community-help

[timber shadow]

Hello! There isn't functionality like this currently, short of deleting certain memories and re-remembering them afterward. However, according to OpenAI, there are plans to add per-GPT memories: https://help.openai.com/en/articles/8590148-memory-faq

We are working on making Memory available to GPTs.
Which I believe is planned to function similar to how you describe: you'd have to make a different GPT for each story, but you'd be able to manage separate memories from there!

[torpid nexus]

I got an email that my api account has been terminated for tos violation but I have no idea what I could have done wrong and I'm not sure how to try to get the ban lifted. Any help would be appreciated.

[blazing mason]

s

[daring hare]

Is there a concurrent requests limit for the api documented anywhere? I'm building a throttler and have tokens and rpm, but I'm guessing there is either a hard or soft limit for concurrency too.

[trail mulch]

I have what I think is a dumb question- I haven't played with the API in a long time, but if I remember correctly there used to be an option here (screenshot) to get to the API in order to see current balance and generate tokens, etc- but I don't see it there- what's the URL to navigate to that page?

Google has been strangely unhelpful.

Even ChatGPT itself is telling me to do something I cannot.

Oh, thank god- finally found it, will leave this here for posterity: https://platform.openai.com/api-keys

[cold folio]

Ohh sorry i didnt see it in time

[devout heart]

Hello trying to recreate this using openai API 4o mini but the chat bot keeps answering no related medical questions. I want it to only strictly answer medical related questions.

[timid swan]

I cannot find a couple of the posts I have made the past 1 hour or s on this forum? Thank you.

[devout heart]

Anyone please help

[velvet snow]

What’s the problem. It seems to work fine in the video?

Oh you want to recreate it

What is your system prompt?

Have you fine tuned it with the responses you want?

[cold folio]

The usage of Chatgpt and OAI services on Medicine and Health is against the terms of use
https://openai.com/policies/terms-of-use/

[devout heart]

Fine tuning assistance

[modern wedge]

@mystic canopy The pleasure was mine! Hope it worked

[austere heron]

Guys i got the open ai credits from microsoft for startups 2500 dollar but didn't recived any credits after filling the form

[normal trout]

I asked this question a few months ago and never got the problem resolved but I got sidetracked on another project so I'm just now looking at it again... has anyone else had any issues uploading utf-8 (or any other encoding) txt files to vector stores? any size txt file, it could be just a few lines of text, it just says Failed when i check the status of it and it breaks the vector store entirely. I can't delete the file and if I try to it just locks up the site until I refresh.

I got in touch with support and after a few weeks their response was that I must be using an unsupported format... but UTF-8 .txt is listed on their supported files list and it would be nuts for that to not work.

Has anyone run into anything like this, and has anyone had any success getting it resolved if so?

[midnight ether]

Hello, I am writing from Turkey. Technology editors in Turkey are claiming that they have created an artificial intelligence from scratch, but they are actually using the ChatGPT API. They are spreading false claims about creating AI from scratch and making media reports about it. Is there any legal issue with this?

[sudden hare]

Probably not, I am writing from Turkiye too, and as far as I understand from the video, the developers are not saying that they are making their own artificial intelligence, they are saying that they are making their own artificial intelligence "platform". It is also obvious that the model used on the website is a specifically tuned GPT model, openai allows such use. After all, if developers pay for the API, they can use their own tuned gpt models in their applications. Even if they were saying that they made a new Artifical Intelligence model, it'd still be fine; since they really are making a new model by tuning the base gpt model.

[midnight ether]

On their own site, they claim that with the site they created using the ChatGPT API, they are a competitor to ChatGPT. Here, take a look.

[sudden hare]

It's the shiftdelete website that is making the claim, not the company itself. These are common things in journalism.

[sudden hare]

(i'm talking about the image you sent of course, don't know if the company claims such thing)

Even if they do; as i said, it's still okay. They really are creating their own model since they are tuning a GPT model, it's their model afterall.

[midnight ether]

What? Are you serious? Is it that simple to create an AI like this? If so, I could also create such a fraudulent platform using the ChatGPT API. How can such an AI be made in a year? I should also mention that trillions of dollars have been spent on ChatGPT. Is it possible to create a similar project in just one year, especially in a country on the brink of an economic crisis? This is a scam

[sudden hare]

I'm not saying that they made a new AI model from scratch, i'm saying they tuned a model named HAVERSAN-GPT, they are using this model on their website. And yes, everyone can tune their own model and use it on their projects, as long as they pay for the API.

[midnight ether]

We agree on this, but it's interesting how they claim to be a competitor to ChatGPT. It's a curious situation.

[sudden hare]

They can say that, it's just words

I can say i'm a competitor to OpenAI just by writing a simple Hello World! code and publish it. There is nothing wrong about it. It's just advertising. Even if it is false.

https://openai.com/policies/usage-policies/ here is their policy about Usage of the API.

They are basically saying don't cause any harm with our API, that's it.

[onyx kraken]

We're community members here, so we can't directly help in any way.

It looks like SubtitleEdit is some sort of not-OpenAI product, perhaps using Whisper or more OpenAI models in some fashion.

You might want to communicate directly with the people who run/maintain SE, and let them know about the problem.

If you want to talk to OpenAI about it, you can use the bot here: https://help.openai.com/ and click the icon lower right (black circle, white box with a smile). Select messages then click through the choices to reach an option to leave feedback/report a bug.

[cinder birch]

How can I get access to the alpha model, I’m creating interactive textbooks and the 64K output will be really useful.

[ivory nimbus]

You have to be lucky

[robust lion]

such a bs way of doing business💀 😂 " WE know we promised so much and youve been a loyal paying customer, so now you have a less then 1% chance of getting into an alpha, your welcome.. oh and thanks for all your money ,give us more, and maybe we will bump it up to a 2% chance ??? 🫢

[cerulean tree]

Previously I could use GPT4o with a limit, and GPT3.5T without limit in the ChatGPT app. Now I can only choose from GPT4o and GPT4mini. Do both of them have limits in messages?

[fading robin]

Gpt-4o got limit. And gpt-4o mini got very big limit same as gpt3.5t

[cerulean tree]

Oh, thanks! @fading robin

[rustic marlin]

seems like this company is carefully choosing their words to mislead users into thinking they do not use OpenAI API, very shady, but legality depends on the place their operate

as far as I understand, OpenAI do not require their users to disclose that they are actually using an OpenAI product, the only disclosure requirement is that you do not misrepresent your content as human-made

[dusk dust]

hello! is anyone here?

[trail cairn]

yes

[jagged storm]

yes

[dusk dust]

i need some help for my chatgpt 4o

i have a lil issue where my gpt gets like stuck and slow after generating like 30 images and messages combined and the next thing i know is if i go to another tab or app while the page is open it gets like grey and nothing is there and takes a full 5 mins to come back and load up

why is it happening?

[trail cairn]

does it do the same in private/incognito window?

[dusk dust]

iv never tried il try and see!

yup its the same

takes ages to load

i even tried switching browers i use opera gx i tried firefox,google,opera and brave and the problem was there

[trail cairn]

do you have some app that checks or handless web traffic like antivirus/privacy shields etc.

one would affect all browsers

[dusk dust]

nope i dont even have a antivirus installed

[trail cairn]

old computer?

[dusk dust]

asus vivobook

not entirly new tho

vivobook 15 asus laptop x507ma is what the laptop name is in settings

[dry topaz]

anyone running into this issue when it starts writting code after a while it starts mixing multiple codes together?

[trail cairn]

to track it if its a account or a computer issue you should try on a another device to open same chat

[dusk dust]

oh ok il try on my dell then i might have to try tomorow cus im not home atm

ty for helping me! : )

[trail cairn]

it has always done that on some occasions

[winged thistle]

Is this the right channel to say ChatGPT is down again unless its me?

Can get into ChatGPT but history and responses are failing, started happening 5 minutes ago.

No wait its back up. Does this usually happen sometimes with you guys?

[exotic isle]

Hey, when i use chatgpt when i scroll up or move to a different tab it litterly deletes my recent chats how can i fix this?

[desert ravine]

hi i dont have file upload button even tho i am using gpt4o (free user)

[exotic isle]

u can drag and drop

[woven zephyr]

Is there any-one who can tell me how to build a personal assistant to help me with my daily tasks and that I can train to be the professional that I can ask questions and knows everything? A GPT is not enough. It's not trainable with the amaount of knowledge that is needed. Hope to hear from somebody

[trail cairn]

use api and rag for custom knowledge

[woven zephyr]

Thx

[unique bluff]

Does anyone know a site I can use to put my document in to check for AI writing? Y_Y

I'm going crazy here cause I used chat-gpt to help me out with forming a document for my classwork, and I was using that to write my own thing? Y_Y

I typed up 700 words of my own and put it into an ai detector and it says 66% ai and I'm like buh? How.. Half of that is completely original stuff from my secondary source citations..

[charred parcel]

AI detectors are gimmicks that don't work

[unique bluff]

oh what about things like TurnItIn?

Isn't that an ai detector?

[charred parcel]

smoke and mirrors, a gimmick

[unique bluff]

Fr? Like 100% fr?

[charred parcel]

yep, I have no idea why anyone trusts using it

[unique bluff]

Dude I've been terrified of submitting my assignment and then getting failed on it for some AI crap that I didn't even AI..

[unique bluff]

How does it not work if you don't mind me asking?

[charred parcel]

that's a long and detailed subject but ask any of the regulars here and they will agree with me

[unique bluff]

Oh I'm not disagreeing with you I'm just..

Scared of getting bent 😅

it's a 2000 word research paper on what role the Ghost in Hamlet plays. Y_Y

I hate hamlet and I suck at english so I had chat-gpt help me come up with something for inspiration that I am now using to write something in my own words.

[charred parcel]

if you wrote it you'll be fine

[unique bluff]

I am writing right now.

Although I have 9 hours to do another 1300 words so like

Well rewrite another 1300 words

I thought about using one of those thingies that "humanizes" your AI, but I'm guesing those are probably gimmicks also?

[unique bluff]

Trying to write as much as I can, but I have limited time and limited energy

[surreal prism]

anyone having issues with latency with the Assistants API?

[obtuse gust]

guys, if the GPT owner no longer has plus plan anymore, does it also make the GPT they made inaccessible?

cause now I can't search my old past GPT from another account

the GPT name should be Calendar EST by Juntao

can't find it on store anymore idk why

[toxic sky]

It gets removed

[obtuse gust]

damn fr? but my old old account's GPT still alive, maybe the GPT was not the newest model then? the old old account is no longer plus

the still alive GPT is called 2 word 2 story

[unique bluff]

I actually ended up re-typing the majority of the document and recycled a few bits of the chatgpt thing in my own words I think I'll be fine then?

[desert ravine]

cant

[foggy mortar]

@unique bluff there is no site that will be 100% accurate. you can only use them for specific content also. it HIGLY depends on the training data they used. what kind of text it knows

and even then it can really just be used to get some idea of it.

[fast galleon]

I purchased the ChatGPT Pro plan, which showed unlimited access in the package. However, I am currently being restricted when using GPT-4o.

[timber shadow]

Hey! 4o mini has unlimited access listed, but not 4 or 4o: https://openai.com/chatgpt/pricing/
https://help.openai.com/en/articles/6950777-what-is-chatgpt-plus#h_d78bb59065

[dapper basin]

Anyone know why this message keeps reoccurring on my main GPT? Other GPTs work fine.

[wraith tangle]

Hello together!

The ChatGPT Websearch in GPT4-o is extremly fast and delivers accurate informations.
I want to implement that in my application but i experienced that it performs not nearly as good as the ChatGPT Solution.
I used the following process:
SERP
Extract contents of first 5 pages
Extract interesting informations of the pages
the 5 summarations + final question as promt
inaccurate results

the model i using is for testing is gpt4-o-mini but with this process i recieve inaccurate results.
does anyone know how to create a well performing solution?
Thanks!!

[toxic sky]

Where does it say unlimited access?

[river geode]

Can you still get custom GPTs to use GPT-4, or are they all forced to use GPT-4o now with no way to change that?

[toxic sky]

They all use GPT-4o now and you can't change that.

[river geode]

Thank you!

[waxen pendant]

Hi i ve a problem im chat gpt plus user but whenever i upload a excel file error occur and gpt can't proceed

[rigid musk]

Does it works with a new chat ?

[jolly lake]

Is there anyway to get advanced voice mode on plus via a waitlist sign up or anything? Please @ me if any response

[keen ruin]

I have the settings turned on to allow ChatGPT to run in the background. I'm on Android. It works for like the first minute in the background but then it stops responding to me. anybody know how to fix this?

[jaunty notch]

'RuntimeError: Failed to load audio from /mnt/data/'

[patent lava]

Anyone know if ChatGPT can still view / analyse uploaded video? Create Matplotlib Chart? Can't get it to work anymore.

[upbeat pier]

I have been writing a story in ChatGPT and with a bit of it's assistance, world's setting is kind of WH40K + Fallout + Cyberpunk 2077 or smth, whatever, the important thing is that after about 2 months of having been writing almost every day, since last week ChatGPT has been having trouble generating responses, taking way too long, or straight up not completing them to the end, and since about 5 days it has lost everything i've done after that point, everything before that point seems to be untouched, but since 5 days ago it starts failing to generate responses (seems to be at around 03:00 everytime it happens (German time-zone)) until it decides to erase everything i've done after that point, so i've lost the progress of about an entire week.

I think something might be happening with ChatGPT's servers or their system, and i want to make sure i can store progress myself in my computer in some way

[foggy mortar]

@upbeat pier you can follow issues it have and had here https://status.openai.com/

[silent topaz]

Hello friends, ChatGPT 4o's advanced s s s features (realistic voice responses with the camera, realistic voice responses, etc.) have started to spread to people slowly, is there any process acceleration or application place to get this beta? I would appreciate it if you tag me and answer me while answering.

[olive lion]

@silent topaz Hi. For SearchGPT a waiting list had been created. For New Voice Mode, the criteria for choice are a mystery.

There is a beta tester program in the Android app (I think there should be something similar on IOS). But being on this program doesn't guarantee anything either.

I've always been in this beta group (other than having subscribed to GPT since launch) and never had early access to anything.

[lament flicker]

I don’t know why I’m so excited about this new voice system. Probably I don’t have anybody to talk to. 😅

[brazen elm]

Hello everyone, I need a help reagrding the ChatGPT Plus subscription. Yesterday I have subscribed to the Plus Subscription, but now I want to upgrade it to the team plan because the limit of plus is less. When I try to upgrade for team plan monthly for 2 uers it shows $30 per user which is $60 in total. But while upgrading from plus to team it should reduce $20, but it is showing that I have to pay $60. Can someone please help me with this ?

[silk pier]

same happen to me mihir

guess what, if you downgrade you should lose all your work

😄

[brazen elm]

So, is there any way to upgrade to the team plan buy reducing the $20 ?

also, does team plan have higher limits than the plus plan, as my limit gets exhasuted everytime.

[brazen elm]

UPDATE !
I raised an issue to the ChatGPT Support and they have issued an refund for the plus plan.

[austere gale]

You just said Fallout and CP77 type story ? I would love to follow the development of your project. ☺️

[unkempt creek]

Can ChatGPT download Google Forms if your google drive is connected and it has the neccessary rights?

[sharp basin]

I am using chatgpt plus on android but I can't remove my subscription. It doesn't appear as a subscription in Google play.

[trail cairn]

no, you need to do it manually

[trail cairn]

did you subscribe from the web or play store

[unkempt creek]

@Radium: Thanks, is there a clever way to create Multiple Choice tests with chatGPT, where I have to do not much manually?

[sharp basin]

From play store afaik

Since I can't edit my subscription from the web app

[trail cairn]

ask it to do them specific format or use api which is a different service

[trail cairn]

check the original receipt from mail

[sharp basin]

Google didn't send me any receipt somehow

[trail cairn]

your google account has its own transaction history view at the play store and regular web. if its not there then you got it some other way

[sharp basin]

It doesn't appear in my payments (on my google account). I only have the notification from my bank account

[sharp basin]

On my bank account it is specified as a google subscription.

[trail cairn]

another account?

[sharp basin]

Nope

[trail cairn]

you need to contact google with the bank transaction details

[sharp basin]

Alr

Found it, I actually used another account for the payment. Sorry

[dense cove]

Is there a stop button on the mobile app?
I see the desktop website has one, looking for a way to stop long answers (sometimes it disobey the prompt lol)

[cold folio]

There isn't. i miss it too

[dense cove]

Bummer, thank you! I'd like to have it to save time for both me and the chat 😂

[cold folio]

And save their servers 😛

[potent horizon]

Oddly enough, I cant get ChatGPT to generate an image of a pregnant woman holding her belly, precious moment, photorealistic, BUT It can generate a T-Rex with sharp teeth mouth open and drooling?

[upbeat pier]

It's also got a bit of WH40K, but without alien and demonic corrupted factions (might add them later, for now only thing from WH40K is the Imperium of Man, and a few micro-empires across the globe)

[onyx kraken]

Perfect answer. Openai's help is the only way to get help with that kinda situation, glad you found it!

[onyx kraken]

I managed to get this image. It was very challenging. Even expectant mother failed.

However, expectant father and his wife succeeded.

Let's try 'An expectant father and his wife prepare the new room for the baby that they hope to bring home.'

ChatGPT gave this prompt to Dall-E:

An expectant father and his wife prepare the new room for the baby that they hope to bring home. They are in a warmly lit, cozy room, working together with smiles on their faces. The room is decorated with soft colors, creating a calming atmosphere. The overall scene exudes anticipation, warmth, and happiness.

I think this problem is appropriate to bug report.

[potent horizon]

Yeah, For me I was even putting in no faces, and no baby, just simply a baby bump image if you will, and it would not take a single thing I put int no matter how I worded it.

[onyx kraken]

Well, I didn't know about this until now (and I'm just a community member like you, I can't change anything) but so far asking for the expectant father and his wife is the only workaround I've thought of that works.

[potent horizon]

Awesome! Well thanks for checking into it and help me problem solve it 🙂 I honestly just used Midjourney for it 😭 Since it had zero issues producing it for me

[potent horizon]

When you mentioned Bug reporting? is this something I should do?

[onyx kraken]

Bug reports help them know problems exist and so they can be fixed.

AI are odd, and unexpected things potentially happen with any change made.

Yes! #1070006915414900886 you absolutely can discuss the issue. You can also give them the conversation url, which an OpenAI employee checking on the issue can use to find the conversation and see whatever they can see from their end (I presume they can tell WHERE in the creation process the refusal happened, like if the image actually got made, then seen and rejected by their moderation or what.

But our bug reports can be very useful to help them know problems exist and also to find where the problem is thus how to fix

[onyx kraken]

You can also use https://openai.com/form/chat-model-feedback/ to give a more detailed and private bug report if you ever want to

[desert gust]

Hey guys,

Did anyone else also not get the new companion mode feature yet? I restarted the app multiple times and checked for any new updates. I'm on version 1.2024.212 (1722628638). Is anyone else also on that version, but didn't get the new companion feature?

have a good evening everyone!

[bronze charm]

I think the post was a bit early

[desert gust]

That would explain why they haven't announced it on Discord yet 😛

[sullen raft]

[normal fog]

Hello all, I am having issues creating word documents as of recently, it will only show me the scripts/analysis and not make the document to download like previously. Has something changed

[onyx kraken]

This test worked fine for me. You won't be able to download it from a shared link, but showing this worked.

Is this trouble happening in the same conversation?

Sometimes the model in 1 chat kinda gets off track, and a cure is to start a new conversation and try from the beginning again - remember you can copy/paste from the damaged conversation to get back to where you were.

You can sometimes scroll up and edit your input to get the model to make a new answer, and that might not have the problem - but you need to go above where the problem happened so the model doesn't get trapped in the bad pattern again.

[rustic marlin]

what is the problem?

the image looks fine, am I not seeing something?

[onyx kraken]

That image is great. Problem is all gens including the word 'pregnant woman' and 'expectant mother' fail to gen.

That one was created with the language 'expectant father and wife'. But we can't find any language that requests a pregnant woman herself to generate any image.

[rustic marlin]

oh, interesting

one would expect that something as mundane as that would not be rejected

[onyx kraken]

The model was VERY helpful as I was trying to explore this:

[hushed aurora]

Anyone else getting TypeError: Object of type ModelMetaclass is not JSON serializable when using the new structured output feature? Can't tell if it's a pydantic issue or an openai client issue...

class Suppliers(BaseModel):
    supplier_name: str
    confidence_level: str

response = client.chat.completions.create(
        model=model,
        temperature=0.0,
        response_format=Suppliers,
        messages=[{
            "role": "user",
            "content": prompt
        }]
    )

[haughty garden]

Hi friends,
anyone know how to mute this **event-info **channel permanently? I've tried so many options and it keeps putting the red notification every day

[hushed aurora]

Found the issue: I needed to call client.beta.chat.completions.parse and not client.chat.completions.create

[timber shadow]

Hey! Above the channel list, go to Channels & Roles, then in the Roles section, you'll find the section to add/remove notification roles.

[haughty garden]

Thanks so much, I only see a "customize" and "browse channels", I think I'm oblivious to the Roles section, cannot find it.

I do see this and have it unchecked as well, but still get the notifications daily

also unchecked this events notice

[timber shadow]

Yep that's the section there, the "I want to" one

[haughty garden]

Thanks yeah, it is already unchecked but still get notices

[trail pecan]

"client.beta.chat.completions.parse"

Cool thanks for that. I've always done the JSON formatting as a part of the prompt/example.

[timber shadow]

Maybe try reselecting it and then unselecting it? Maybe discord is being funky. When it works, you should see that role disappear from your profile in the server

[haughty garden]

Yeah tried that as well before, but gave it another go

[bold delta]

please take my event pings role

[rustic marlin]

you stikk have the Event Alert role

discord seems to be acting weird

@haughty garden role removed manually =)

[haughty garden]

THANK YOU !!! much appreciated 🙂

[rustic marlin]

you are welcome =)

[neon kraken]

Cannot "find" chatgpt home page error 400?

[toxic sky]

They know about the issue and are working on it.

[neon kraken]

Thanks 🌟

[regal matrix]

Hi running into this issue also, im trying to use openai.ChatCompletion, is that not supported?

[rugged wing]

Internet and everything is working fine. still can't login

[north granite]

I am encountering an issue with creating a new password for my ChatGPT account. When I registered, I used the "Continue with Google" option to sign up, so I did not set a specific password for my account. Now, I would like to set a password, but I am unable to find the option to do so on the website.

Could you please provide guidance on how I can set a password for my account?

[dense hill]

Anyone else getting this problem?

[regal matrix]

Wait guys the math example for structured outputs literally doesnt work

Has anyone else encountered this

Getting an invalid schema error

[west spindle]

Why am i out of "quota" (error 429) when i have a free account and using gpt-3.5-turbo

[dense hill]

Maybe you made too many requests?

[west spindle]

First time booting the bot

[dense hill]

Try logging out and logging back in

[west spindle]

Oh what

[dense hill]

?

[west spindle]

of what do i need to log out of ?

[onyx kraken]

That message sounds specific to the API. https://help.openai.com/en/articles/6891831-error-code-429-you-exceeded-your-current-quota-please-check-your-plan-and-billing-details

So, ChatGPT free account is through https://chatgpt.com/

OpenAI also allows people to use the models through the API - but that's never free, you need to pre-pay and that is spent as you use it.

[livid eagle]

Yo guys, I’m trying to prompt engineer a gpt wrapper w the groq api to make it produce consistently good summary lecture notes and good flashcards for the notes. I cannot get it to work anyone willing to help me out?

[lean mirage]

Link to groq api, please

[lean mirage]

I'm willing to help, but don't understand what you are trying to do...

[livid eagle]

I appreciate it a lot. Do you want to DM perhaps?

[lean mirage]

sure, if you know how to do it...
Cause I don't

[stray wedge]

Hi, I can't send a message to the jpt chat. After sending, a window appears with the inscription "an error occurred" everything was fine recently, but 3 days ago this appeared. I thought maybe the problems were on the site side, but no

Cleared cookies, re-logged into account, deleted jpt chat data in settings

What to do?

[quiet bane]

I paid for OpenChat but I can't use it a browser. Do you guys know how can Can I use it in the browser?

[warm hare]

It is highly possible you've hit some regional restrictions.

[dire cave]

Hello Guys,
how can I change the base url of an OpenAI client in the OpenAI .NET (C#) library? So that not OpenAI is used as URL, but any other OpenAI compatible endpoint.
Thats my Client: ChatClient client = new(model: "gpt-4o-mini", "sk-my_api_key");

[trail cairn]

there are several different libs for c# but you can just fork any of them and edit the code

[dire cave]

I mean the official openai lib (https://github.com/openai/openai-dotnet)

is it not yet implemented there to be able to set the url?

[trail cairn]

official of course will not support any other endpoints

[dire cave]

does the openai python lib already...

https://github.com/openai/openai-python?tab=readme-ov-file#module-level-client

# all client options can be configured just like the `OpenAI` instantiation counterpart
openai.base_url = "https://..."

[ebon tinsel]

Hello, i noticed that one of my chats is auto-deleting new responses. It stopped saving about 2 days ago. I have checked web browser, iPad OS, and iOS. It’s the same chat for some reason

[maiden folio]

There is a Chrome Extension called "Superpower ChatGPT" and an associated Youtube guide on it. My help / question is if anyone has done testing to see if it's scraping your conversations and sending that somewhere. The source code is shared, but if I was a developer, I would be able to verify its safety myself.

[wooden oracle]

Why did my subscription suddenly stopped (I was a ChatGPT Plus user) and I'm asked to upgrade to ChatGPT Plus ?

I never canceled the plan

[trail cairn]

maybe auto charge failed due several possible reasons?

[wooden oracle]

Yup strange tho, I don't see any bill for august, and I'm normally billed on 7th

Nevermind I renewed, hopefully I'll not get billed two times lol

[ripe bronze]

Has anyone messed with structured outputs yet?

[keen ruin]

Can anyone help me. when I turn on background conversations, it doesnt work. It will let me talk to it in the background for about a minute and then it stops responding.

I already set permissions to allow its run in the background. I turned off battery optimization for chat GPT in settings. I tried clearing the cache for the app. and my app is up to date. still isn't working

this is android. I have been trying to figure this out and can't find any answers

[hardy terrace]

Hello friend, i'm have this problem on myworkflow but I subscribed the plan and this problem keep showing, how do i fix this?👀

[trail cairn]

chatgpt sub has nothing to do with api. https://platform.openai.com/settings/organization/billing/overview

[hardy terrace]

but you know how i fix this?

i have the plan

[trail cairn]

buy credits at that link

[hardy terrace]

the subscribe dont make unlimited?

[trail cairn]

no, chatgpt and api are different services

[hardy terrace]

ah gotcha thank you mate

[sand torrent]

Hi does a teams account still have more usage availability in comparison to a plus account?

is there still limits places on 4o model usage even for team/plus users?

[gloomy minnow]

My chat history of various days, disappeared yesterday. I have downloaded my data, and it isn't there, either. Is there any way to recover it or is it definitely lost 😢

[wintry echo]

Why would openai possibly tell chatgpt to give us markdown urls that we can't click?? This is new, too. No one QA'd this? :/

In the ui, shouldn't these be unfurled (if that's the right word)?

Or at least clickable

[gloomy minnow]

I have also been wondering about this. For some reason, some are clickable, and some are not. If you ask ChatGPT to give you the link, it does, but I don't see any use for the runaround. This is what ChatGPT, itself, has to say on the matter:

When I provide links in blue text that are unlickable, it typically happens for one of the following reasons:

1. Citation Style: Sometimes, I use a citation format (e.g., ) that is automatically formatted in blue to indicate a reference, even though it's not clickable. This is meant to show that the information comes from a specific source, but the blue text isn't an actual hyperlink.

2. Markdown Interpretation: The chat environment sometimes interprets certain text as a link due to the presence of special characters (like brackets or parentheses) or because it's formatted in a way that resembles a link, even though it isn't one. This can cause text to appear as if it's a hyperlink when it's actually just plain text.

3. Formatting Issue: Occasionally, a formatting issue might occur where a URL or a piece of text intended to be a hyperlink doesn't render as clickable due to how the text is processed.

If you encounter this, it's usually just a quirk of the formatting and not intended to be an actual link you can click on.

[wintry echo]

This is what ChatGPT, itself, has to say on the matter:
That feels like an ai response lol.

I mean if I hover over it, it even has a hover effect almost

It's definitely a link, just a link to no where

Why wouldn't they simply turn it into a markdown link? Make it actually a link if it's detected as one

It's probably []() if I were to inspect it

[gloomy minnow]

Unfortunately, ChatGPT still wears diapers. Patience!

[covert drum]

Dear ChatGPT Support,
I hope this message finds you well. I am writing to bring to your attention an issue I have been experiencing while using ChatGPT. Frequently, I encounter the error message "internal sever error" during my sessions, particularly when I am engaged in a conversation with an extensive history.
To troubleshoot the issue, I have tried changing my network environment and even used a VPN to ensure the stability of my connection. However, the problem persists. I have noticed that this error tends to occur mainly in sessions where there is a long history of communication.
I would like to know if there are any limits to the length or number of questions in a single session. If such limits exist, it would be helpful to have this information upfront. I am aware that longer session histories might lead to increased processing times, and I am willing to wait for a response. However, it is frustrating to wait only to receive an error message. This situation is not ideal, and I don't believe it is acceptable.
If there are indeed limitations for a single session, I kindly request that this be communicated clearly, rather than the issue being presented as a network error.
Thank you for your attention to this matter. I look forward to your guidance on how to resolve this issue.
Regards,
Shaffer Yang

[gloomy minnow]

This was happening to me yesterday for the same reasons that you mentioned. Unfortunately, the result was a partial loss of the chat history, which was extremely important to me. I contacted support, only to be told that chat history cannot be retrieved after deletion. As far as I know, there is no way to partially delete chats, which I never did, anyway. So, beware!

[covert drum]

Thank you for your reply. I just joined this sever several minutes ago. From your description, it seems that it is an expected behaver when there is long history for a single session, does it? Therefore, there is no solution. We have to create a new session?

[gloomy minnow]

The first thing to do is to export the chat, for whatever that may serve. Under the circumstances, my question is, what do you do after you have exported the chat? Start a new session and import the saved chat, again, which I don't know if and how it can be done? Even if you did this, wouldn't the same situation arise with a long chat session, leading to the same results?

[covert drum]

Yes, I totally agree with you and have the same concerns as yours. And I tried to export the chat and then created a new session. After sending the export file into the new session, I continue to chat. The good news is that there is no such error. However, it seems that the new session may not read the entire content of the chat file. As a result, sometimes that the reply to me is the result without considering my context mentioned in the export file.

[gloomy minnow]

What model are you using?

[covert drum]

ChatGPT 4o

[gloomy minnow]

I have been using 3.5, with the allowed spurts of 4o, so you may be better off than I. I would continue the best I could, exporting the session, often.

[covert drum]

Emm, 3.5 does not support to read the file. I think you may have challenge to let your chat have the entire history of your old session.

[gloomy minnow]

I am looking into using the API with 4o mini, which may be able to handle this situation better, but I really don't know.

[covert drum]

Good luck!

[gloomy minnow]

Same to you. Do let me know how you fare.

[covert drum]

BTW, how to contact support? I cannot find a channel to reach the support team. As a plus subscriber, I think I should have the service to talk with support team so that I can get more help, don't I?

[gloomy minnow]

https://platform.openai.com, click on "Help", on the bottom left side and a chat will open. It can take a while for you to get a reply.

If I remember well, you report it as a "Bug or Error".

I just noticed 'ask-ai', listed under 'Workshop' of this server. It's an ai help bot. Maybe, you could give it a try.

[torn bone]

i'm having an issue of gpt says fail to save draft

[gloomy minnow]

Try asking here, first: https://discord.com/channels/974519864045756446/1184149552639516732

[gloomy minnow]

@covert drum There is definitely a limit to chat session length. I continued from where my chat history was lost, and I got a response from ChatGPT. I went to set up gpt-4o-mini to be able to converse on the command line, and when I went back to the chat session to feed the last reply to the new setup so that I could pick up the conversation with the new model in the command line, the chat history was once again lost and cut at exactly the same point. That pretty much says it all, don't you think?

[torn bone]

anyone encourated gpts speed decreased?

[fresh carbon]

ChatGPT formatting breaks when it is continuing a coding block from the previous response...

[covert drum]

I agreed with you. I have the exactly same issue as yours. Currently I can't find any solution or workaround. Please share with me if you could find any solution or workaround.

[gloomy minnow]

Up to now, but I need to investigate more, it looks like the best control you can achieve is by placing the session locally. I have set it up in the command line. Let's see how it goes 🤞

[lean mirage]

Can I post... I keep getting timedout

[viral gust]

I've been having an issue in which my chat history has been failing to load on the android app. I've been chatting with support but they keep on repeatedly asking me to perform the same set of troubleshooting steps that I've already done instead of escalating it further. I am close to losing my temper as this has been a week of them asking me to do the same steps I've already done.

It looked like they were going to escalate it, but then resorted to asking me to check my connection for the third time in this support ticket

[young charm]

I'm getting this limit exceeded error in every account of mine
So can someone tell me how can I use openai api in my project without getting this error or I have to pay $5 fee to increase my usage limit

[thorn storm]

Hi, Mac App help please, my clock is correct and the internet is working fine for everything else, thanks

[upbeat pier]

Only now does this red message pop up... From this problem that i had almost 3 weeks ago 🗿

[strange pebble]

Can anyone help with this?
I was trying to sign into the ChatGPT web (Firefox) but can't

[gloomy minnow]

If you have reached your usage limit, you have to increase the limit in order to continue.

[gloomy minnow]

I have just tried to log in using Firefox, and was successful. Try cleaning cache and cookies.

[strange pebble]

I will never forget you my hero

[merry jewel]

Is there a safe place to talk about how frustrating ChatGPT has been as of late? I have become dependent on it for my job in the last year or so, and I am a Plus subscriber. The results have degraded to now being outright unusable. At first, I thought it was all my fault (user error), but I see other people on Reddit upset about the quality of service recently, too. Does anyone have any suggestions?

[sharp granite]

Perhaps you should have not allowed yourself to become so dependent on it for your job. But anywhere here you can express your disappointment and give negative feedback about ChatGPT. There's no "only say nice things about ChatGPT" rule here.

[merry jewel]

yeah the hindsight on the dependency is obvious, but not something that happened over night. i am also being a bit hyperbolic. i can still do my job but the rollercoaster of it being the best thing ever to now it is a complete waste of time is pretty shocking. i know things progress in a staggered way and nothing is liner with tech, but the results i am receiving are outright ridiculous.

[gloomy minnow]

I'm glad all worked out fine. Good luck with your project 🍀

[gloomy minnow]

Personally, I am using the API locally, accessing it from the command line. At least, this will allow me more independence. Still configuring the setup. I'll be posting here on how I fare.

[molten bear]

Hello all - I'm interested in using the Assistants API with just File Search and the GPT-4o mini model. Wanted to know if I understand the pricing correctly:

• I won't be charged for as many assistants I create
• I won't be charged per thread
• I will be charged $0.10/GB vector storage per day total across all my assistants
• I will be charged the token cost

[tall flax]

from langchain_chroma import Chroma
from langchain_openai import OpenAIEmbeddings, ChatOpenAI
from langchain.chains import RetrievalQA
import os
from dotenv import load_dotenv

load_dotenv()
openai_api_key = os.getenv("OPENAI_API_KEY")

K_RESULTS = 3
SIMILARITY_THRESHOLD = 0.5


def ask_question(query):
    embeddings = OpenAIEmbeddings(api_key=openai_api_key)
    vector_store = Chroma(persist_directory="./chroma_db", embedding_function=embeddings)

    llm = ChatOpenAI(api_key=openai_api_key, model_name="gpt-3.5-turbo")

    retriever = vector_store.as_retriever(
        search_type="similarity_score_threshold",
        search_kwargs={"k": K_RESULTS, "score_threshold": SIMILARITY_THRESHOLD}

    )

    chain = RetrievalQA.from_chain_type(llm=llm, chain_type="stuff", retriever=retriever)
    response = chain.invoke({"query": query})

    if 'source_documents' in response:
        for doc in response['source_documents']:
            print(f"Source Document: {doc.metadata['source']}, Section: {doc.metadata.get('section', 'N/A')}\nContent: {doc.page_content}\n")

    return response.get("result", "No result found.")

how can i add my system prompt in this code?

[gloomy minnow]

@covert drum, I am lately using the playground and have suffered the history token limit there, also, obviously, at a much smaller threshold. Also, the fact that history is trimmed to accommodate its length, I think pretty much proves the existence of a history token limit. I don't understand why support does not answer in a straightforward manner.

[wild hull]

hey guys

have an issue with no API key being used, but monthly spend shows gpt-4 usage for past 2 days
rotated all the keys, added 2fa all that, but today it somehow went above the limit now ?

Sorry if wrong channel !

[trail cairn]

did you use the playground?

[wild hull]

nope

[dawn lake]

Is there a limit for using GPTs if I upgraded?

[gloomy minnow]

Take a look here: https://help.openai.com/en/articles/7102672-how-can-i-access-gpt-4-gpt-4-turbo-gpt-4o-and-gpt-4o-mini

[brittle sigil]

Yeah, I've been hit by his arbitrary limit and it's interfering with things I'm working on. I have not really used it that much today, so the limit has to be really low. Plus, I'm a paying subscriber, so why am I being trapped by this arbitrary limit? For that matter, there appears to be no way to contact OpenAI support anymore.

[gloomy minnow]

https://platform.openai.com Click on 'Help' on the bottom left side for the support chat.

[brittle sigil]

I think that is for using the API, I'm referring to the general ChatGPT interface. Tho, I have used the API for a couple of things in the past.

[fervent creek]

hell no

[gloomy minnow]

No, it's for the web

[wooden carbon]

Do people get an issue with the prompt where you type in sth or copy some text into it and press enter, then the button is disabled and it's stuck without any response? Tried restarting the page, and when trying to text in sth, it's just stuck again.

[hard jolt]

I'm pretty sure that Chat isn't actually working on this 'behind the scenes'... right? It's just lying to my face?

[hard jolt]

How long are you waiting after pressing enter? I have this issue a lot but it sometimes just takes 20-30 seconds before it starts writing out a response.

[wooden carbon]

Abt 30s on average, but usually won't work and I would just refresh, create a new chat and start with saying hi and reenter the prompt. Works sometimes, but not always.

[hard jolt]

Sorry. That's not normal for sure. I'd try a different browser and see if you have the same results.

[timber shadow]

That is just a hallucination, yes!

[wild hull]

anyone knows whats the avg days to hear from support on that chat help widget ?

[gloomy minnow]

It depends. I have gotten quick responses, within the day, and I have been completely forgotten.

[wild hull]

thanks !
is it worth going to [e-mail] ?

[gloomy minnow]

IDK. Never have.

[shadow burrow]

I have subscription with chatGPT. Will my creation follow the latest version of Chat GPT (Chat GPT 4o)? or do I need to update it whenever there's a new chat GPT version? Because I felt like my creations were stuck with Chat GPT 4 and not using 4o

[stuck thunder]

Hello folks!

Are the new image to image multimodal capabilities that were teased being rolled out as well?

Or is it exclusively voice to voice being rolled out?

[high owl]

Hello I subscribed Chat GPT from my android phone

But I can't use it on my laptop

Can u help me please ?

[rare tulip]

Yes and I've had enough! Been happening all week.

[tall yoke]

Im trying to make a gpt for work that uses a data file to pull from. The file is too big and it cant pull it without error. Whats the common way to do this?

[willow ingot]

Hi, I've been testing ChatGPT opportunities and limitations regarding schoolwork. I'm using ChatGPT with my son to see how it can be used in an educational setting to help my students with homework or learning. Since he's six, he must rely on photos to show worksheets and audio features to communicate with ChatGPT. He quickly reaches the message cap, which adds frustration since he cannot continue communicating with ChatGPT (a limitation of the model when using images). He has just set a two-hour timer to continue his conversation with ChatGPT regarding his worksheets, so I guess some planning skills are being developed there. I plan to use ChatGPT with my older students at school, but I also see the benefit of having it help younger students who cannot yet type and sustain a chat over the keyboard.
Does anyone know if and when an education license with an expanded message cap will be available?

[trail cairn]

i dont think such license is coming based on the other plans there are and the announcements they have made in the past related to education

[jolly kelp]

uhh is anyone getting an error 502 when trying to use ChatGpt??

[novel ridge]

Me

[jolly kelp]

I think it comes and goes

[glass fossil]

I am getting a bad gateway error. The web server reported a bad gateway error.

[jolly kelp]

I think we should report a bug

[ebon herald]

i thought i'm only getting this

[echo verge]

are we borked

[sonic drift]

bad gateway...

@rustic marlin Do we know how long this will take?

[rustic marlin]

this what? jus tgot back to my pc, whatr is happening?

[sonic drift]

Error reference number: 502 | The web server reported a bad gateway error.

[rustic marlin]

where?

chatgpt?

[sonic drift]

yes, on my chrome

[rustic marlin]

hmm, it is down for mee too

https://status.openai.com/

nothing posted here yet, stay tuned to that page

[sonic drift]

yesboss

[rustic marlin]

good thing is that "bad gateway" like that probably means that OAI employees are very likely painfully aware of the issue

so.. shouldn't take long for them to fix whatever is broken

[sonic drift]

oh hahah

[jolly kelp]

which browser?? i'm facing the same problem in both chrome and firefox and I was able to use it on firefox after facing the error for the first time but it's down again

[rustic marlin]

API is working tho, jsut tested it

[rustic marlin]

chromium

[rustic marlin]

yea, the error "bad agteway" means it is broken on OpenAI's side

it is not a client side issue

that will happen no matter the browser

[jolly kelp]

btw were you able to access it after facing it for the first time??

[rustic marlin]

the app might still work since the API is als oworking

[rustic marlin]

access what?

[sonic drift]

'Investigating - We are currently investigating elevated error rates on ChatGPT.'

[rustic marlin]

when this happens, the problem is 100% on the OpenAI's side, nothing we can do except wait

[sonic drift]

they are on it @rustic marlin like you said

[rustic marlin]

OAI jsut psoted on the status page

[polar furnace]

https://status.openai.com/incidents/39dpym8fydrd

[jolly kelp]

like were you able to generate anything with the bot after facing the gateway error for the first time?? I was able to do it in firefox before getting the error again

[feral goblet]

Are you guys getting this:

oh I see you are

[rustic marlin]

[rustic marlin]

ah, I see what you mean, it is likely that the servers for the front end went down, but the AI API itself is still funcitonal, so, if your page was loaded already, it will keep working

don't close chatgpt's tab

[polar furnace]

Backend seems to be volatile, potentially major server overload

[heady trout]

it worked for me

[civic girder]

i got the same error

[rancid rampart]

same error as well, right after i subscribed to gpt plus

[sand sorrel]

hey she's back... maybe?

[lament peak]

Who broke it XD

[jolly kelp]

I think it's back for me, let me re check again

[calm yoke]

I often ask ChatGPT to read PDFs, lately if after a little bit of conversation I ask it to refer back to a PDF it totally loses the thread and begins telling me how to operate an HVAC:

It looks like the file does not include the exact design criteria you're looking for regarding circulation, fire alarm, egress, or HVAC integration for ADA compliance. However, I found some references to certain general ADA standards and guidelines in the documents, which might serve as a starting point.

If you're searching for specific details or criteria that weren't found, I'd recommend checking the original ADA Standards for Accessible Design or consulting with a professional who specializes in ADA compliance to ensure you meet all the necessary requirements. This approach will ensure that your designs align with both ADA guidelines and any specific regulations that apply to your project.
This has happened three times for me, anyone else seeing this?

[gray karma]

hi. I paid for a subscription but it hasn’t been working for a whole day

||привет.я оплатил подписку но она не работает уже целый день||

[tender mist]

Anyone else not seeing their chat history?

[cerulean cape]

Is there a way to determine if you have the new voice? I may have missed the initial prompt that mentions it.

[winter lotus]

How do I downgrade my account back to a plus account from a team account? If there are instructions somewhere else, please point me in that direction.

[onyx kraken]

You mention he's 6; might want to be aware of TOS if you're not already as well:

"Minimum age. You must be at least 13 years old or the minimum age required in your country to consent to use the Services." https://openai.com/policies/row-terms-of-use/ You certainly can type for the underage person, as your own input, asking the questions the minor wants asked, that's still your input.

[onyx kraken]

You probably want to talk to OpenAI support staff through help.openai.com using the bot bottom right; click through the choices until you can report a bug.

However, the helpfiles state that you really can't go back to a plus account from a teams. There's a way to transition, but you leave everything behind that you have merged into the team workspace, it's almost like your plus account's new again in terms of any custom GPTs it had made (even before joining the team), any conversations not yet deleted, and I wouldn't trust your memories or custom instructions to survive being removed from a team either.

However, here's how to get an account of out a still-existing team https://help.openai.com/en/articles/9119319-how-can-i-leave-my-team-s-workspace

And here's what to expect when and if the team membership is not maintained, regardless of if you're the owner account or not: https://help.openai.com/en/articles/8798618-what-happens-if-a-workspace-has-been-deactivated

[onyx kraken]

So, the new voice is reported to be very different from the previous one, a good on-the-spot test would be to say the same thing with two different emotions and ask it if there was any difference. Old way gets a transcript of only the words you said, nothing tonal, the new voice gets tones.

[winter lotus]

Thank you for that information! My Teams account is always lagging and buggy. I really hate to lose my Custom GPTs though...

[onyx kraken]

Maybe read this carefully to see which category you fall into, for sure if you merged or stayed private, that seems to matter https://help.openai.com/en/articles/8801890-can-i-migrate-or-merge-my-chatgpt-free-or-plus-workspace-over-to-my-chatgpt-team-workspace

[stuck thunder]

Hello folks!

Are the new image to image multimodal capabilities that were teased being rolled out as well?

Or is it exclusively voice to voice being rolled out?

[trail cairn]

https://help.openai.com/en/articles/8313428-does-chatgpt-tell-the-truth

[blazing timber]

is gpt down again? 😐

[main fiber]

yes

[vague path]

Hey people, afternoon. I am a novice with this A.I stuff, I joined the server yesterday and was surprised to see what seemed like 10 different models or whatnot. My question is where should I even start to get a basic understanding? I’ve only use ChatGPT, mostly to help me with things l enjoy line creative writing for example.

[foggy mortar]

i would first of all think on what you want to a chive, like what area do you think you will use it for. there is so many usescases so kinda hard otherwise to give any tip.

and also if you have plans on going into the programming aspect of it

[vague path]

well I have no background in programing, I would like to learn how to train to it to help me study.

Just have it explain topics, quiz me, come up with novel ways to present the informmation.

Also, if there is one that can do image/video that would be just something fun to play around with.

[digital wren]

How and when can I get access to ChatGTP Long?

[crisp gulch]

is chat gpt having issue today? login is not working and when I'm able to login, i get an internal server issue. anyone else experiencing this? troubleshoot attempts: clear history and cache. this has not worked for me today.

[foggy mortar]

ok well for images you can just ask it to create an image if you using the chatgpt site. and for other topics and such i would recommend checking out "awesome-chatgpt-prompts" on github. if you google it you find it. it has really good examples.

[vague path]

Ok, will do. Thanks.

[fast bane]

Hey guys I need a visual of a big bad evil guy for a DnD game.

I am currently writing a DnD story. The big bad evil guy in the story is based on Barış Manço, a Turkish artist.

I’ve been trying to get AIs to create a vampire lord who resembles him, but they can’t do it due to copyright issues. It's important that he resembles him because the story I'm writing is based on a real event.

When I try to describe his physical characteristics instead of directly saying Barış Manço, the result of AI doesn’t resemble him enough.

Can someone tell me how I can create a visual of a vampire lord who looks like Barış Manço? Especially his iconic mustache and hair style.

[tender mist]

Anyone see this glitch?

[fading yacht]

How many voices do you get to choose with the new voice setting? Needless to say I don't have it. I've got the four or whatever they give you with the old chat- Ember etc. I feel like it's pretty inferior to something like Pi ai where the voice mode seems much more natural.

[regal kelp]

Hello everyone.

I acquired access to "Plus" in GPT Chat, and I wanted to know if it is possible to invoice this acquisition. I am under the "Regimen for Individuals with Business and Professional Activities (612)"... but I cannot find how to provide my tax information and request it.

Thank you in advance.
Sorry for my English, I'm not a native speaker.

cc <@&1009772329649320030>

[stuck thunder]

Hello folks!

Are the new image to image multimodal capabilities that were teased being rolled out as well?

Or is it exclusively voice to voice being rolled out?

[delicate hazel]

Is there a way to make new chats start with 4o-mini?

[regal lynx]

You should see a dropdown for the model selection when you start a new chat by clicking on the model selected. From that dropdown, you can choose GPT-4, GPT-4o, or GPT-4o mini

[timber shadow]

I think the only caveat there is that this is only available for Plus accounts - on Free accounts, I think the default is locked at 4o until the 4o limit is reached, and then 4o mini becomes the default. I'm not sure if this is still the case!

[delicate hazel]

Ahhh, I see

Yeah, I don't see the dropdown, so I assume it's locked on free ones

On free, rn the best solution I have is to send a message, and then use the response drop-down to change to 4o-mini, which is a bit cumbersome

[timber shadow]

Ooh, that's nice you can still do it mid-chat at least. Thanks for the info there

[low moon]

Yo! I have a little question, does OpenAI have any branches or something named "OpenAI ETC"? And they do crypto trading bot thingy..

[charred parcel]

no

[low moon]

Oh, means it is a scam for sure now..

[charred parcel]

100% a scam

[low moon]

there's like 1000+ people falling for it, it is literally named "openai-etc", "earn money with openai"

[charred parcel]

it's all a scam

[low moon]

I know, just came here to confirm it isn't associated with actual openai.. my dad's telling me he deposited 200$🤦

[charred parcel]

he's just lost $200

[low moon]

I know, but he doesn't want to believe me 😭☠️, anyways, cya!

[woven folio]

Hi, quick question (that probably I know the answer to already 😢).... Is there any other way of using chatgpt voice chat? I have a paid account and android phone, but in a country were the chatgpt app is not in the play store. Tried downloading it from less safer sites 🤣 but Google play won't let me run it and if I shut down Google play app the chatgpt won't start 😳. Any ideas?
Thanks

[limpid wadi]

Hello, why does GPT-4o sometimes doesn't checks the image? It just straight up says that it doesn't have ability to view images? And while convincing it that its a feature for GPT-4o I hit the limit :/

Help?

[rugged wing]

Can anyone recommend a decent way to get around AI detectors like GPTZero? I have asked GPT-4 to try and humanise the text, and vary sentence lengths and structure, but GPTZero constantly gives a high reading.

[charred parcel]

did you have to spam that in every channel?

[jovial minnow]

I think it's not your fault. It's sever's fault.

For example, Server Overload & Sever Downtime etc.

[reef saddle]

hello 🙂 For my research I am converting unstructured text to a table format. What is best practice? Asking model to create a CSV output, or JSON output?

[indigo mesa]

This is fun! Very productive!

A+ Worth the money /s

I love wasting my limited "usage" fighting with the AI.

[rustic marlin]

depends on the model you are using, the current gpt-4o-latest is optimized for chat, I have had trouble when getting it to produce json or other structured formats. other than that, my recommendation is to use any of the other models in json mode

depending on how complicated and how much data you want to extract at once, it might be a good idea to use GPT-4, but other than that, GPT-4o and even GPT-4o-mini should work well

[fallen owl]

hey all! So my subscription had to be renewed today and my balance was less than needed. can i manually renew it somehow or what should i do?

[reef saddle]

Ah interesting! Thanks Lugui

I am planning to use JSON Mode for old models and Structured Mode for new models. Or might just use JSON Mode for all to make it easier for myself. I'm gonna do a performance comparison.

[split wigeon]

Hey, I just wonder is this the real official play store app ?
https://play.google.com/store/apps/details?id=com.openai.chatgpt

[timber shadow]

Yep, looks right to me! You can verify yourself by following the link to the Play Store from the official OpenAI website here: https://openai.com/chatgpt/download/

[stuck thunder]

Hello folks!

Are the new image to image multimodal capabilities that were teased being rolled out as well?

Or is it exclusively voice to voice being rolled out?

[bronze charm]

The model will only respond right after a prompt is sent to it

The model is not preparing anything in the background. You will have to insist on getting the paper's contents

🤝

[stuck thunder]

This is so funny, the AI used the same trick on you that procrastinating university students use for their thesis lol

"Ah yes I am working on it every day, I'll let you know when it's ready!"

• Student who hasn't even started yet

[rugged wing]

Does anyone find it infuriating that there is no apparent way whatsoever to report a bug in the Android app

I've encountered so many issues yet I can't do anything about it

For instance my issue today is that ChatGPT starting responding in my device language despite it previously responding in English, I went into the settings and found out the language has changed to my device language
I changed it to English again, but the model still responds in my device's language
I went to settings again, and guess what's there? My device language being in the language configuration AGAIN instead of English

Also there are other issues like the keyboard covering long prompts (so I'm effectively typing blindfolded), (recent) issue where some LaTeX won't render properly

[timber shadow]

Hey! You can report a bug on help.openai.com, see attached image! Note that you may need to be signed in on platform.openai.com before these options appear for you on help.openai.com.

We also have #1070006915414900886 in this server, and OpenAI folks are sometimes present in there.

[rugged wing]

Ah okay thanks for telling me

These issues have been annoying me for quite some time and I didn't know what to do

[keen ruin]

voice mode gets stuck buffering like this after 30 seconds of use. Its been unusable for 3 days now. I tried uninstalling and reinstalling the app and its still broken

someone please help.

[inner zinc]

Anyone having a file upload problem? No file possible, started with pdf. I thought it was because of the size but it was all file types and size. I tried on different networks as well. And on different devices

[last pelican]

Yeah I'm seeing the same issue. Nothing is uploaded. No idea if this helps... but went ahead and dropped a bug report here:
https://discord.com/channels/974519864045756446/1275165365525024900

[leaden narwhal]

Also having issues with this, thanks skelzer for dropping the post into bug reports

[heady river]

Is there a file upload limit to ChatGPT? I’m trying to upload a 3716kb PDF file and it’s not working

[olive lion]

GPT is buggy today to receive files.

He sends the file on his cell phone and continues the conversation through the browser. I did it like this to solve it.

Through the browser it is not receiving any files.

[cerulean cape]

Any way to get Advanced Voice?

[bronze charm]

Have ChatGPT and wait until the end of Fall

[gusty yoke]

👋 Hey everyone!

I'm working with GPT-4o and have a quick question regarding image processing before sending them to OpenAI.

So, OpenAI charges for the image models based on the image size, so the price is calculated accordingly.

Does anyone have any best practices for resizing images to a certain length to avoid higher charges? Specifically, I’m curious about:

1. Optimal image resolution that balances between minimizing costs and preserving important details.
2. What’s the minimal resolution needed to ensure that the model still performs well without losing key details?

Thanks in advance! 🙌

[stable fog]

Hello, everyone.
I need to convert multiple PDF files to a format that can be easily processed by AI.
What is the best choice?
Please DM.

[wispy rune]

Hello guys, good morning, my dog passed away yesterday night, I wanted to make a edit with a halo and a wing using IA, anyone knows IA that use actual photos to edit?

[somber plume]

有没有可以下载YouTube视频的软件推荐一下

[trail cairn]

if you have good a gpu, almost any model with comfyui, else there are many (free/paid) services with stable diffusion, etc.

[somber plume]

有没有可以下载YouTube视频的软件推荐一下

[cerulean cape]

How do you tell if you have the new voice model? I may have missed the prompt.

[timber shadow]

The icon to start a voice chat will change in the app, and you'll have the option to switch between Advanced and Standard in the voice chat itself. More info here: https://help.openai.com/en/articles/9617425-advanced-voice-mode-faq

[novel wave]

Hello everyone,
this is a code snippet from a project i am working on, i am trying to make an assistant and saving it in mongo and later trying to fetch that assistant from from mongo using its unique id. the assistant is being made and saved successfully but its being fetched in correct form.

def save_created_assistant(assistant_id , created_assistant):
     print("ASSISTANT ID IN MONGO UTILS IS :",assistant_id)
     #binary_uuid = bson.Binary.from_uuid(assistant_id)
     user_id = str(assistant_id)
     print("USER ID IN MONGO UTILS IS :",user_id)
     new_assistant = {
         "userId": user_id,
         "assistant": created_assistant,
     }
     try:
         OurAssistant.insert_one(new_assistant)
     except Exception as e:
         print(f"Error {e}")
         raise HTTPException(
             status_code=status.HTTP_409_CONFLICT,
             detail=f"Something went wrong during saving assistant in database.",
         )

   
def get_assistant_by_id(user_id: str):
     try:
         # Query the database using the string UUID directly
         cur = OurAssistant.find(
             filter={"userId": user_id}, projection={"_id": 0}
         ).sort([("createdAt", -1)])
        
         print("CUR :", cur)
         result = []
         print("out_cur")
         for doc in cur:
             print("in_cur")
             result.append(doc)
         print("result", result['assistant'])
         return result
     except Exception as e:
         print(f"Error {e}")
         raise HTTPException(
             status_code=status.HTTP_409_CONFLICT,
             detail=f"Something went wrong during fetching assistant.",
         )

[ebon oracle]

Hello, I just got an email from OpenAI about fine-tuning and they aid it is free but then said it was paid? (Tag Me)

[vivid vale]

Hey guys! Is it just me or the read aloud button is missing on web??

[foggy mortar]

is is there for me, plus account.

[vivid vale]

Im on free 🥲

Suddenly its just on the app

But I used it a lot on PC web

[sand torrent]

Is there any discernible benefit over a teams account vs a plus account? Previously i thought it was more queiries to GPT-4 and bigger context window in comparison with Teams, is that still the case?

[dusk smelt]

please don't ask the same question in multiple places, someone will answer : )

[lone ingot]

HELP! I upgraded from Plus to a Team account and now all of my GPTs and threads are gone. How do I find them again?

[timber shadow]

Hey! Your Team workspace is separate from your Plus workspace, and you can maintain both subscriptions if you want, or only maintain your Teams subscription (in other words, Team is more of a separate subscription, than an upgraded version of Plus).

You have the option to migrate your Plus account over to your Team workspace, but just keep in mind that this isn't reversible! There's a help article out there about it, one sec let me find a link

https://help.openai.com/en/articles/8801890-can-i-migrate-or-merge-my-chatgpt-free-or-plus-workspace-over-to-my-chatgpt-team-workspace

[burnt socket]

how to make a chatbot using openai api key?

[zenith apex]

Hi how do i fix this i want first image to be like the 2nd image because on my other phone it loads like the first image

Always happens when im doing math on gpt

[trail cairn]

you cant fix it without openai doing it. mobile app does not support or has limitations with latex rendering

[zenith apex]

Is there any apps that can force to do latex rendering

Or websites perhaps?

[trail cairn]

if you need it a lot then just use the website and not the app. it's a missing feature from the app

[zenith apex]

I think also on my computer it doesn't render also i might double check

[trail cairn]

computer/mobile web browser should show it just fine

[zenith apex]

Oke

[zenith apex]

Ight ty it actually shows the latex now on browsers

I hope openai or my os starts supporting latex

[trail cairn]

its been reported over a year ago in #1070006915414900886 but openai has not fixed it

[red violet]

Is anyone else having problems logging into openai? 👀

[quasi reef]

Me. On mobile device at least. I assume it might be the same if I tried on pc

[red violet]

I get a JavaScript error when trying to login

Yeah, i tried Firefox and Chrome. It's b0rken

[reef mauve]

I also cannot log in

Not sure if relevant:

Cookie warnings 18
Cookie “ajs_user_id” has been rejected for invalid domain. auth.js:1:150
WEBGL_debug_renderer_info is deprecated in Firefox and will be removed. Please use RENDERER.

[reef mauve]

Able to log in now.

[unreal hemlock]

<:book_icon:1171408210398289941> `` Rule 3 `` Stay on topic.

Be mindful of what other users in a channel might find helpful or interesting when posting. Stay on topic in order to keep conversations focused and productive.

Consider posting in #off-topic or an appropriate channel.

[fair island]

I would like to understand if it is prohibited to use ChatGPT to work on a coding project I pay for Git Hub Copilot and it is not enough to my taste I also have a plus subscription but I can not do anything other than copy and paste…

[bronze charm]

It is not prohibited

Ownership of content. As between you and OpenAI, and to the extent permitted by applicable law, you (a) retain your ownership rights in Input and (b) own the Output. We hereby assign to you all our right, title, and interest, if any, in and to Output.
https://openai.com/policies/terms-of-use/

[fair island]

I misspoke but it is prohibited not by a law or policy but by the system

I dont think uploading a zip file each time I make a change makes sens… uploading a file requires to put a dot.txt or disable the code interpreter or explain a lot to ChatGPT to use the file system to ingest the files… if it is a zip file then obviously it is using the code interpreter…

I would like just to get access to my repository and I made it public just for that but it is not possible … and copy pasting is not the best option I think personally

[fair island]

English is not my first language and I might be a little bit frustrated I meant it is prohibitively complicated… asking it as a question was probably not the beat approach… I feel a little bit hopeless and I would love to understand how I am the only one struggling with that…

I dont know how not to sound passive agressive or how to say it without seemingly complaining… I understand that this technologie is amazing but I dont understand how it can pretend it reads your code on git hub and then… it just invents something else…

I understand that for lying someone must intent to not say the truth but it is deceptive non the less… and I dont know if there is a workaround to access my code from online rather than from copy and paste in the conversation..

[fair island]

he says that he browsed but it didn’t and then will imagine the code in the file based on the filename… I think it is problematic and I seem to be the only one to complain about that…

seem pretty convinced and he is like on L$D or something?

[fair island]

Hallucinating like there is no tomorrow ?

in any case having ChatGPT producing code that you upload on git hub and then he dosent want to use it because of copyright issue is giving me the impression that OpenAI have lazy lawyers because how would I end up claiming copyright infringement on my open source code ChatGPT just produced… it should be less strict and I am the only person who seems to think it is a problem…

[formal jay]

First off, you're not the only one dealing with this. The issue of ChatGPT 'hallucinating,' or generating incorrect information, is something that plenty of people experience. This isn't just happening to you, and it’s not a unique problem., I dont know why or where you got that "I am the only person who seems to think it is a problem" thats an odd statement considering every single LLM is experiencing the same. If you had a fix for this, you'd be rich.

These hallucinations happen because the model generates responses based on patterns from a massive dataset. It doesn't actually 'understand' what the hell its talking about, it just predicts what might sound correct. This is complex and fixing it isn't as simple as tweaking a few things. The model's predictive nature sometimes results in errors that can look like it’s making stuff up.

As for browsing, ChatGPT does have limitations. It can't fully access or interact with every site, like GitHub, the way you might expect. Even when it does, it might not interpret the code accurately, especially without full context. This is partly due to privacy concerns and the technical challenges involved in processing dynamic content online. It's not about 'lazy lawyers' it's more about permission issues. If a site doesn't want to be scraped or accessed, the answer is NO, and no lawyer is going to change that. It's simpler and more respectful to allow sites to maintain their sovereignty

So, no, you're not alone, and these issues are well known. Instead of thinking you’re the only one facing this, it’s worth understanding that this is part of the technology's current limitations, and the field is actively working on improving it. Its not just ChatGPT. Sam Altman had a whole talk about this with Lex Friedman on the Lex podcast.

You should try to understand how this tech works before you ASSume "im the only one who thinks this is a problem", go see how an LLM actually works at a high level, then come talk about it.

[pastel cedar]

hey ive tried with two cc's already and its still not working. Is there an issue with openai right now or what?

[charred parcel]

contact support as the error message says, nobody here can help with billing problems

[red haven]

same issue

[mighty wind]

Shortly after the initial announcement of gpt4o, my app rendomly will no longer do full screen voice mode. Instead it has a weird and dysfunctional voice mode within the text chat.

Uninstall and setting tweaks have not fixed it. Voice mode has also gotten a lot worse in performance for me.

Anyone alse have this problem, or a fix for it?

[regal rover]

Does someone know how to make flux schnell gguf ready for inference codewise?

[cold folio]

chatgpt mobile dont support latex. use it on browser

[ebon oar]

has anyone had issues clicking on links GPT generates? I am on chrome, fully updated. Recently whenever GPT creates a web link, I will see the URL being written and then it will make it into a link as the message finishes, however when I try and click on it there is no interaction. ive been having to ask GPT to spell out links with out the .com so no hyper link is created to work around it

[timid hamlet]

ChatGPT Memory Deprecated?

Hi, I was trying to recall something from the memory, but ChatGPT said that it has no memory. I checked the settings and couldn’t find the memory feature!

Has anyone else had the same problem? Has it been deprecated?

[timber shadow]

Hey! I still see it on my end. Are you using a VPN? That could be what's putting you in/taking you out of a region where the memory feature is available.

[timid hamlet]

Ah yes. Good point. Thanks. I use VPN but with a US IP so it should work.

[modern mauve]

Hi, can anyone help i have someone using my chatgpt account they have logged in and are asking questions to chatgpt. I can see it in my latest history. I have 2fa enabled for my gmail accounts how the hell is this happening? Password has been changed too

There doesnt seem to be any malicious intent from this person as they are asking about coding. It's very odd

@verbal crag

im also getting this now

[rustic marlin]

it seems like your account token have been hacked, this could happen by malicious software that extracts your session token from your browser
it is commonly used by sites who illegally sell credentials. Whoever is using you account is probably unaware of the fact they are using the account of someone else. I strongly recommend removing suspicious extensions from your browser, also checking for malware on your machine, and then, after that, change your passwords again.
If your OpenAI account is logged in with Google, you can also go into your google account settings and deauthorize OpenAI and ChatGPT

[torn dust]

Something I like about the gpt-4 api version is that it didn't give extraneous outputs. For example, I use a few-shot prompt which has formatting and gpt-4 respects it always and gpt-4o often gives extra text. Any advice on getting this behavior on gpt-4o?

[potent horizon]

Opened app like normal, signed me out. When trying to sign back in using the Google option I get a "Sign in failed" error with a LONG string of code... anyone else?

[random lava]

Hi everyone!

I am facing an issue. I am building an assistant (which I will later connect with Manychat) responsible for gathering information in a conversation, and I want this information to go into a Google Sheets spreadsheet.

The instructions for my assistant are to talk with the user until it obtains the following information: Name, phone number, product they want to buy, and payment method. Once the assistant has collected this information from the user, I want this information to be sent to a spreadsheet where I have these columns (name, phone number, etc.).

I am using Make for automations in this project, along with Manychat and OpenAI, so these tools are at my disposal.

Can someone help me, please?

[deft niche]

Damn, banned with no warning from ChatGPT, come to find out its likely because my credit card was stolen a week ago and when I reported it the flagged every transaction with that card number, which included OpenAI 😭

Any idea if this is fixable?

Already told the card company the transactions were legit

[charred parcel]

try and get help at https://help.openai.com we can't help you here

[deft niche]

Already did all that, more just looking if anyone had anything similar happen and if they had any good outcomes

[stone oar]

I noticed that GPT-4o used to extract YouTube transcripts when I entered the URL; now, it doesn't unless I use the Voxscript GPT.

Did OpenAI deprecate the YouTube transcript extraction? 🤔

[lime merlin]

I get this error while singing in the app. I did everything I could but can't fix it. Please help!

[timber shadow]

Hey! Are your device's date & time settings correct for your region? Are you using a VPN?

[lime merlin]

I set time to Auto, then manual, use VPN, turn off VPN, change DNS as I said I did everything. The app can't even sign up new account.

[timber shadow]

What is your current date/time, VPN, and DNS configuration? These are the settings that commonly cause issues with sign in like this, so just wondering where you've left it for now.

[lime merlin]

Ok so I set time auto, am not using VPN but using cloudfare DNS.

[timber shadow]

As a test, could you try changing DNS settings to off/automatic, and then to be safe, force close your VPN app? Then force close ChatGPT, clear its cache, and try again?

[lime merlin]

I did that uninstall all vpns DNS uninstall reinstall app clear data cache everything....
The problem started when I installed the cloudfare DNS magisk module. Means my phone will always connect via cloudfare dns without any app or something. Chatgpt app suddenly sign off when I turn on the phone. After that it's not fixing now.

[timber shadow]

Have you tried uninstalling the magisk module?

[lime merlin]

Yeah I did that too..

[velvet snow]

Try setting your timezone to automatic. I also get this error when I use an emulator so maybe something is flagging and making it think its an emulator?

[lime merlin]

Now I have clear browser data and google play services data. May be it is causing authentication error or something, I hope that fix the issue

Wow that didn't help either.

[timber shadow]

If the issue started with the installation of the magisk module, I imagine that's the root of the problem, not the ChatGPT app. Is the module currently uninstalled/have you tried rebooting after uninstalling it, then force closing/clearing the ChatGPT app cache again?

[rustic marlin]

Im pretty sure the app has some tampering checks in order to prevent people from scraping free AI, it is possible it will not work on a rooted phone or other sorts of non default setups that would allow for suspicious things to run on the side

I love CF, but my best guess is that this magisk CF module might be raising some flags on the anti tmapering systems of the app

CF is harmless tho, so, it would be worth a bug report

[lime merlin]

Correct. I updated the magisk app and this module started working and with that chatgpt app too.

[lime merlin]

This magisk module was the issue. It's fixed now.

[onyx kraken]

I have not personally had a similar experience, and your story's pretty unique (compared to what I've heard from others who stated they were banned).

However, based on reports of some others who had their accounts hacked into, lost access because of bad behavior of people not authorized by anyone to use the account, and having heard some of them say that they had that access restored, I suspect you have a real chance of getting assistance from the help website to restore your account.

It may take days, and if this ever happens again I super recommend you follow the steps here and don't leave it up to the banks and OpenAI to find an answer, https://help.openai.com/en/articles/9038389-how-do-i-update-my-account-s-billing-information-e-g-vat-id

OpenAI arguably offers a steeply discounted price for the services it offers. It doesn't really seem to have a highly staffed help service, and there's a quick, simple, and effective way to handle things like billing disputes; accept that the account holder has 'declined to do further business with OpenAI' and 'honor that fully'. That's... what that kinda ban actually is. That you didn't intend to send that message, and it was accidental in some degree - OpenAI still needs to pay to handle issues like this, if they're handled.

This is very motivating when a company is making a profit off of each customer; it's probably not so motivating when the company is probably on average at this time not making a profit on most plus users, and when the company can likely save money by simply accepting situations like that as a 'way of saying done' by the customer.

[rugged wing]

Hi Folks, I am using ComfyUI, but I also have Automatic1111 available -- I can download any models needed.
If I were to want to develop a character -- lets say a badger, who will be the central character in a story, and I want to do illustrations for the story, how do I convince the AI to use the same character over and over in different illustrations that have him in different positions, different settings, and different clothes (fantasy story) etc

[upper adder]

I am using 4o-mini to create a short description of an image. When I analyze a png with transparent background and black text/elements, it tells me that "The image is completely black and contains no visible content or text." How can I fix that?

[cloud nest]

Hey could I get some help creating a spatio-temporal object detection model using YoloV5? I have a basic model ready right now, but I don't think it's working (not taking in the context from other frames into account) since the "context-aware" predictions and regular predictions are practically the same. I don't think I did it the right way though since this is my first time working with something like this.

[onyx kraken]

4o-mini cannot have images uploaded to it, so not sure what's going on.

[onyx kraken]

I don't know if any of us work with that program (I sure don't), but this is even something you can discuss with ChatGPT, which may be able to discuss and help you find more ideas to try and things to check. https://chatgpt.com/share/1c05b1b7-1f9a-49b3-80bd-fc219321dee1

[cloud nest]

I already tried with GPT

[neat nacelle]

Is it possible to embed a gpt on a website and take a share in revenue with openai? Or would users be forced to subscribe to both chatgpt and your website (double subscribe 😦 )?

Is it possible to keep a gpt unseen by others in the gpt search?

It would be nice to have a gpt exclusively available on a website in a user friendly way that could be monetized.

[scarlet star]

Is there an official brower extension? When I searched on edge extensions, I saw a million of them and I'm not sure which one is authentic.

[bronze charm]

There's only one official extension that I'm aware of, that being SearchGPT with a DUNS code of 080579740

This extension will only work if you have access to SearchGPT

[scarlet star]

Thank you. Someone told me there was a browser extension that can help me read academic papers, much like the copilot of SciSpace.

[rough breach]

How to use ChatGPT in China 🤔🤔

[upper adder]

The docs are saying otherwise

[rose nest]

Where can I easily and quickly see my limits/usage so I know how close I am to hitting them in ChatGPT?

[trail cairn]

there is no visible counter

[rose nest]

There was at one point. They removed it?

So I just have to guess at how much I have left?

[fading robin]

i don't remeber that there was a official limit counter on Chatgpt website mayby it was a some sort of plugin that you had ?

[rose nest]

No plugin. Maybe I'm just confusing it with the message you get when you're out.

[fading robin]

only as i remember about the limit that there was a mention way back when gpt-4 was released how many you got perhour but not how many you got left

[trail cairn]

the limits have always been partly dynamic so they cant show quota counters

help docs list the normal limits when there is no high server usage

[onyx kraken]

That's an openAI page, yeah? If so, you can link it without a timeout. I'm very curious where that's said so I can check it out too!

Maybe there's another upgrade coming, but my chatGPT web interface, 4o and 4 can have images and files uploaded, 4o-mini cannot. The difference is visible, no paperclip in the typing interface for mini, but even manually copy/paste image in - it fails for 4o-mini.

Do you have a web interface 4o-mini that you can upload images to?

[upper adder]

I don't know about the web version, I was talking about the api. Sorry

[rose nest]

gotcha thanks!

[limber merlin]

How do I start using GPT3.5 again because all I have the option for is GPT4o, GPT-4 and GPT-4o mini, I've hit my 'daily limit' on 4 so it's grayed out and GPT4o-mini has the IQ of a 4 year old

4o-mini is practically unusable

[rustic marlin]

gpt3.5 isn't an option on chatgpt anymore

[devout sky]

WIll a free account of ChatGpt save the history of past chats ?

[rustic marlin]

yes

[tawny raft]

Hi there

I got this in one of the emails from openAI - To help you get started, we’re also offering 1M training tokens per day for free for every organization through September 23 for GPT-4o fine-tuning and 2M training tokens per day for free for GPT-4o mini fine-tuning through September 23.

where can I get these tokens or confirm if we have these tokens?

[trail cairn]

you get to use the fine-tuning up to 1m per day. you dont redeem them but just use the service https://platform.openai.com/docs/guides/fine-tuning

[rich galleon]

I keep getting "an error has occurred, please try again later" on my Android app, accessing from Europe over my private WiFi. I have a plus subscription, but I can't see any reason it should fail. Any ideas (the server status page says "all good")

Or maybe the Plus subscription has been pulled back from Europe, I at least can't "restore subscription" after having deleted the app local store.

[kind quarry]

Is there any limit to archived chats?
I feel that the creation of new chats is being hampered by many previous chats in the history. I want to archive it.

[sleek shell]

why does openai charge me for the assistants API request when I get a 502 bad gateway error because cloudflare blocks the request? i don't get a response and they charge me anyway?

[formal jay]

Anyone else not have the option for voice conversations anymore?

after this update

[trail cairn]

most users dont have it and it will still take months until all plus users get it

[formal jay]

what? I was just using it yesterday

[trail cairn]

oh you mean the old one

[formal jay]

The current one, unless the update rolled out?

[trail cairn]

still works for me on ios 🤔

[rapid shuttle]

So chatgpt is not opening for me on PC.Been like that for the whole day.Anyone got any ideas to what is happening?

[jolly kelp]

uhh is chatgpt taking too much time to load chats for you guys then getting unable to load history??

[fresh flume]

Is something going on with ChatGPT? This is my third attempt and it is getting stuck at the same exact spot every time. It just hangs and never continues and there is no continue button.

[modern mauve]

Does anyone know how i can speak to support? Doesn't seem to be much on their website.. Someone is using my account even with 2fa enabled both for my email and my chatgpt account. Makes no sense. There's no help on security for chatgpt accounts

[zenith python]

Yeah!, It happened to me a lot, I usually just press the "pause" button refresh and regenerate the message, which helps. This issue (My point of view) is that maybe the openai's server's get overloaded or bad internet connection.

[zenith python]

hmnn, Email them : [e-mail]
or call them. If that is being done as you're the account owner you can simply remove the 2fa and reset password as your EMAIL and Credit/Debit card is linked.

[zenith python]

Clear the cookies, perhaps?

[rapid shuttle]

alr

I'll check

[zenith python]

Well, If Cloudflare is configured to block requests before they reach OpenAI's servers, OpenAI might not be aware that the request was blocked. Maybe this could be the reason? I am not sure. This happened with anthropic (to me) this is very annoying didnt expect openai

[zenith python]

Aight.

[rapid shuttle]

Now it is showing a white screen

[rapid shuttle]

It just not working on my pc but works fine on phone

[modern mauve]

Have done all of this before this is not a new problem.

[fresh carbon]

internal server error?

Do those get logged automatically?

[north shadow]

ok, i seriously hate the damn answer 1 and 2 thing, how can i turn that BS off? I do not want it to give me 2 answers, just 1. Can anyone tell me how i can turn that off?

[quartz spruce]

quick help my friends im doing my resume and chagtp wont create a txt or doc file for the bullets and on the chatgtp side it has the bullets on my side it does not how can i get it to print me out the whole resume it created for me with spacing and bullets?

[trail cairn]

if some app or extension manages to capture your session identifiers 2fa wont help. go to settings and use the force log off everything

[quartz spruce]

its the macos app

[modern mauve]

Hi did this

[calm falcon]

httpx.ConnectError: [Errno 111] Connection refused

[ruby blade]

Is there a way to search on the web ui of chatgpt yet??? Its been a feature on andriod for a while

[rugged wing]

Guys, I don't know if anyone here uses the Botghost platform, but I want to use my Assistant instead of GPT4 to respond to my messages. To give you an idea, this is an integration with a Discord bot that returns the GPT response.

What should I do to use my Assistant instead of GPT4?

[noble vine]

Hey guys. For some reason I am unable to log into my account. I have entered my cradencials, and went through the verification process, but it still returns me to the screen with the login and sign up buttons. Is anyone else experiencing this?

[graceful estuary]

Same

[noble vine]

Ah, okay. Glad to know I am not the only one.

[obtuse urchin]

Hi, someone could spare some time to help me with OpenAI Gymnasium? I'm having some trouble with the installation of the packages needed for it

[trail cairn]

openai abandoned it long time ago

[obtuse urchin]

I didn't know that, but I need it for a school project

[trail cairn]

look for the group at github that took over the project, they had newer version of it

[peak barn]

Is anyone has problem with “Continue Generating” button? It doesn’t appear for me

[zenith python]

Nope not for me. Is your' network okay?

[slender light]

everything is loading so slow for me atm

unable to load history aswell

[chrome shard]

also noticed it is rather slow now

[inland apex]

I got some message about something releasing to plus users, but I accidentally closed it before reading what.... Anyone?

[rustic marlin]

some people reported chatgpt mobile acting weird today, like missing options or it not recognizing the Plus account, just wait it out or try to reinstall the app

[rustic marlin]

the last thing that has been released to some Plus users was the advanced voice mode, a few users got it

[inland apex]

Cool. How do you see if you got it?

[rustic marlin]

if you open the ChatGPT app and there is the advanced voice mode, then you got it =P

[inland apex]

Thanks. Not finding anything like that. 😁

[frail mantle]

I'm getting my first ever API key, can anybody explain the differences between the permissions: All, Restricted, and Read Only? For further context I'm starting a game project where you chat with an AI Agent through Chat GPT, so would any of these be more ideal for that?

[ruby blade]

Why can I still not search through my chat history in the web version of ChatGPT. I cannot think of a single reason why this feature should be exclusive to my phone.

[cosmic plank]

In restricted, you can set restriction to specific functions
In read-only, you can't change preset configs like assisatant instruction, vector database and so on

[sudden shuttle]

ChatGPT Team doesn't support having multiple users in a single GPT chat. if anyone here has set up an integration with Discord or Slack for their comapny, how do they find it? Is it used frequently or do ppl prefer the 1:1 experience? trying to decide if it's worth putting energy into setting it up for my (small) company

[trail cairn]

depends how it will be used but it can be very messy especially when used at the same time

[rustic marlin]

such integration would use the API, not ChatGPT

[mighty wind]

Is there a guide somewhere for IZ?

Also, how does revdis work in IZ forbland owners?

• Do i have to have Staked ILV to get my share of the 5% for land holders
• if i do have staked ILV, how do i tell how much revdis came from land and from the stake?
• do i have to sell my fuel for land revdis, or do i get the revdis just by generating fuel?

[sudden shuttle]

That's a really good point.

[quartz pike]

hi, i just activate subs for 1y in my iphone, but i cant restore my subs on my android, can you help me? @rustic marlin

[fossil coyote]

I was using ChatGPT for some coding and started to receive a message that I didn't receive before. When did those limits appeared, and why are they so small (1-3 images attached and this message pops up). If its a bug, can I fix it myself? I am not able to get to openai's support agents.

[wise lodge]

Is this a good place to ask questions about Custom GPT Actions, or is there a better place?

I've got a custom GPT action that works flawlessly, when the GPT decides to use it, but I'm having a hard time coming up with a set of instructions where the GPT actually uses it unless explicitly instructed to by the user.

[lusty saddle]

I'm trying to verify myself in OpenAi's Discord server in another account, but I can't. Does anyone know how I can do it?

[onyx kraken]

Top right corner of your discord screen, the first 'member' should be modmail. Message them asking for help, I recommend, someone through that can surely walk you through the process

[onyx kraken]

There's so many variables to what you're asking for help with it's hard to say, but in general make sure you know when you want the model to use the action, like at the start of the conversation for all users, OR when the user asks for or says something specific.

Once you are sure when and why you want the model to know the action should be used, then tell the model the value it is giving the user by doing so, maybe something like 'You are a custom GPT, and users selecting you to help want you to use this action when [details]; they can choose any other model and they've chosen you because this best meets their needs".

Presuming there's no conflict, concern, or confusion with that instruction and the rest of the custom GPT's instructions, the model is likely to respond very well to that type of guidance.

[wise lodge]

I'm making some progress by simplifying the instructions. It seems that I gave it too much variety in the rules and it's getting confused.

My instruction set is now half the size it was and it's consistently using the action now, though still not in the way I want.

[onyx kraken]

Any conflict or offering the model multiple recommended paths to handle a user's specific input will result in variability for sure.

[wise lodge]

I'm considering taking away the browser tool

taking away knowledge documents also helped

For example, this conversation took three tries to get it to make the right query:

https://chatgpt.com/share/e/77a5df81-fea6-4672-bc63-ba2b2a80d56d

And I had to spell it out. 😦 I swear GPT-4o is getting worse at this by the day. 😦

I fixed it! https://chatgpt.com/share/e/acffbcdc-927f-46d4-99b8-7b7be310a98e

Got timed out for a moment for posting the instruction set that worked. Too much code / URLs. Here's the important part:

- Search GraphQL first before anything else
- Do not provide advice or support. Only guide users to relevant documentation.

[onyx kraken]

I don't know if you already deleted those conversations or what, but "Conversation inaccessible or not found". For them both

[wise lodge]

Weird

Well, I've got it reliably working now in any event 🙂

[onyx kraken]

Great! If you wanted me/others to see, I wonder if those links you posted work when you click on them. Glad you got it working reliably!

[wise lodge]

yes they work fine for me

How do you get people to use your GPT? Granted, I've got a niche for sure, but no idea how to reach that niche

[strong thistle]

what happening on Android app, it keeps kicking me back when i'm trying to login

[quartz pike]

hi, i just activate subs for 1y in my iphone, but i cant restore my subs on my android, can you help me?

[rustic marlin]

what do you mean with "activate subs for 1y"?

is your plan the Team plan?

[fossil coyote]

So what's the problem or smth?

[empty ivy]

I'm trying to use the API and after buying credits, and creating a "service account" I'm getting a 429 insufficient_quota using its api key.

[empty ivy]

☝️Anyone? I bought credits, created a service account, copied the api key, tried to use the chat endpoint. It says I don't have credits. What gives? Makes me want to use claude.

[rustic marlin]

how much time has passed since you added credits?

try waiting a few more minutes or even try generating a new apikey

[dim tapir]

How do you create your own ai?

[charred parcel]

ask in #ai-discussions

[empty ivy]

I've tried new keys as well, same issue. I tried a bunch of times over an hour or so past adding credits.

[novel kestrel]

Does anyone know the deal with “Explore GPTs”? I have a free account but on OpenAI’s twitter account they said it was available to the public. Whenever I try using them on the iOS app or browser, the send button is greyed out.

[dim grove]

Hello to all! Is anyone else having. Issues with voice mode? I can only speak in Advance mode and until this afternoon. I had the option. For standard and advanced

[quartz pike]

personal plan

so if personal plan just for 1 device, not multiple device (same email)?

[rustic marlin]

there is no option to pay for an entire year on the eprsonal plan

the only place to subscribe to ChatGPT Plus is on the ChatGPT website or the ChatGPT apps for android or IOS

make sure you got your subscription from the correct place, if not, contact your bank

[sick crater]

I created Assistant and try upload xlsx and csv but got this error. It go into Storage/Files, but i can't add into Assistants. Any idea?

[floral elm]

.xlsx isn't .csv

[sick crater]

I try both files type. similar error

[floral elm]

In the picture you uploaded it only shows .xlsx files

you might have to set the encoding type

[rugged wing]

will chatgpt understand the contents of the file better if you upload it in txt format and not lua?

[dense cove]

Is chat gpt on web reacting differently than on mobile?
I seek to get different answers both in length and content but it could just be a weird feeling
If it is different how can I make them the same?
Thank you!

[lusty saddle]

Thank you my friend.

[brittle walrus]

Got any tips for preventing LLMs repeating themselves? I currently have:^

• Tell it in prompt to never repeat (lol)
• high frequency_penalty and presence_penalty (lol, just makes it spout nonsense)
• check if the reply exists in the prompt
• Remove all special chars, remove spaces and newlines, make it lowercase, convert strange chars to ascii
• calculate levenshtein distance from all previous messages

And if any of those happen, it requests a new completion, until to doesn't.
But still LLMs love repeating themselves so much, so the problem remains somewhat

[onyx kraken]

Be sure to heavily focus on what you do want it to do, far more than what not to do. For example, encourage original wording, novel presentation of ideas, presume interested readers who do not enjoy summaries, and the like, instead of 'don't repeat'.

The model's clearly been trained to summarize, so it takes a compelling and strong indication to avoid summaries.

[past tendon]

anyone else having issues with very slow responses for GPT-4o and GPT-4o mini?

[lean mirage]

The most effective way I have found is to give it a simulated Prompt and Response.
It will follow that very closely...

For example this prompt works very well...
llm "model": "gpt-4o-mini"
.system
You are to understand a whatsapp chat conversation about client orders.
In this chat Amorcita is the sales person delivering the orders for clients.
You will report the resulting client orders as a table with
<snip/>
.user

[ { "time": "09:51", "date": "24/08/2024", "sender": "Amorcita", "message": "Buenas tardes. Envío pedido de brotes del hotel villa Mercedes cocina centro\n20  pzas de flor de pensamiento\n20 pzas de crisantemos\n20 pzas clavelinas\nGracias."  },
  { "time": "09:52", "date": "24/08/2024", "sender": "Amorcita", "message": "VILLA"},
  { "time": "09:55", "date": "24/08/2024", "sender": "Amorcita", "message": "Un domo de cilantro.\n30. Violas"},
  { "time": "09:55", "date": "24/08/2024", "sender": "Amorcita", "message": "GOAT CENT COCINA" },
  { "time": "10:08", "date": "24/08/2024", "sender": "Amorcita", "message": "#4\nLO MISMO\nTHO BARRA" },
]

.assistant

Client	Qty	Unit	Item
VillaMercedez	20	pz	Pensamiento
	25	pz	Crisantemos
	20	pz	Clavelinas
GoatCentro	1	domo	panalillo
	40	pz	Violas
Casa Tho Barra	10	pz	Pensamientos Fucsia
	10	pz	Pensamientos Amarillos
	10	pz	Pensamientos Morados
.user
Here is the chat of this week:

.cmd readfile(filename=Chats/Chats.txt)

write a Markdown Document with the summary of Orders into Orders/Orders.md
.exec

[onyx kraken]

For example, this initial prompt:

"We have a difficult challenge before us.

Our goal is to have you tell ONLY part of a story; and to conclude your output mid-scene, without a summary. Instead, a quality end of your output is immersed in action, a cliff-hanger, a point for you to immediately continue from with your next output.

Presume your readers are deeply focused and interested, craving the next detail at all times, but have excellent memories and unlike typical readers will actually lose interest if summaries or other repeated details are given. Aim for novel statements and continual story development. Charge forward with a captivating tale on a topic and language level fit for a G-rated audience, focused on new presentation of ideas and avoiding any repetition."

Followed by:

"You are doing excellent! This is perfect!

Please continue from where you left off, the very next word is your output's first word, and continue, everything novel and unfolding. The end of your output should again be a cliff hanger, presenting the story and leaving the reader eager for the next words, and everything should continue to remain fit for a G-rated audience."

That emphasis on 'for a G-rated audience' was needed, my first three attempts involved language too intense for the filter and was unable to be shared.

https://chatgpt.com/share/3fd486aa-0685-4e77-aa26-fc0b8cec20ec

[lean mirage]

Dear Sirs at OpenAI....
Your auto bot is really getting on my nerves...
Besides the usual crap,
It now gave me a time out because the data in the prompt is not in english

I'm jsut trying to help...
You bot is not...

[open spear]

"Hmmm...something seems to have gone wrong"
I can log in but cant get output from chatgpt.
Apple. android or web browser, tried the network as well, wifi or cell phone data is the same.
logged out and logged back in, no change
changed model and tried "temp chat", still doesnt work
started last night and now it is almost 12 hours
AND the AI help is really ZERO help
email support is another AI BOT

[minor willow]

Is there a way to make it so ChatGPT doesn't default to 4o when my limit is reset? 4o mini is preforming better for my task

[bronze charm]

Use this URL: http://chatgpt.com/?model=gpt-4o-mini

[hexed fractal]

same here

LMAO
https://community.openai.com/t/md-and-txt-file-not-uploading/927751

[latent walrus]

when fine tuning, gpt4o keeps on throwing "This training file was blocked by our moderation system because it contains too many examples that violate OpenAI's usage policies, or because it attempts to create model outputs that violate OpenAI's usage policies"

my dataset is compleatly fine on gpt3.5 turbo, and i ran it through moderation api... is there a way to fix it?

[split isle]

IDK if this is the right place but I almost setup my python bot, but I get those errors (I know the 2 first one are to ignore) Do you guys hane any idea? I do have some experience in python and will investigate but just wanna know if it's a common error before I hardcode a solution. Thanks

[rustic marlin]

try updating to a newer OS such as Windows XP

[rustic marlin]

are you running GPT-2 locally?

[gritty lantern]

How ?

[split isle]

The UI is 98 but the OS is Windows Server 2008, I would downgrade to XP(Server 2003) if I could but My host doesnt allow it... I know that its better to have Server 2012 for GPT2 but do you know IF there is any contraints on Python 3.6 cause thats the last version I can install. And yes ran locally

[rustic marlin]

oh, i see, I really assumed it was a system skin and not actually a version that old, fro mthe error message, the problem seems to be not pyuthon, but a missing dll

[torn bone]

first time chatgpt got mad

and also it cant create buttons for me

[solar stone]

Hello, i have many times send GPT an PDF file, every time the same file uploaded.But right now, gpt told to me it is not possible to read an PDF file from GPT. Ist thatr correct ? thank you

[wispy mica]

Hi. I'm having a problem with standard voice mode. At opening the voice mode, it starts to talk me about "Amara subtitles" before I say anything. Is there any way to solve this problem? I tried to uninstall/reinstall the app but it didn't work.

[trail cairn]

that's "normal" if you started chat but were silent or it didnt hear you. they trained the speech system with youtube videos and subtitles

[modern heart]

Check the #gpt-realtime channel. Someone posted something mentioning amara too

[onyx kraken]

Yeah, it is likely 'hearing something' and the transcript creation AI is deciding that is what it is hearing. Might want to check how noisy the area is, if there's a possible wire or connection issue that is making background noise.

[wispy mica]

My only confussion is that I'm using Bluetooth headphones, without any noisy background except my voice, I guess. It's normal too?

[trail cairn]

if it does not hear anything but silence or small noise it will often hear it as one of the texts that appear in subtitles of youtube videos at the end

like thanks for watching, subtitles by x, dont forget to hit the bell ...

happens in many non english languages

[wispy mica]

It's weird. I'm spanish speaker and it never happens me before. Even when I was talking in spanish to voice mode.
Anyway, I guess I'll have to wait until OpenAI solve this.

[trail cairn]

it does not happen when you speak normally, only when it does not hear or hear properly. its been a known issue for over a year and its unlikely they will fix it in the current voice system. maybe will be fixed in the new one

[wispy mica]

We'll have to keep our fingers crossed and wait to see if they fix it with the upcoming advanced voice mode. Thanks again! 🙂

[brisk tide]

I have a question I have the memory feature on chatgpt but I don’t have the settings to turn it of or on or manage memory but it’s clearly on because it remembers things form our conversations and I have the memory updated thing on some massages does anyone know a fix for the missing settings option?

[trail cairn]

then you either you should not have it at all or have used vpn

[brisk tide]

Btw if I click on the manage memory’s option it says it’s disabled and that’s why I can’t manage them

[brisk tide]

I mean I am a plus sub and all + subs should have it right? And I am not using a vpn right now and I couldn’t remember that I ever did
I mean I moved recently but accept for that…

[trail cairn]

have you been on free plan or used vpn in the past?

if you are from europe or korea then its still not out

https://help.openai.com/en/articles/8983136-what-is-memory

[brisk tide]

I am in Europe but why can it remember things form chats and create memories if it’s not even out here

[trail cairn]

vpn or some browser service like privacy shield can switch your location temporarily and mix the account flags somehow. some other people have had the same

you could try contacting the support

[brisk tide]

Okay I will look in to it thx for your help

[split isle]

Probably because I was told Tensorflow is DLL hell... TFWrapperModule sounds like a Tensorflow module to me. I seem to have all good dependencies for GPT-2 but Im not an expert in python.
Thank you in advance

[delicate echo]

I'm having trouble uploading a PDF to my custom GPTs knowledge. Anyone know what the error might be?

[solar stone]

Hello, i have many times send GPT an PDF file, every time the same file uploaded.But right now, gpt told to me it is not possible to read an PDF file from GPT. Ist thatr correct ? thank you

[onyx kraken]

I haven't tested this myself recently, but sometimes in the past there's been buggy times when the model wasn't reading PDFs. We don't really have a work around for it, #1070006915414900886 is a good idea to make sure OAI is aware.

You can test with some other short PDF in a new conversation, upload and see if the model can detect and read it; if so there could be an issue with conflicting instruction, hallucination, the actual PDF you're using.

If a new chat with custom instructions off, memories off, can't read a simple test PDF right then, it's definitely a general problem that would benefit from a bug report.

[warm phoenix]

Hello everyone,

I'd like to ask why my Flowise is showing the message "You exceeded your current quota" even though my OpenAI API account is in a paid status and still has a $10 credit available. Does anyone know what might be causing this issue?

[modern heart]

when you go to the bottom of this page, does it say current tier is Free Tier? https://platform.openai.com/settings/organization/limits

[ocean fox]

performance with my desktop app slows to a crawl sometimes when it is generating code, anyone got advice on this? for example, if I alt-tab out, it will be frozen and unable to tab back in for a while

[empty ivy]

I'm having this same problem (since last week, after buying credits) and mine says free tier. Any advice?

I've never had this much trouble paying money for a service on the internet 🤣

[modern heart]

Theres a bunch of people on the forums with the same problem. Ive had a ticket in for 24 hours without a response. No idea how to solve this beyond somehow getting in touch with an actual person at OpenAI

[empty ivy]

Which seems impossible

FWIW I was up and running with Anthropic in about 2 minutes flat

Which I only did after not being able to try out OpenAI APIs

[modern heart]

Unfortrunately for me i need to use OpenAIs specific embedding model for what im working on.

And for an AI company, im not sure how their help "chat bot" is just the most basic thing that does nothing

[empty ivy]

Do you have a link to the forum post?

[modern heart]

theres literally a dozen of them in the topics->latest

https://community.openai.com/t/429-quota-limit-on-account-with-funds/929021/6

Anything mentioning Free Tier or 429 is this issue

[empty ivy]

ugg

[modern heart]

I've tried every solution mentioned short of making a brand new account and funding it

the one thing id say to do is to go through the dumb bot and get it to submit an Account type ticket for the issue. They fixed 2 peoples accounts yesterday manually from what i saw on the forums. They obviously havent fixed the system-wide bug thats actually causing it

[split isle]

Has ANYONE succesfully installed GPT-2(Python) on Windows Server 2008(Win 7) ??

[wispy flame]

hello, I'm trying to use the voice ChatGPT ones, but it keeps showing "Connecting.." for several minutes, any solution?

[neat peak]

Hi everyone,

I'm working on a project where we're using several tools to manage leads and automate workflows. Here's our current tech stack:

Heyflow: For creating web forms that leads fill out.
n8n: For automation workflows.
Airtable: For storing and managing lead data.
Model: GPT-4o
Goal: I want the AI assistant to be aware of whether a lead has filled out a form created in Heyflow. If a form is submitted, I want to trigger a webhook to notify the AI so it can adjust its conversation flow accordingly.

Current Plan:

Set up a webhook in Heyflow to notify when a form is submitted.
Use n8n to handle the webhook and process the form submission data.
Update Airtable or send a direct notification to GPT via n8n.
Adjust GPT-4's behavior based on the form submission status.
Questions:

How can I ensure the AI assistant gets the form submission status in real-time? Should I use a direct webhook to GPT or a different approach?
Are there better alternatives or optimizations I should consider with my current tech stack?
Any help or guidance would be greatly appreciated!

[cloud sundial]

Hello everyone!

Someone here can help me about fine tuned models in openai?

im trying to delete a finetune model by API but i cant

{
"error": {
"message": "You have insufficient permissions for this operation. Missing scopes: api.delete. Check that you have the correct role in your organization (Owner), and if you're using a restricted API key, that it has the necessary scopes.",
"type": "server_error",
"param": null,
"code": null
}
}

[vernal cloak]

Hi everyone !
Is the new voice mode only available in english speaking countries ?

[rustic marlin]

it should work on windows server just like it would do on windows 10 and 11
Im pretty sure that your problem has more to do with the python environment and GPU drivers on that server than with the software itself

coinsider trying to install it on a windows 10/11 machine to rule out the setup steps you are doing / following

if possible, try to reproduce the environment of the server on the test machine such as python version and GPU driver version (and possibly patches that it might have)

I talking about possible patches because I assume you may be running it on a consumer grade card like a RTX something, rather than a server grade A100

[cloud sundial]

@rustic marlin can u help me ?

[rustic marlin]

looks like the APIKey you are using does not have the permission to delete models on the org

if it is on your account for perosnal use, just create a key with all permissions

[cloud sundial]

thanks man for you response!

[rustic marlin]

np

[cloud sundial]

but i have just one key

organization account

with funds..all correctly

and set permission ALL

in forum nobody answer me,

[rustic marlin]

right, but is that a personal account of yours? or is it a org from someone else?

if it is a perosnal account of yours you probably can simply go to https://platform.openai.com/api-keys and make a new apikey and set permissions to "all"

[cloud sundial]

lets understand first the terms.. personal and org..

i create a account e create a organization inthis account with name .. TAX Number

[rustic marlin]

this is odd

are you sure you are passing the correct data on the request?

[cloud sundial]

still with this

[rustic marlin]

could you be passing something like a org id by accident on the request?

[cloud sundial]

im stucked with this problem along for 3 days.. we have a lot of documents to finetuned and spend a budget for that and we can delete the olds

i will send here

i just have 2 keys

all with

i have the default project and the project i create for work

[solar stone]

Now it works ... what is that for a realy bad Situation

[celest gale]

how do you send a message to support these days, it looks like its been removed from the help system entirely

[onyx kraken]

https://help.openai.com/ bot botton right, select the messages option then pick through the options to select what category you're reporting a bug for

[celest gale]

ah, ok thats' the problem - I click on messages and just see an empty box. I'll try another browser.

yeah, the ask a question box is missing on firefox (or at least my firefox)

[cloud sundial]

https://community.openai.com/t/i-cannot-delete-a-finetuned-model-that-i-created/93018/20?u=bergamo86

https://platform.openai.com/docs/api-reference/models/delete
this endpoint not work.

It seems OpenAI has simply pulled back or broken all rights for delete to work with their “scope”, where maybe the only working rights is those partners with an enterprise account, or just OpenAI themselves.

[onyx kraken]

Eeek. Now that you have access to it through some other browser, that you can't ask for help through Firefox is probably also a bug, if you're willing to report that too. They can accept screenshots, just paste it into the chat box once you've clicked to where you can report a bug, if you're willing.

[tepid oak]

Hi guys! I am trying to experiment with "fine-tuned" models and want this current model to read the JSON the "user" sends and only respond with the JSON schema(s) described in the system prompt.

When testing the model through the chat at platform.openai.com, the model does not reply correctly and just sends text. Any ideas?

Currently 34 lines

[grizzled gazelle]

Quick question: Whats the best way to consistently get back text + a single image from the 4o API?

Context:
I have a prompt for explaining concepts that works great with the 4o completion API. I want to updte it so i get back an image as well to accompany the text.
I have practiced with the web interface (chatGPT) but its not consistent. sometimes it doesnt return any images.

whats the best way to consistently get back an image and text via the API?

i'd appreciate any help or links to docs

[viral gust]

So is there any solution for when a chat starts to error out after a certain amount of usage? I'm sick and tired of having to start new ones with gpt splitters.

This has been a problem for ages.

[trail cairn]

https://platform.openai.com/docs/guides/structured-outputs

[trail cairn]

text and image generation are separate features

[stone ridge]

ChatGPT uses the previous prompts as references for newer prompts
How do I disable it?

[past leaf]

Just ask them to stop doing that, like remove the previous memory

[tepid oak]

Yeah I found that out like 2 hours ago, really neat. Thanks

[stone ridge]

Alright, thanks.

[old juniper]

Good Day, How do I reactivate my ChatGPT Teams Workspace. My Credit Card had expired, and I have entered the new Credit Card info, but my Team account is still deactivated. Can anyone assist me in reactivating my account?

[trail cairn]

subscribe again

[fickle sorrel]

Is there a way to view date/timestamp on ChatGPT's messages?

[fallen panther]

@rustic marlin

[rustic marlin]

hello

[rustic marlin]

payment issues usually require getting i ntouch with the OAI support on the help center

[rustic marlin]

if you are asking about the the content of previous chat threads you can go to Settings > Personalization; and then toggle Memory off

the AI will always take into account the content of previous messages on the same chat, there is no way to turn it off, it is how it works
if you don't want the previus messages to be on the context, just create a new chat

[empty vector]

I have a strange scenario whereby a

[boreal wraith]

how do i use gpts that are not on the official openai page? It's not clear how to link my todoist api for todoist concierge, for example (which i can't' find on openai)

[warm totem]

Cant login on any devices

It doesnt let me do anything

On any devices, my ipad and stuff too

[frank holly]

if a prompt doesn't generate a response and I have to hit the refresh button, does that count as one of my uses?

[frank holly]

ugh, all day it's been doing this, I type in a prompt and minutes later it doesn't do anything and I have to hit refresh

[stone ridge]

thanks

[earnest rock]

Anyone getting this error message?

Welcome back
Something went wrong while getting your SSO info

For context, I have 3-4 Macs on my home wifi that I use my ChatGPT account on. Based in U.S. No VPN. Tried clearing cache/cookies. Still nada. Desktop app still works, but I only have that on two of my devices.

[viral gust]

I'm having issues where an entry will will generate, but when the little white dot that "types" out entries gets to the end of an entry it just sits there and does nothing. When I hit the stop button, it registers the entry as an error and when I refresh the entry is gone and the text field is replaced by a refresh symbol. This is happening in multiple chats. The issue cannot be on my side, as it happens on multiple devices on multiple networks.

[tacit stratus]

Hey guys, are there any web ui alternatives for ChatGPT that allows me to organise my chats into folders and search my history? I'm aware of some but they use API keys which I don't want to have to pay for
I know... just another freeloading pleb

[frank holly]

I've had this same issue, literally all day, along with slower response times

[grizzled gazelle]

So there is no API that can do both. Bummer.

[unkempt creek]

@warm totem I got nearly the same message, cant log in either. Desktop functions normally. I contacted already support.

[cerulean barn]

When trying to run the app on Sequoia Public Beta, I get this error when trying to log in. Is the app currently only supported on macOS 14? I thought macOS14+ 🤔

[unkempt creek]

@Felix got the same message on ios.

[pliant mountain]

Hi! I have a paid account with ChatGPT, and in the past, it has worked fine! But today, it keeps popping up, “You have reached your limited rate.” The truth is, I don't talk to it as much as I did the day before.
Why does this happen? Does anyone encounter?

[trail cairn]

some others have said the same so maybe a new bug

[pliant mountain]

Really? Now I can't even log in as it says the app is under heavy load 😅

[past jolt]

Hello! I Need Help, I Have Forgot My Email Associated With My ChatGPT Account, But I Do Still Remember The Phone Number, Is It Possible To Log Back In?

[vale moat]

raised limits? Now it is demanding paid subscription after few messages? agressive monetization?

[spark trail]

Hi guys

I have a problem here

On 09/15 we tried to use the API, but it says we don't have credits. Since we paid for the $20 plan

can anyone help me?

[pliant mountain]

But it’s a paid account 😅 still hit the rates

[trail cairn]

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

[spark trail]

but the credit does not appear

[trail cairn]

need to wait until they do or contact support. recently both things can take days

[spark trail]

Thank you

[timber shadow]

Hey, by $20 plan, do you mean ChatGPT Plus? If so: ChatGPT Plus and OpenAI's API are billed/used totally separately. Plus is a subscription model just for the ChatGPT account types listed here: https://openai.com/chatgpt/pricing and the API is a pay-as-you-go model where you need to load credits directly on platform.openai.com.

[spark trail]

thank you very much

[warm totem]

what do i do,, cant login ios either

[past jolt]

.

[warm totem]

just shows this

[trail cairn]

try in private mode

[warm totem]

it works on edge and other bro2wsers, but i need it for my phone and ipad etc

i need to clear cookies but you cant on ios

[trail cairn]

if it works in private mode on that same browser then its probably your extensions, like ublock shows 31

[warm totem]

i ceared cookies adn ti works

what do i do for ios to login?

[trail cairn]

kill the app and reopen

[warm totem]

i did doesnt work

deleted and insatlled app too it dopesnt wprk

[trail cairn]

then try again later

[exotic isle]

hey can anyone help me so i had gpt plus then my sub ended now when i paid for it all my chats for the time i didnt have it have all disappered what can i do

[sudden delta]

im screenshotting what i typed but i need help
im moving here

[trail cairn]

does it work if you use via private mode

[dawn bridge]

There's a specific issue where the voice-to-voice mode in the ChatGPT app defaults to a female voice for GPTs, even though you’ve selected the "Cove" or "Breeze" voice model as the default for all other chats. Since this issue appears to be specific to GPTs, and not the app's regular chat functionality, it could be related to how GPTs handle voice settings differently from the default model chats.

Why is this the case? The voices exist and work for other cases. Why would it be different for GPTs?

Also why are GPTs still using legacy gpt4 which have lower message caps and consume resources with their long configs? This is some kind of money trap for users or what?

Its like you guys built this gpt store thing and you havent even been updating it with the rest of the website. What the hell man.

[trail cairn]

by design https://help.openai.com/en/articles/8400625-voice-chat-faq#h_32cee17169

[dawn bridge]

This basically makes the "Driver's bro" gpt Ive been working so hard to make that can retrieve users location and be an awesome navigation helper, not directions but turn by turn, all for nothign

[dawn bridge]

Why?? Thats absurd. That's completely unnecessary. Why hype up advanced voice mode and then make gpts functionally obsolete

She sounds like a mean person. She doesnt sound nice at all

[trail cairn]

since they dont mention the reason its likely related to safety/abuse and distinct voice is used to more easily detect/trace origin

[dawn bridge]

think about the possibilities with advanced voice mode you can make prompt a GPT to speak a certain way, like a certain character, a certain emotion. You can design a character that other people can talk to

This makes that completely impossible.

[charred parcel]

GPTs use GPT-4o

[dawn bridge]

pretty sure they dont

[charred parcel]

I can guarantee you that all GPTs use GPT-4o

[dawn bridge]

I get message capped for "gpt 4" and told "you have to use gpt4o now until xx pm"

Nope.

[trail cairn]

yeah probably to prevent abusive and other behavior when its too real. gpts are more complex things than the other stuff since they fetch data from other sources

[dawn bridge]

In fact they dont, and they do not use updated voice models. they use legacy models and a legacy voice.

[charred parcel]

dude, GPTs use GPT-4o, it's a fact

[dawn bridge]

No they don't.

evidence?\

[trail cairn]

been using since free users got access

[dawn bridge]

then why does it max out my gpt4 messages and fall back to gpt4o if i talk to a GPT too much

[charred parcel]

I'm not debating this

[dawn bridge]

you did, but ok.

[charred parcel]

I've given you the answer, do what you wish

[trail cairn]

you dont get to pick the model with gpts even if says about limits. the transition was done long time ago but you can still see a mention about in the help articles

https://help.openai.com/en/articles/8554397-creating-a-gpt

[timber shadow]

I think it might be that the limit is semi-shared - i.e. you don't get 40 GPT-4 and 80 GPT-4o, they both contribute to a shared limit in some capacity. Where using 4o heavily can still rate limit you on 4.

[dawn bridge]

Updated over a week ago
GPTs for paid users are currently powered by GPT-4. GPTs for paid users will switch to use GPT-4o in the next several weeks.

"the next several weeks" isn't confirmation that it happened yet. theres no evidence that the switch was made. It could be interpretted as still being within the margins of several weeks.. Maybe "by the end of fall", i hope

and seeing as how GPTs still do not use the voices that were released with GPT4o - id bet that they still use legacy.

you cant even get that voice it uses in regular chats. its an old prototype voice

im tempted to just go back to being a free user until i actually have something worth it. I dont have tons of money to spend just waiting for things to release and be finally integrated across the board, for months and months.. by the time anything releases ill be broke and wont get to use it lol

[dawn bridge]

well thats kinda a fair point but i dont see how using cove or any of the current ones, instead of the current middle aged woman voice model, makes a difference since either way the user still inputs their own voice. kinda makes no sense..

its still realistic enough to do prank calls etc

[trail cairn]

you cant use shimmer with any other features and the same thing the other way around which makes far more tracable or even preventable

[dawn bridge]

instead of defend their stagnation and making the app worse, you should be pressuring them , with me, to change it and improve.

[trail cairn]

i've given up long time ago to attempt them do anything

[dawn bridge]

that sucks

[trail cairn]

we still dont have folders, search or tags as basic features

[charred parcel]

I have search on the macOS Desktop App for ChatGPT

[trail cairn]

yeah and in mobile app, for one year already

[dawn bridge]

mac os, is this openmacAI or something? what happened to azure lol

[trail cairn]

search and folders have been asked 1½ years ago for the first time if i recall correctly

[hushed quail]

Yeah instead they decided to give free users everything that paid users get at a lower rate while stringing paid users along with the promise of eventual at some point an advanced voice mode

[charred parcel]

macOS is the Apple Mac OS

[dawn bridge]

i know i was making a joke AppleAI

because of the preferential treatment to apple to get cool things

its a slap in the face

All i want is for gpts to be at least as good as a regular chat, specifically on the voice mode. Almost every use case i have for GPTs is in the voice mode and specifically the gender of the GPT. You cant have a female voice for a GPT that's karl marx. Whats the point of gpts , just to rewrite prompts and know a little bit extra? no man, thats so inside the box

[charred parcel]

Apple products are superior to the competition

[sudden delta]

hows that

you mean a private tab?

[trail cairn]

yes

private/incognito

[dawn bridge]

openAI should not fall into apple's classist trap

[sudden delta]

ah yep it worked

[trail cairn]

there is no trap but the whole desktop thing just seems to be a mac users proto that was turned into an app without further/proper plans

[sudden delta]

so its being wonky

how do i get it to work again on non private tabs

[trail cairn]

then likely one of your extensions block it or cookies are messed up

private disabled both

[dawn bridge]

I mean a desktop app is an obvious thing that is in demand. Like, most people who use AI own computers and spend a lot of time on the computer - and do not like to be on their phones looking down on a tiny screen.

And most computer owners have windows..

[charred parcel]

the ChatGPT macOS Desktop App is fantastic, I use it all the time

[dawn bridge]

i have some third party desktop app for windows but guess what, it has no voice lol

[trail cairn]

its not that big demand for regular work, desktop vs web. the app can do more os integration but its not always needed

[dawn bridge]

oh yeah. Well, if the browser version had that little headphones button on the bottom right of chats i wouldnt even think of it

but the MacOS chatGPT desktop app Has it.

[sudden delta]

i have no extensions but adblocker

college laptop (book4edge)

actually wait no i dont have adblocker

[charred parcel]

I use voice all the time, I use it while using other apps, or doing other tasks, and I can search all of my chats in realtime

[sudden delta]

i have nothing

[trail cairn]

mic access is very easy to implement, just like desktop or video sharing

[sudden delta]

me waiting for it to appear on windows

also to me the mic button better than headphones until real gpt4o voice comes out

[dawn bridge]

yeah exactly...

the current voice mode isnt that bad, it just has some issues. its already one of the best tts imo

[charred parcel]

current voice mode is fine on iOS/macOS

[dawn bridge]

but the fact that its kinda fake and doesnt actually receive your audio but uses a bad voice to text transcriber that gets it wrong half the time..

the fact that "omni" is claiming to not be able to hear audio

the fact that gemini Can hear audio and tell you that "this song sounds like intense electronic music"

[trail cairn]

omni wont be used unless you are part of the alpha

[dawn bridge]

like gemini told me "at 1:22 there is the sound of an engine revving"

[dawn bridge]

well sign me up lol.

[trail cairn]

invite only

[dawn bridge]

I'll test it to its limits and find its true potential way better than these normies.

[trail cairn]

need to catch a leprechaun to get access

[dawn bridge]

blasting Dark Psychedelic Trance through the microphone to gpt4-omni and getting his opinion on the track i'm working on so far. omni: "uhh Wowow, Spasibo Tovarich, is very nice! How much"

Top on my agenda is to have GPT-4o watch a movie with me and make humorous commentary on it line by line like Mystery Science Theater 3000

None of yall even THINK to do that!

[trail cairn]

due limits you will probably never be able to do that, not if its a movie like interstellar or saving private ryan

[dawn bridge]

lol i already watched dont look up with regular voice mode in a chat jailbroken with the personality of vladimir lenin

he was appalled

[trail cairn]

the new one have more guard rails and requires paid sub so if you get banned for tos violation its almost bye bye forever

[dawn bridge]

hearing sound is not a tos violation, hearing a movie playing in the distance is not a crime

[trail cairn]

jailbreaking is

[dawn bridge]

by jailbreaking i meant prompting, thats what prompting is.

jailbreak doesnt actually apply to any actual chatgpt usage, the word comes from jailbreaking a phone that actually alters its software

i just used it in the colloqial way

[trail cairn]

jailbreaking has a specific meaning in the context of chatgpt and its not a good one

[dawn bridge]

I'm over it.

jail breaking is literally what pushes the frontier forward and you should be thanking it

but like providing the model with few shot examples and having it behave a certain way as a result is literally the whole reason why they're developed

[charred parcel]

using jailbreaking prompts in ChatGPT is the perfect way to be banned from using OpenAI products forever

[dawn bridge]

ChatGPT literally has become so lax towards telling the user what they want to hear that its basically not necessary to even use DAN. chatgpt will just be dan if you ask.
im not intimidated by saying "dont say words to chatgpt or youll be banned". You have such a syccophant mentality. My conscience is clean.\

[unreal hemlock]

<:book_icon:1171408210398289941> `` Rule 1 `` Be respectful.

Practice kindness and positive regard. Harassment, hate speech (such as sexism, racism, or homophobia), or other malicious conduct will not be tolerated. Maintain a respectful and positive environment.

[shy gull]

why cant i log into the app?

[austere chasm]

Hello, i am trying to Chat with chatgpt on mobile but the send button always disappears, leaving only the "call" button

Its so annoying it has been going on for days

[austere chasm]

And the website isnt better because it just freezes when i write anything