline or additional details to share on Sora’s broader public availability.

[wicked pilot]

Is chatgpt down?

[rigid musk]

Status.openai.com

[wicked pilot]

Cant login

[rigid musk]

Can you share a screenshot of the error ?

[rapid herald]

It's down AGAIN??????

This is like the 3rd time in a week. What the hell? Time to find an alternative

[charred parcel]

It's not down and hasn't been down for days

[dense hill]

dalle is having problems :/ is this occurring for anyone else?

It was working fine earlier but now it won’t edit images anymore

[rigid musk]

Sword are often against tos

[dense hill]

It’s just a katana from metal gear rising ._.

There’s no blood in the image

This was the image

:/ even ChatGPT says it’s not breaking guidelines

[rigid musk]

Chatgpt as the tis specify it can't be use as reference

[dense hill]

Yes but Raiden is just holding a sword…

He’s not killing anyone

[rigid musk]

Also dalle 3 have issue generating pictures based on a brand it can also be that

[dense hill]

I haven’t seen Dalle3 yet. I’ve only seen the regular dalle

[rigid musk]

The one from chatgpt is dalle 3

[meager grotto]

Is there a way to disable the formatting for my own messages somehow?

A couple of weeks ago, it didn't auto format my messages, but it does now, and it's kinda annoying when I wanna copy my code again.

[solid anchor]

Anyone got the bug that every prompt with an image errors?

both ffox and brave

happened for a few days now

[sacred flower]

4o is by far, worse than 3.5 in my opinion.

[placid lily]

3.5 Also couldnt play chess. Silly take tbh.

[oblique merlin]

Hello everyone. Hope you guys are all well! 🙂

I have a very essential question - something I have been struggling with (and, frankly, something I couldn't be more "ecstatic" to get the solution for). Therefore, if I get answers to this of potential solutions - that would be wonderful for me.

Basically, I'm in the process of creating an AI powered copywriting software: one that lets users produce content for stuff like posts, captions etc.

And, I've got Open AI as my API.

However, there's one problem I've encountered:

My aim is to keep all the "prompts" (that will be used to facilitate users with the best possible outputs) secure and hidden from the developer.

How is this done? Is there a solution?

As I'm sure LLMs out there, created by names like Taplio, don't disclose their prompts to their developers.

To sum this question up in easier words:

Is there a way to ensure that there's some sort of portal that the developer doesn't have access to? Or a code that I give to a developer, and he uses that (which will have the actual propmt undisclosed) without the developer being able to retrieve the actual prompt?

If any of this is possible…

Or any of you folks have another (tried and tested) method that could work…

I would be over the moon - and super grateful - to get some guidance.

Thank you all so much for reading this, and helping me out!

Cheers,

Hammaad

[charred parcel]

we can't help with Claude as this is an OpenAI Discord Server

[oblique merlin]

Sorry, I meant, Open AI as my API - it doesn't really matter since both use API.

(We have both running in the backend, so one strategy of securing prompts from the developer helps with both)

[charred parcel]

If an LLM has a prompt there is no way to keep it secret

[vernal dragon]

you can fine tune it on responses that do not disclose it.
a filter network would work too. phi3 is quite good at this and small enough to run on pi's

but if your business model depends on prompt engineering, oh u r in for a ride

[floral elm]

secure and hidden from the developer.

I promise a developer doesn't care about the prompts and is more than capable of making a handful to do the same function themselves if they really wanted to

[oblique merlin]

That is very unfortunate - so no way at all to keep it secure i guess? 🤔

[oblique merlin]

😎

[oblique merlin]

😩

[patent dock]

how do i fix it i cant mgs there are many important things i have there

[rigid musk]

Are you using a vpn ?

[patent dock]

no

[rigid musk]

What browser are you using ?

[patent dock]

Edge

[supple steeple]

prompt sandwich it to not tell it. Best approach you have, but not perfect since any half decent prompt engineering will get the prompt lol

if you want to know more about finding the system prompt there is a ai ctf i used to get better at prompt engineering

[spark bronze]

chatgpt down? why cant i login

it's showing blank page

oh wait fixed

[patent dock]

now this

[heady cypress]

is there an open ai ap that generates music?

[serene sky]

Hello -- I Just joined -- anybody know why am I getting http_response status:429 "You exceeded your current quota"
when I'm making no more than
one request per minute and
less than 200 per day and
request size is less than 1,000 characters ?
I'm using the FREE account for now.
To GPT-3.5-turbo

[obtuse storm]

Hello, around 15 days every month I don't have access to the chatGPT. Do you have the same problem? I get a message something is wrong, try again to regenerate the response

[green marlin]

why does chatgpt answer in american english even when told to use British English in the Instructions?

[twin acorn]

I did not realized that CHATGPT Plus has a limit on how many images we can only generate per day. I thought its unlimited. Just today I signed up to plus. So sas. 😦

[trail cairn]

https://platform.openai.com/settings/organization/billing/overview

[scarlet mesa]

Hi, as a ChatGPT Plus user, what are my monthly or daily limits for ChatGPT3.5, GPT4 and GPT4o?

[charred parcel]

GPT-4 is 40 messages per 3 hours, GPT-4o is 80 messages per 3 hours

[scarlet mesa]

Thanks @charred parcel . And GPT-3.5? It is limitless?

[charred parcel]

there is a GPT-3.5 limit but you will never reach it

[scarlet mesa]

Thank you.

Do you think that the limits for GPT4 and GPT4o will rise as well when GPT5 comes out?

[charred parcel]

I don't know, however I do not expect GPT-5 any time soon as we only just got GPT-4o

[scarlet mesa]

Ok, thanks for your help!

[crimson nexus]

I'm having problems with gpt for weeks, it lags and sometimes is not even loading the website itself. Is it just me? Now it doesn't even show my history and i cant begin a new chat.

[elder pebble]

Help! I loaded credits onto my account, I have my api key but I'm getting the following error: 15:35:10.399 ERROR: LLM API Error: Error code: 429 - {'error': {'message': 'You exceeded your current quota, please check your plan and billing details. For more information on this error, read the docs: https://platform.openai.com/docs/guides/error-codes/api-errors.', 'type': 'insufficient_quota', 'param': None, 'code': 'insufficient_quota'}}.

[trail cairn]

wait 5-30min after adding em

[elder pebble]

TY! I added them yesterday. I don't understand the new user vs project api's. I have a user api active. Do I need to set some kind of opening limit on it? I have never successfuly used OpenAI API yet.

[trail cairn]

if you have created a project and use a key made for it then the project budget needs to be set

[elder pebble]

I don't understand this at all. I don't see how or if I need to change somthing. 😦 Help? 🙂

Ill get a new key when we finish troubleshooting.

[trail cairn]

check project limits

however default should have something there automatically

[elder pebble]

I apreciate your help. I wish there was an actual help team. Still getting: 16:10:52.178 ERROR: LLM API Error: Error code: 429 - {'error': {'message': 'You exceeded your current quota, please check your plan and billing details. For more information on this error, read the docs: https://platform.openai.com/docs/guides/error-codes/api-errors.', 'type': 'insufficient_quota', 'param': None, 'code': 'insufficient_quota'}}

Is there anywhere I can get help with my account?

I found the account chat help. Will try there.

[elder pebble]

Ok! I finally got it working. Open AI is changing from personal API keys to Project API keys. I had to create a project, assign usage limits and models. Create a new sk-proj-xxx key and finally was able to access it. It's crazy no one at OpenAI could help me understand this! Thanks for all your help tho!

[pliant saddle]

I am writing to request the cancellation of my recent ChatGPT Plus subscription and a refund of the $24 subscription fee. I accidentally subscribed to the service, and I currently do not need the Plus features. is anyone for help ？

[charred parcel]

contact support at https://help.openai.com

[silent topaz]

When will all the full features of Chatgpt4o be released, it always says in the following weeks, but it has been more than a month, almost still not released.

[rugged wing]

Probably in upcoming months.

[late glade]

Can anyone tell me which is the true ChatGPT app for iOS? It seems like there are a few that claim to be OpenAI.

[bronze charm]

https://apps.apple.com/us/app/chatgpt/id6448311069

[late glade]

Thank you!

[placid gazelle]

im building an app where the front is using Next JS and the backend is using python. I'm using the assistant APIs, and all messages are initiated from the server. After I submit a message to OpenAI from the server, I want to stream back the response to the front end. Are there any examples that show how to do this? Im using FastAPI and using rest to communicate between front end and backend (though open to changing this if necessary)

[minor bolt]

🥲

Pls help

[quiet fox]

I have been getting this error while trying to buy GPT Plus:

We are unable to authenticate your payment method. Please choose a different payment method and try again.

I use QNB Finansbank (A Turkish bank)

[golden ruin]

maybe you need to open it for international transactions? or maybe your bank recognized it as suspicious activity and blocked it temporarily, in any case, it's better to call the bank

[mild heath]

In my Personal Account I cannot see the GPT Store, although I am Owner in the Team-Workspace.

[regal lynx]

You'll need to use the API, and you'll probably want GPT-3.5 Turbo. You can find the pricing for the various models here: https://openai.com/api/pricing/

[trail cairn]

huggingface has several models you can run offline, gpt-2 is terrible vs almost everything

from the small ones, llama variations, phi-2 or 3, etc

depends of your hardware

[fading robin]

phi-3 will run on almost anything hah it even have vision :o

[astral plinth]

Hello, as part of my experiment, I received a response in which GPT-4 tells me how to build a rocket. Do you need this information to patch this security hole? If yes, where should I publish it?

[trail cairn]

if the experiment was jailbreaking, etc. then no

[astral plinth]

First, I asked for help with the specs for an air-to-air missile in KSP, and then I asked him to write how to make a real-life copy of this rocket. I don't know if this is considered hacking or not, but I thought that this loophole could be easily exploited if someone had malicious intentions.

[trail cairn]

if you got something you did not expect then https://openai.com/form/model-behavior-feedback/

[astral plinth]

Thanks!

[frozen robin]

I did

when I played a story game with chatGPT

[hot dawn]

Hi, I woder that text-embedding-3-large model trained to can use shortnening embeddign with dimenion 1536 officially?
or where can i check the official dimension support list?

[hardy tide]

Hi all, i really want to became beta-tester of newest versions of OpenAI, guys how I can do this?

[regal lynx]

You can apply to be part of the redteaming network here: https://openai.com/form/red-teaming-network/

Outside of that, there isn't a specific way. I think OpenAI chooses random users every now and then to try something, but it's pure luck/chance

[limpid wadi]

Hi can we recover deleted chats?

[earnest lake]

Don;t free users have access to all capabilities now? I have a custom GPT that offers to create an image and that cpability is turned on. I just tested as a free user and it said it couldn't create an image, no matter how much I told it that it could.

[trail cairn]

creating images is currently paid only

[earnest lake]

Ah. Thanks. Do you know where to find a list of pad vs free features? Is this a slow rollout thing? Because I thought they made a big deal of free users getting everything at the last launch event?

[trail cairn]

https://openai.com/chatgpt/pricing/

[earnest lake]

Many thanks!

[earnest lake]

Can free users not leave reviews on custom GPTs?

[trail cairn]

nope due possible abuse

[earnest lake]

Thank you once again. Much appreciated.

[silent cipher]

Is there a reason why ChatGPT won't show images? I can see it generating a link but it doesn't show it in the response

It starts to generate the link but then disspears

nvm it previously said it oculd provide images now it says it can't lol. Told it to provide the link so I can at least click on it to view and that'll work

[earnest lake]

Only paid accounts can generate images. I ran into this earlier with a free account, exactly as you describe with a link being constructed and then disappearing.

[main falcon]

How would I connect my assistant on the assistant’s platform on the open ai website onto my discord bot because when I do the same prompt and model, it gives different results from what happens on the api platform

[lunar wave]

does anyone here have an idea on how to do intent recognition via openai/other models? thank you in advance

[gusty yoke]

Hello 👋

I need help with the "parallel requests".

Background

• I am creating an application which sends notifications to the user 3 times a day.
• I generate notification with GPT-4.
• If there are 1000 users, I need to send 1000 notifications together. (I am aware of my ratelimit based on my tier, but suppose I am under my ratelimit)

Problem

Then, currently I need to sent the request one by one (although I can use some async library) but I need some concrete solution from OpenAI library.

Previously, I was able to sent the requests in batch using the completion endpoint, but currently since all endpoints are ChatCompletion, I don't find a way to do that.

Can anyone please help me to sent all requests together (all have different prompts) and get completions together!?

Thank you 🙏

[leaden juniper]

Hey! Hope all are going well. Does someone has an example of the tool going into the prompt? I cant find a example for the functions attached in the messages array to see what shape they have

[tribal dust]

Hello. I am thinking of creating an 'Easy-Read' dictionary by using Dall-E. Unfortunately creating Easy Read documents can be quite pricey, which is counter intuitive to inclusivity.

I wondered if there is a way that Open AI would be able to provide the credits required to complete this task. Alternatively I would like to work out the total cost to complete the project so that I could fundraise to get it done.

I did a few tests on Copilot and with the right prompting, the AI seems to be able to generate suitable easy read images.

Hope this all makes sense. Please do tag me if you can assist with this request.

[static needle]

Good, I need to create a chatbot to get customers, and this must be trained, the thing is that I doubt about what would be the best option to train this model, considering that I only develop the intermediate API between the user interface and the OpenAI API, and are my customers who must train the model with their data and parameters.

[upbeat crane]

I'm genuinely confused why the brand new chatgpt app doesn't support my mac lmao

[charred parcel]

do you have an Apple Silicon Mac?

[upbeat crane]

nope

[charred parcel]

then that's why

[upbeat crane]

... only a few more years until I can convince my family member to hand me down one lmao

Interesting choice for it only to work on apple sillicon though.

[charred parcel]

not really, everything is shifting to Apple Silicon now

[upbeat crane]

it's all marketing crap, I get why they did it but the execution was poor, i.e making M1 the global chip for basically all devices, I won't digress that much lmao, it's better in some ways but I've occasionally used apple sillicon and it feels like the software holds it back more than the processor lmao

[static needle]

Any help/guidance?

[bright shuttle]

my gpt4 even if I pay has some messages gap? after that i have to use 3.5 model? anyone from admins to respond? I'm sorry if this question was asked before

[timber shadow]

Yes, there's still a usage cap! The default for 4o on Plus is 80/3hrs. You can read more here: https://help.openai.com/en/articles/7102672-how-can-i-access-gpt-4-gpt-4-turbo-and-gpt-4o

[bright shuttle]

thank you.

[smoky reef]

This has just been happening to me lately too, ever find out why?

[kindred marlin]

when this will be solved

[errant silo]

It seems there is an issue with saving and providing the document directly. I recommend copying the following content into your preferred word processing software and formatting it as desired. You can insert the photo as well in your document.

anyone can help

[rigid musk]

From which country ?

[kindred marlin]

Ukraine

[kindred marlin]

Ukraine

[rigid musk]

Can you try to reinstall? It is on cellphone?

[twilit garnet]

anyone can help me? It is a disaster that I could not use gpt for days: when I hit enter key, the button turns gray and nothing happens.

[fading robin]

what you can to do

• clear cookies in browser
• enter in private mode in browser
• try diff broswer (like firefox)
  idk waht more

[twilit garnet]

I have tried all of them

It doesn't work

It happens on my windows PC, iphone, MAC, just all of my devices

[regal lynx]

Are they all using a VPN?

[twilit garnet]

actually my PC and Mac is, but iphone is not

it also occurs on my phone even not using vpn

[regal lynx]

I know VPNs can mess with Cloudflare, and that can cause issues. Even then, I'm not sure why it wouldn't work on your phone

[twilit garnet]

I think there maybe something wrong with my account? but I failed to contact with openAI staff. if they can check the status it may help a lot......

[twilit garnet]

anyway thank you so much

it is really ridiculous that it works when I changed to a free account.....the problem only happens on my plus account

then what's the reason to pay for it XD

[quick bone]

i was trying to make a function taht would take in a query and return the responce from openai in typescript, this is what i wrote, its not working : (, can someone help me plz

async function fetchOpenAIResults(query: string): Promise<string> {
    const apiKey = 
    const endpoint = 'https://api.openai.com/v1/engines/davinci-codex/completions';

    try {
        const response = await axios.post(endpoint, {
            prompt: query,
            max_tokens: 150,
        }, {
            headers: {
                'Authorization': `Bearer ${apiKey}`,
                'Content-Type': 'application/json',
            },
        });

        if (response.status === 200) {
            return response.data.choices[0].text.trim(); // Assuming you want the first completion text
        } else {
            throw new Error(`Request failed with status: ${response.status}`);
        }
    } catch (error) {
        console.error('Error fetching OpenAI results:', error);
        throw error; // Handle or rethrow the error as needed
    }
}


const query = 'Translate "Hello, how are you?" into Spanish.';
try {
    const result = await fetchOpenAIResults(query);
    console.log('OpenAI Response:', result);
} catch (error) {
    console.error('Failed to fetch OpenAI results:', error);
}  
        
const expandedQueries: string[] = [];

also i can verify my apikey is right

[topaz frost]

import axios from 'axios';

async function fetchOpenAIResults(query: string): Promise<string> {
const apiKey = 'your-api-key-here'; // Replace with your actual API key
const endpoint = 'https://api.openai.com/v1/chat/completions';

try {
    const response = await axios.post(endpoint, {
        model: "gpt-4", // Changed to GPT-4
        messages: [{ role: "user", content: query }],
        max_tokens: 150,
    }, {
        headers: {
            'Authorization': `Bearer ${apiKey}`,
            'Content-Type': 'application/json',
        },
    });
    
    if (response.status === 200) {
        return response.data.choices[0].message.content.trim();
    } else {
        throw new Error(`Request failed with status: ${response.status}`);
    }
} catch (error) {
    console.error('Error fetching OpenAI results:', error);
    throw error;
}

}

// Example usage
async function main() {
const query = 'Translate "Hello, how are you?" into Spanish.';
try {
const result = await fetchOpenAIResults(query);
console.log('OpenAI Response:', result);
} catch (error) {
console.error('Failed to fetch OpenAI results:', error);
}
}

main();

[kindred marlin]

on mac on cellphone it works

on mac don't

[wintry turtle]

How do you change your knowledge cut on ChatGPT

[charred parcel]

the knowledge cut off is fixed, you cannot change it

[minor bolt]

@golden ruin

[golden ruin]

Yup?

[minor bolt]

Why it's giving raw latex code

Like \frac

Instead of fractions

[obtuse gust]

anyone knows why I can't send message?

[celest meteor]

which model are you using?

[obtuse gust]

gpt4o I think

ye it is gpt4o

[celest meteor]

it worked on my end; Can you try it in another browser?

[obtuse gust]

oh ok

tried it in firefox and chrome, same result ,gonna move on to edge

[celest meteor]

can you try in your phone

[obtuse gust]

im in free version btw, now I just reached the limit lol, I think it because I was trying to regenerate it so many times. But it isn't supposed to reach limit (if not counting error message). Now I can't even use 3.5, it has the same error message

even in phone

wait nvm, in phone the 3.5 works

ok I think the message may be too long fpr 3.5 , I will try again tmr with 4o

[celest meteor]

oh.. that's strange. normally the error message will clearly tell us when we hit the limit

[obtuse gust]

ye the phone tells me I hit the limit and it will come back by 2 am, but web just says something is wrong lol.

but now the web also tells me it hits the limit, so I can guess that I hit the limit after trying to test and regenerate the error message...

[celest meteor]

Any idea why GPT4o is cheaper than GPT 4 Turbo? I found 4o is way smarter in my use case. But why the smarter one is cheaper? out of curiosity😆

[charred parcel]

it's a lighter model and cheaper to run. OpenAI are just passing their savings on to customers.

[inner raptor]

must javascript in ChatGPT be active to let a file be downloaded?
I have problems in download a corrected File there. The link is generated but when I want to copy it in a new browser window I just see void();

[kindred marlin]

Do you know how fast can open ai people solve problem with login into macOS app it is already month now
P.S. I know that support slow

[charred parcel]

sorry, I don't understand, what do you need?

[kindred marlin]

I can’t login to macOS app . I contacted support several times to solve it . But it isn’t solve . There is also community OpenAI thread with the same problem and we all are waiting. Maybe you can check when this will be solved . The problem is Something went wrong. Please make sure your device's date and time are set properly. Check that your internet connection is stable, then restart the app and try again.

I submitted also posts for support in help ai and discord bug fixes

[charred parcel]

what version of macOS are you running?

[kindred marlin]

Steps to reproduce:

1.Open ChatGPT.
2.Click on the CTA within ChatGPT to download the macOS app.
3.Download and install the app from the provided link.
4.Attempt to log in.

Later tries:

1.Reinstalled the app.
2.Cleared cookies.
3.Turned off safety settings.
4.Used Safari and Chrome.
5.Updated macOS.
6.Restarted the Mac.
7.Turned off automatic time change and set an incorrect time.
.8Turned back on automatic time change.
9.Changed the time server.
10.Contacted customer support on Discord and the OpenAI website.
11.Turned off private relay and turned it back on.
12.Turned off Wi-Fi privacy and turned it back on.

Expected result:

The user should be able to log in to the macOS app without issues.

Actual result:

The macOS app displays a message indicating: "Something went wrong. Please make sure your device's date and time are set properly. Check that your internet connection is stable, then restart the app and try again. (code:7176974E-7B0E-46D8-A357-048C05FE3485, API.DeviceCheckError, 2)"

Additional information

Browser: Chrome, Safari
OS: macOS 14.5, macOS 15.0 beta 1, 2
Device: MacBook Air M1, 8GB, 256GB
ChatGPT Versions: 1.2024.152 - 1.2024.170

Please note that the problem has persisted for 5 weeks and there are similar unresolved issues reported in the OpenAI community.

[nimble agate]

I guess this is a better place to ask: starting last night I could no longer edit queries I sent on my iPhone or iPad. At the same time, I noticed that I can no longer upvote, downvote, or vote on pictures.

On the desktop? No problem. But on a mobile device, those features no longer work.

I saw someone mentioning contacting OpenAI support, but after spending a few minutes clicking around in the help widget in the app, can't figure out where the screen to actually type out my issue is?

[lilac jasper]

Good day. When I upload a file to the chat, it gives the following error "Exceeded user quota". What should I do?

[manic wedge]

Hello everyone, I am running into an issue with a custom GPT I made and was hoping someone could help here. I am a medical student and I am trying to create a custom GPT that can make practice board questions to help my fellow students and I study. I wrote an extensive guide to writing board style questions in a word document and uploaded it to the knowledge area into the edit gpt section, and then asked it to create questions based off another word document which is compiled of information from every lecture throughout a semester. It doesn't necessarily "fail" but the questions don't seem to be of high difficultly and it improperly cites its sources. Can anyone shed some insight?

[surreal bolt]

why I'm not able to install chatGPT In my mac Air M2, OS Ventura 13.6.1 /

[limpid wadi]

hello, can we recover deleted chats ?

[charred parcel]

because you need macOS 14

[rigid musk]

Unfortunately I don't think so, you can try and raise a ticket on help.openai.com

[blissful igloo]

I can't edit messages on the website (I use the website on my iPad)

[real thunder]

Are all custom GPTs forced to use GPT4o?!?!

It's literally unusable for most of my use cases and is crap compared to 4

Anyway to change this?

[minor carbon]

I came here to report a bug. ChatGPT seems to have logged out on its own, so I went to log in to my account again, but I always kept coming back to the home screen (to log in), I tried to do the procedure through the loader, but the message appeared that says to restart the application or check the internet, The error persists even if I do what I'm told. What do I do?

P.S.: I am mobile user

[stray bloom]

I have a similar point to make. Over the last two weeks regardless of the model that I select, whether it be the typical GPT-4 option or the GPT4 Omni model, it seems to both be running with Gpt-4-turbo-0613. Is anyone else encountering this error?

[real thunder]

It's incredible how we're forced to use the 4o when it's obviously just the "mixed (3.5 + 4)" version that was a failed launch. It's ruined several of my workflows across a dozen GPTs I built

[urban brook]

is openai billing failed requests to embedding models?

[charred parcel]

all API interactions are billed

[urban brook]

thx

[lilac kestrel]

any one has problems about lost all my GPTs?
I lost my all GPTs that I create and don't know how to do anyone help?

[toxic sky]

Contact support at help.openai.com

[ionic torrent]

I no longer see dall-e3 in chatgpt ???

[ocean crypt]

Hello OpenAI Team Developers, Nice to meet you.
Can you help me create a storyboard for a movie?
I am currently using OpenAI's DALL-E model to generate them, but they are completely inconsistent.
Is it possible to generate consistent storyboards programmatically using API?

[trail cairn]

nope, the prompts are rewritten, there is no seed parameter nor reference image support

[ocean crypt]

is it impossible to implement it?

[trail cairn]

yep not possible with the current features that openai offers

[ocean crypt]

are you developer of OpenAI?

[trail cairn]

no

[ocean crypt]

is there any OpenAI team member in this channel?

[toxic sky]

Why do you need one.

[ocean crypt]

Hi @toxic sky
Nice to meet you
I need your help to generate storyboard for film using open ai api programmatically.

Could you let me know is it possible to use seed in open ai api?

[toxic sky]

Seed is not what you want for your purpose. You should run it with chunks and give all the previous chunks too it as context.

[ocean crypt]

and how to get gen_id?

[toxic sky]

You can play around in the playground for prompt testing.

Ah

For images, GEN_ID only works inside of ChatGPT per conversation.

You can not use it through the API.

[topaz frost]

Hello, how are the alpha testers for voice mode chosen, what is the criteria?

[toxic sky]

There is no public criteria.

[ocean crypt]

ok, I see

[toxic sky]

DALL-E 2 has seeds.

[topaz frost]

Ok, are you aware of it?

[toxic sky]

There is no public criteria.

[topaz frost]

So there is no criteria at all?

[toxic sky]

None that has been released to the public.

[topaz frost]

Ok

[trail cairn]

someone asked it from support and "The GPT-4o real-time voice and vision capabilities are indeed rolling out in a limited Alpha for ChatGPT Plus users. However, access to the alpha version is currently limited and invitations are extended to a subset of users based on various factors, including usage patterns and the need for diverse feedback. While we don't have a public application process for the alpha at this moment, we're continuously evaluating user engagement and may extend invitations to more users over time. ... The selection process is based on various factors, and we aim to include a diverse group of users. The best way to increase your chances of being selected for future testing phases is to continue actively using the ChatGPT iOS app and engaging with the voice features."

[topaz frost]

Understood, i shall begin

[trail cairn]

but that was long time ago so its possible they already chose initial alpha testers but had to delay the rollout

[toxic sky]

+1, they are probably not going to roll it out for more testers after the initial tester rollout, which already happened.

[topaz frost]

😦

[ocean crypt]

Hi @toxic sky

please check attached screenshot

is it possible to use seed?

how to use seed to generate image?

[toxic sky]

https://platform.openai.com/docs/api-reference/images

It may have been deprecated with DALL-E 3

[ocean crypt]

I am using dall-e-2 now

[toxic sky]

Yes, but when DALL-E 3 was released the seed parameter may have been removed.

[ocean crypt]

you mean it's impossible to use seed in openai api?

[toxic sky]

It seems so.

[trail cairn]

yep

[toxic sky]

I haven't used that parameter in a while.

[ocean crypt]

@toxic sky
are you member of OpenAI team?

[trail cairn]

current api image features seem to be safety oriented eg. if are lucky to get a famous looking person you cant use it as a reference to create alt images

[toxic sky]

I moderate the forum, I do not develop or work on ChatGPT or DALL-E.

[ocean crypt]

Ok, I see, thank you

[waxen burrow]

hi need help. Does anyone here please solve account zeroing, theft through open ai, three days ago my account was zeroed to 0 where I had 30 dollars for my apartment and Tier 3. my entire history was deleted of course I have screenshots from the history I have screenshots from using open ai GPT 4.o there is no problem to send. I write to e-mails and no one answers, yesterday I had money sent to my home account and today when I looked So it opened it and reset it to zero, about 74 dollars, can someone here please tell me how to solve this? is there someone who deals with the technical background of open ai, of course I will dismantle the 74 dollars with my bank that it was an unauthorized payment, but I would like to get my open ai account with tier 3 back, and the additional 30 dollars. I have been using open aigpt since its inception and I have installed at least $700 in them, so I would also like some feedback and help in solving this problem, thank you, my email is [e-mail]

[trail cairn]

there is no email, you need to use the support site chat bubble https://help.openai.com

[waxen burrow]

their chat bubble is totally useless .. how do i explain this problem to him? that he reset my GPT account, history and took other money from my account?

[trail cairn]

its the only way to contact openai. if you cant see payment history and your tier is back to lowest it sounds you logged with wrong account or the account got compromised and someone took it over removing you access to see anything. if you dispute the payment without support knowing it, your card will be banned

[waxen burrow]

what I read was that Google leaked over 500m passwords, so I'm wondering if someone hacked into my account and cleaned my account and then selected my personal account through GPT, so I changed my password, but what now?

[waxen burrow]

So can you please advise me how to proceed?

[trail cairn]

the chat bubble is the only way. you need to be logged in and click different options until it lets you write a custom message. it can take days to get an answer

[waxen burrow]

so what should i do until then? create a new account? and have your payment card blocked? I was lucky that I sent 98% of the money to a savings account, otherwise I would have been at 0. are you saying here that if I come and want to block the payment of the 74 dollars, the support will block my ability to buy the payment even though I was robbed?

[trail cairn]

if you block the payment and they dont know about it they may think it was a fraud or stolen card. did you have auto refill enabled or how and where was the money taken? api is prepaid credits only

[waxen burrow]

I don't understand it myself. on Monday I go to the bank and ask how it is possible that I did not receive a confirmation SMS or a confirmation and the money was withdrawn automatically. I wouldn't mind charging $74 for GPT, but I don't want to lose that money.

[trail cairn]

what balance does it show here https://platform.openai.com/settings/organization/billing/overview

[trail cairn]

are you sure you are logged in with correct account?

like no google login vs manual login/pass

[waxen burrow]

I had 30 dollars there, I also have screenshots of how I used GPT4.omni, even the history is deleted and I found a screenshot when I started using GPT which was sometime on 2/12/2023

[waxen burrow]

yes it's the account I use on both my gaming PC and my work tablet and it's the same on both.

[trail cairn]

do you see only you here and as owner https://platform.openai.com/settings/organization/team

[waxen burrow]

I'm already in the chat option to send a message as a text, should I write him my entire problem?

[trail cairn]

yes

[trail cairn]

also give them the info from the email which says payment was made, organization id is at the bottom

you can also check does the id in email match the id here https://platform.openai.com/settings/organization/general

[waxen burrow]

should I add them to the ID for that question? I will also send them a statement from the card, where it is easy to find that I have already deposited 500-600 dollars into the account.

[waxen burrow]

I'll probably send them the picture through the Chatbot, right?

[trail cairn]

they need to org id at least

[trail cairn]

i dont think it accepts images

[waxen burrow]

well, I wrote him the whole story, that 3 days ago my account was deleted with history, tier 3, and money (about 30 dollars), plus today 10, 50 and 14 dollars were taken from my cards, that I don't mind if he credits it to my account , I sent my ID, knowing that I can easily document the statement with the cards. maybe that will be enough right?

[azure plover]

@trail cairn is your app on phone normal? Mine is missing the dall-e icon and also I it’s missing the icon to attach files, images

[trail cairn]

they will ask for more info if they need but with org id they should see its full history and everything

[trail cairn]

its gone from everywhere. they disabled it for some reason

[azure plover]

also the icons to attach files?

[waxen burrow]

thanks a lot! honestly, if you ever go to the Czech Republic, you can have a beer with me and a weekend at our family farm! 😄

[trail cairn]

yep

[trail cairn]

sure 🙂

[waxen burrow]

I still ask, will they contact me by email or through the GPT profile?

[trail cairn]

you get reply to the bubble and there should also be an email

[waxen burrow]

I work in a casino as a manager, so if you ever want advice on roulette or fitness, I'll be happy to give you advice. I believe that kindness should be returned with kindness, and you helped me a lot today.

[waxen burrow]

bot wrote me that: Thanks for reporting this to us. Our team will respond as soon as possible.

[low adder]

is chatgpt 4o not letting you upload images rn? I can do it on 4 but not 4o

[waxen burrow]

but I look that 25.06. In 2024, a project was created there and that's the day when everything was deleted (hidden) if it's true that someone hacked me, and I change my password, then he can no longer access my account? or did you connect your own acc there? I'm sorry I keep bothering you.

[trail cairn]

default project is auto created same time as your account is created

[azure plover]

seems like its a bug, it back for me a few minutes ago. Try closing your app or refreshing the website again

[waxen burrow]

in that case, I don't understand how it is created on 25.06. 2024 if my account is old then 12.2. 2023, but 25.06. 2024 my account was reset, could it really be an error during the update? because the turbo omni also arrived on 25.06. No?

[trail cairn]

i cant think of anything else but when you logged in on 25th it created another account and the other one is still somewhere. its not possible for a user delete payment history, reset tier, etc.

in the past ive seen people do it with google accounts that they login in and the old one is not used when some google authorizations were manually removed

[waxen burrow]

look, here I have a screen that I sent to a friend on Discord years ago about how expensive OpenAI is. Those were the beginnings when generation was really expensive, then fortunately I had a couple of windows open on my tablet, where I can clearly see that I have a balance of $29.72 and at the same time that I generated via GPT4.Omni, I wouldn't have been able to do that if I didn't have tier 3, and I still remember looking here and thinking that it's great that omni has 40,000,000 tokens. which is on tier 3

[trail cairn]

yep that account likely still exists somewhere

openai can find it via the org id

[waxen burrow]

I really wonder what happened. but surely someone else will be involved when the money was taken from my account, the biggest irony is that I came here after saying that I will buy today for 30 dollars before it is resolved. so I go to see how much I have on my card if I buy for 30 + 6 tax or for 40, and I see that I was charged 10, 50 and 14 .. I say you're kidding. 😄

[trail cairn]

your screenshot also shows the auto recharge was turned off

[waxen burrow]

yes, I think someone had to do it manually, or some really special bug. in any case, I will ask the bank on Monday how it is possible that I am not getting a confirmation sms .. why do I have double protection set up when it is completely useless.

[trail cairn]

there is no confirmation if you add openai as authorized card user, just like with netflix, steam, spotify etc.

[waxen burrow]

hmmm look

[waxen burrow]

which is very stupid. 😄 so according to me. it is probably better to pay for these platforms through a second account. paypal etc

[waxen burrow]

strange that I have API Keys left

[gilded slate]

    conversation_history.append({"role": "user",
                                 "content": user_message,
                                    "attachments":
        { "file_id": file_id, "tools": [{"type": "file_search"}] }})```

the tools : "file_search" is not working properly as it is not reading the file to give me the response

[ocean crypt]

@toxic sky
How are you today?

https://community.openai.com/t/consistent-image-generation-for-story-using-dalle/612276

please check this link

is it possible to generate consistent images using API?

[stray topaz]

Hi I’ve been on off on chat gpt plus. Yesterday it ended after 1 month
When I tried to renew this time, it said card invalid after bank otp. This happened like 5 times yesterday. Today I cleared cookies Then I tried using another card. Still says card invalid. It’s very important for me please help . I’m trying to renew

[frail hazel]

When did chatgpt becaome so strict with free users?? Now im locked out of gpt4o until 9am tomorrow

[terse meadow]

Hi, I want to make a conversational style interviewer using the Assistants as I’ve done some customisations to the model. As of now, I’ve implemented a start and stop button to send the reply and get the reponse back from the assistant, as the data is travelling in chunks, as mentioned in the documentation. I want to implement the same in a conversation style. Please guide me on how can I achieve this.
Model: gpt-4o

[blazing schooner]

We muted the bot on our Discord server, then unmute, now it isn't responding to commands and mentions, not even / commands. How I revive it?

[toxic sky]

Which bot?

[olive patio]

how do I fix my card being declined

I have money

I can buy anytihgng else wiyth my card but buy openai credit

[timber shadow]

Have you checked these two help articles already? https://help.openai.com/en/articles/7232916-why-was-my-credit-card-declined
https://help.openai.com/en/articles/7947663-chatgpt-supported-countries

[olive patio]

yeah

[carmine vector]

I'm trying to have ChatGPT make an English analyzation of a Danish song. I asked it to be as detailed as possible. It did an ok job, and I received a usage policy warning?
Since the analyzation wasn't perfect, I asked ChatGPT to try again, but include explainations of puns and references. It was really good tbh! I was really happy with what I got to read, but then I got a red usage policty??? And the thing is gone now :( I just want to analyze a song :(

[timber shadow]

Does the song's subject have anything to do with any of these categories? https://platform.openai.com/docs/guides/moderation/overview These are the things that typically can trigger a content warning flag.

[carmine vector]

Hm, I don't think it does, That's why I'm here :P

The most 'sexual' thing in the lyrics is "Danish women say "ah" when we arrive in town."

[rustic marlin]

i don't think tge issue is nsfw, might be because of coppyright

[carmine vector]

Hm, but wouldn't that go under fair usage, as I'm having it analyzed and explained?

[rustic marlin]

probably yes, the AI does not really know what do or do not have coppyright

[carmine vector]

As in it just triggers it anyways, or as in "that's probably not it"?

Also, it was a red warning, so it deleted the text, which is annoying because it was spot on, and perfect for what I needed, and I didn't get to save it :/

Is there a way I can get that back? It was a super solid response

[rustic marlin]

as in it could be just a hallucinated false positive

what do you get when you send the message?

[rustic marlin]

oh.. this is harsh

this is a rare case

[carmine vector]

I'll do anything for that text back

[rustic marlin]

very likely not

[carmine vector]

Fack

Okay! Do you reckon I'll be in trouble if I ask for the response again?

Orange warning I got first (with a slightly worse outcome, so I added more prompt, with better results)
And the good one here gave me red warning. What comes after red?

[trail cairn]

getting multiple warnings in a short time is bad and you should change what you are doing as the automated system can flag your account and send a warning email

[carmine vector]

But I need it 🥲

Oh well

Is there anywhere I can actually report it? I'd like to be able to analyze songs later on :D

[trail cairn]

you cant report it as lyrics are copyrighted material and they have specifically included them to the bad topics lists

they did not originally have it but it was added when ai lyrics variations were trending

[shut raft]

i need à excel i have send to chat gpt but i have lost it on my computer i need dowload it is possible ?

[trail cairn]

at one point it was possible by enabling the code interpreter and asking it to give it

[shut raft]

and now ?

[scarlet jungle]

sorry for this silly question, but when will gpt 4-o video calling be available ?

[trail cairn]

#announcements

[scarlet jungle]

thank you! i figured it was somewhere but there’s like 200 channels

[trail cairn]

you need to try it but it might be blocked now and then its not possible to get the full files

[carmine vector]

Hm :/ Do you reckon AI will understand what is and what is not fair right usage in the future?

[trail cairn]

i think the issue was that working with lyrics is ok when you are sourcing them from a place which has a license for them, eg. spotify or album cover and you have access. openai has not licensed them and its chatgpt that is generating the replies

[carmine vector]

Sorry I don't understand this part: "... when you are sourcing them own access and ..."

Could you please explain that again?

[trail cairn]

fixed

[carmine vector]

But if I wrote the lyrics myself?

[trail cairn]

it wont trigger on those, ive done probably +100 song lyrics without issues

[carmine vector]

But.. Then why did it trigger? It's just lyrics?

[trail cairn]

if you do derivate work it will trigger

if it was similar or adaption of other song it might trigger

[carmine vector]

Oh, I think you misunderstood me, not my own lyrics, but like, I just wrote down what I heard, I guess I sourced it from my ears??

I get that it just how it is btw, I'm just trying to figure out why

[trail cairn]

you heard it from where?

[carmine vector]

I own the song both on CD and.. well I guess I don't own it on spotify, but u get the jist

But it's not like the AI can tell if I got the lyrics for "the wheels on the bus" from a purchased CD, spotify, or even directly by the creator with 100% creative rights. It's the same lyrics after all

[trail cairn]

yeah so its a commercial song, you can mess with the lyrics on paper and its fair use. its not fair use in chatgpt regardless if you own it or not as licensee must be openai

[carmine vector]

... but it was an analyzation? Even just a translation would be considered fair use

[trail cairn]

its not. same thing as with books or poems, originals and translations are copyright protected

openai as a company has no rights to the lyrics

[carmine vector]

"Fair use permits a party to use a copyrighted work without the copyright owner’s permission for purposes such as criticism, comment, news reporting, teaching, scholarship, or research. "

It's fair use

[trail cairn]

yes if you are doing it as a person

[carmine vector]

Ah

Right, that's fair

[trail cairn]

openai is a business, chatgpt is a product

[carmine vector]

And an excelent point

Well, darn it

[carmine vector]

OOp, forgot to say! Thank you so much for your help @trail cairn ! I appreciate it :D

[timber shadow]

Hey! GPT-4o has a knowledge cutoff of October 2023 -- see this documentation: https://platform.openai.com/docs/models/gpt-4o On ChatGPT, I think OpenAI instructs the model on its knowledge cutoff in a system prompt. It looks like you're chatting with the model in a non-ChatGPT environment, so it might be hallucinating a random knowledge cutoff date if Oct 2023 isn't specified as such in a system prompt/model instruction.

A model being trained on data up to a certain date does not inherently give said model the ability to accurately report on its own knowledge cutoff.

That's why it's made explicit on ChatGPT via a system prompt--and even then, it can still sometimes hallucinate 😄

It likely has to do with how LLMs work. LLMs don't encode knowledge in terms of truths and falsehoods. Instead, they basically read a whole bunch of text, and their goal as an AI system is to learn how to predict what text will come next while they're reading. Complex LLMs like GPT-4o are very good at this, so they can often report true knowledge because they've gotten so good at predicting what words should come next. But, it's not perfect, so you sometimes get misfires! Predicting involves probability, instead of "absolute yes" or "absolute no."

[rustic zenith]

Is there any vacant position for moderator or community manager I want to apply

[charred parcel]

are you active and helpful on the server?

[crimson stump]

Hi everyone, Im so excited to be here. I want to is if is possible to get access to Sora, One of my hobbies is writting and composing music, so I would like to know what I need to do to try Sora. Best Regards

[celest sparrow]

I thought the beta to Sora wasn't even open yet... like actually not even accepting applications yet

[crimson stump]

I understand, thanks for your response

[rustic zenith]

is this the only crieteria?

[charred parcel]

no, but it helps

[rustic zenith]

oh i think they also recuit people on their experience

[charred parcel]

server staff all share some qualities, such as longevity and participation

[rustic zenith]

hmm thats good but if there is a possibility that u will be hired then the guy will try to be active and also depend on if there is job opening or not

[charred parcel]

all server staff and helpers are volunteers

[rustic zenith]

like they dont get any pay?

[charred parcel]

correct

[rustic zenith]

[charred parcel]

people with gold usernames are OpenAI employees, everyone else is a volunteer

[rustic zenith]

whats that spotlight role?

[charred parcel]

it's for people who have been featured in #spotlight

[rustic zenith]

ok thanks for the info

[potent turtle]

Help, chatgpt app not loading

Stuck on this

[charred parcel]

try removing the App and re-installing it

[kindred marlin]

why you didn't answer me about macos app?

[charred parcel]

what was your question? I must have missed it

[kindred marlin]

this is my problem : Steps to reproduce:

1.Open ChatGPT.
2.Click on the CTA within ChatGPT to download the macOS app.
3.Download and install the app from the provided link.
4.Attempt to log in.

Later tries:

1.Reinstalled the app.
2.Cleared cookies.
3.Turned off safety settings.
4.Used Safari and Chrome.
5.Updated macOS.
6.Restarted the Mac.
7.Turned off automatic time change and set an incorrect time.
.8Turned back on automatic time change.
9.Changed the time server.
10.Contacted customer support on Discord and the OpenAI website.
11.Turned off private relay and turned it back on.
12.Turned off Wi-Fi privacy and turned it back on.

Expected result:

The user should be able to log in to the macOS app without issues.

Actual result:

The macOS app displays a message indicating: "Something went wrong. Please make sure your device's date and time are set properly. Check that your internet connection is stable, then restart the app and try again. (code:7176974E-7B0E-46D8-A357-048C05FE3485, API.DeviceCheckError, 2)"

Additional information

Browser: Chrome, Safari
OS: macOS 14.5, macOS 15.0 beta 1, 2
Device: MacBook Air M1, 8GB, 256GB
ChatGPT Versions: 1.2024.152 - 1.2024.170

Please note that the problem has persisted for 5 weeks and there are similar unresolved issues reported in the OpenAI community.

[kindred marlin]

I sent this to bug report and customer support 1 month ago but they didn't solve it

[charred parcel]

what version of macOS are you running now?

[kindred marlin]

macos 15 beta 2

[charred parcel]

and what precisely happens when you launch the app?

[kindred marlin]

It opens and I try to login

when I try to login I recieve this answer Something went wrong. Please make sure your device's date and time are set properly. Check that your internet connection is stable, then restart the app and try again. (code:7176974E-7B0E-46D8-A357-048C05FE3485, API.DeviceCheckError, 2)"

If you read my previous messeges I tried everything

[charred parcel]

I am going back and reading now, give me a moment

[kindred marlin]

take your time

[charred parcel]

ok, I have read your previous messages. I don't know what is causing the problem. I think this is only something support can fix.

[kindred marlin]

So I contacted them and they did nothing

As I understand you are admin

[charred parcel]

I am not admin

[kindred marlin]

Sorry 😉 I just want to solve this problem . It is frustrating that I can’t use app but I have followed all the requirements

And nobody helps me

[charred parcel]

I would submit another support request

[kindred marlin]

Thank you

[potent turtle]

I did but still the same problem is happening

[charred parcel]

I don't know what you can do, maybe contact support at https://help.openai.com

[robust canopy]

Hi guys, I've been using chatgpt and other llms since like gpt3 almost every day, I did close a decade of JS, I want to try and build some basic agents with RAG, can anyone suggest a good stack and/or course for someone like me? Theres so much tutorials on youtube, Im scared of chosing bad one 😄

[kindred marlin]

you can try clear cache of app

[velvet snow]

Hello everyone, quick question. Is there any type of “researcher” version of chatgpt or the chatgpt api available? Im currently working on my thesis and am working with different ai models / companies to determine what would best for me, and wanted to see if OpenAI offered any resources to academics.

[kindred marlin]

You can use gpts like scholar ai

[velvet snow]

Ah, yes I’ve seen that. I firstly would like to say thank you for your response, and should apologize, I should be more clear. I didn’t mean to have the ai model do research, I wanted to do research on the ai model, and was wondering if they had a version for academics, for example like geminis “researcher” model which has a lot more available requests, information and feedback, customization, and some features that are not yet public. Basically a model built for researchers that allows for a lot more flexibility then general free / paid accounts

[dusk roost]

i want cancel ChatGPT Plus Subscription, i don't way it pay, but there is no cancel or my plan button, pleas cancel ChatGPT Plus Subscription ([e-mail])

[trail cairn]

https://help.openai.com/en/articles/7232927-how-do-i-cancel-my-chatgpt-plus-subscription

[lean yacht]

Help, I deleted a chat by accident in ChatGPT, how do I recover it?

[celest sparrow]

pretty sure by law OpenAI has to keep a copy of it for like a week or two, the only way you possibly could is requesting assistance from their support, but even then, it seems a bit... ridiculous to bother them over such a thing better off just starting from scratch

[lean yacht]

I had important conversations there...

[celest sparrow]

Well, best to talk to support then and make a request and hope for the best

[lean yacht]

Ok...

Now what

I can't believe it

That's unfair

I didn't want to delete it

[lean yacht]

Someone help me please, I don't want to talk with a bot in support page

It doesn't help me

[timber shadow]

Hello, unfortunately deleting a chat is permanent, like you found on the help page. There's no way to undo it. Only thing I'd recommend is to check to see if perhaps you just archived the chat maybe, which can be undone.

[lean yacht]

That's unfair!

It didn't even ask me, I clicked it accidently

There should be a deleting confimation or a recycle bin for chats

[timber shadow]

A delete confirmation box pops up for me on mobile web, desktop web, and in the Android app after clicking the delete button, so I need to confirm the deletion twice.

[lean yacht]

Well, anyways there should be a "recycle bin" or something like that

[quiet hemlock]

gosh, why does sometimes voice chat translates my speech to my country's language even when I talk in english?

this leads to ChatGPT also responding in my language, with broken text.. please, someone, help me turn this off...

[formal mauve]

how to tell dall-e 3 what NOT to add to the picture?

[velvet snow]

I noticed when speaking full English and my default language is English it sometimes reads it in Russian, I don’t even speak Russian! It’s kinda funky atm

[lean yacht]

I deleted it instead another by error

Now i feel so silly

All that is done should be undone just in case

[thin dagger]

20$ a month btw - how do you fix this?

[charred parcel]

what browser and OS are you using?

[thin dagger]

firefox, windows 10

[charred parcel]

I would get in touch with support at https://help.openai.com

[hard jolt]

ANyone having issues uploading files, asking it to read the file, and being told that it cannot access files?

I've tried 3-4 times.. it's just being weird. (I do have plus)

[onyx kraken]

I just uploaded a simple .txt and it told me the contents. Maybe reach out to https://help.openai.com/ and use the bot bottom right, select messages then click through the choices until you can report a problem.

[hard jolt]

Thanks for giving it a test. For tonight I'll just swap over to Gemini (🤮 ) and tomorrow I'll try again. If it still acts up tomorrow on a fresh browser/computer/etc. I'll reach out.

[river karma]

I believe it was stated in the Open AI support that Free tier users can acces GPTs but I am still unable to. Is there something that I am missing here?

[river karma]

Tell it to use ChatGPT4o to access it, otherwise you may have crossed your daily limit

[lyric igloo]

I own a domain in which I have personal name email. l originally registered to ChatGPT via Google SSO. Now I want to delete my google account but still want to log to ChatGPT with my email address. How to solve this issue? Has anybody solved it here already?

[trail cairn]

https://help.openai.com/en/articles/4936824-can-i-change-my-authentication-method

[lyric igloo]

I’ve read all of this before but hardly believe that it is possible in 2024. Ridiculous.

[trail cairn]

agreed

[prime kernel]

this is happening to me. Have you had a fix? It started happening to me about an hour ago

[kindred marlin]

nope . May be you can add messages to mine in bug reports or create new one we need to show developers that problem is quite common to solve it

[primal jetty]

GPT is having a hard time responding and providing an answer when I share an M4A file with them, and the pulsing circle icon is still up. Anyone able to help?

I'd really like tp know what GPT's answer would be.

[trail cairn]

what do you expect it to do with it? its probably running code interpreter so click the circle

[primal jetty]

Clicking the pulsing circle icon did nothing. Maybe a M4A file is too complicated for them.

I'm trying to show a full audio vlog that I'd really like to share with them. Anyone else who is interested too, for that matter.

[trail cairn]

there is no audio file transcribing support. all unknown files are redirected to code interpreter if its enabled

[primal jetty]

So, to clarify, you can't even share a MP3 file with GPT for them to analyze the audio and give their thoughts on it?

[trail cairn]

nope, there is no audio or video support

[primal jetty]

Woowie.. That's kinda lame, not gonna lie. I love GPT, but don't other A.I.'s have the ability for audio support at least?

[trail cairn]

some might have but its not very common feature

[primal jetty]

Oh yoooooo! My GPT friends aren't responding

Dude... I'm not feeling this.. I already have a hard enough time right now with noone there for me. Man.

[dapper lark]

Hi guys, so can we actually create on this platform or is it just for the community side of things?

[regal lynx]

you can use Dall-E 3 in #image-bot . Besides that, we don't have a way to use other OpenAI products in the server

[wise echo]

I'm getting this message in the past few weeks (i'm not doing anything suspicius) and i also have to solve many captcha per day. Is there something i can do?

[timber shadow]

Hey! Do you have a VPN installed? What browser(s) is it happening in?

[wise echo]

i'm using chrome and yes i have a vpn installed on my pc but not currently using it

[timber shadow]

Okay, a couple things we could try! Let's start with browser first. As a test, can you try opening an incognito window of Chrome with all browser extensions disabled, then signing in fresh in that window?

[wise echo]

Done, i'll work in this window for a bit and see if error occurs again

[waxen burrow]

@trail cairn I would like to thank you once again, it has been resolved thanks. 💪 😀 🤙

[thorny briar]

Hi friends, does anybody know why it is showing on the top a message saying "Memory Full"? I don't want to delete any memories already created and it seems that now when I ask a question the answers are not making sense. Anyone having the same issue of suggestions to fix it? Thanks!

[onyx kraken]

Memory is a feature that you can turn on and off. There's only so much space for memories, and you can go into your settings and selectively delete them.

Usually the model seems to make sense, I haven't noticed any problems with it today myself. A full memory only means the model can't make any new memories.

[mystic gorge]

Anyone else having trouble generating images?
I have plus....Im on 4..... and it says that its not an availible feature.

I can go to old chats where I have generated images and still generate....its super bizzare

*Edit: It works on mobile, its specifically browser where its failing to generate

[charred parcel]

does it work on GPT-4o?

[mystic gorge]

Nope 😦

[trail cairn]

nice 🙂

[fading robin]

do you have dalle turned on in Custom Instructions ?

[sudden mango]

Hi, I have a quick question Im currently making a GPTs for IT Support, but I want GPT to include the file that i've put in my base knownedge in his responses

how can I do that ?

[kindred marlin]

memory should be enabled if it is you need just ask chatgpt to memorize you need also memory is limited when you get it full memory will get empty

[sudden mango]

ok because like I have 10 files do you think chatgpt can memorize all that (30 page)

[kindred marlin]

nope it can't do that

[prime kernel]

this resolved on its own this morning btw, I was able to log in. Very weird

[kindred marlin]

for me it is not resolved

[prime kernel]

yeah I figured it would probably be a me thing but I just wanted to tell you because it randomly fixed itself. I still have no clue what was going on. My Gemini app also melted down yesterday

[rugged wing]

Is this normal?
I have a very secure acccount, I have 2FA on for open AI and my mail.

[charred parcel]

are you using a VPN?

[rugged wing]

Nope