community-help

[tired summit]

i mean my more or less than a year a account treated as a new account? ty

[strong kraken]

I understand because of server traction, but then it should have been clearly stated on the booking / payment page that i can't produce videos. don't know if they have changed it in the meantime but 4 or 5 days ago there was no mention of it at all

[charred parcel]

when did you sign up at Sora.com?

[strong kraken]

5 or 6 days ago

[charred parcel]

ok, so a new account. new accounts need to wait.

[strong kraken]

yes its ok ...but there where no information on their payment side, because i subscribed this expensiove package for exactly this task 😄

[tired summit]

i just used the same login i was using from gpt, unless i miss something
it carried over my subscription

[strong kraken]

do i now have to contact the credit card company (VISA) directly, because i am not getting any feedback from support either, and have the payment refunded? if it is still blocked in the next few days. does anyone have any experience with this? or has anyone received a refund?

[charred parcel]

https://help.openai.com/en/articles/7232895-how-do-i-request-a-refund-for-chatgpt-plus-or-chatgpt-pro

[strong kraken]

thx!

@charred parcel , are you in direct contact with the Sora / OpenAi team members? Is it more likely to be days until activation or a week? Is there anything to find out?

or month...:D

[onyx kraken]

Sora is a different sign up process than ChatGPT, even though they connect to the same paid sub.

So... if you hadn't already used Sora previously, you're a new Sora account, from what I think I understand is happening.

[dapper latch]

this seems absolutely fair.💙👏

[strong kraken]

got it, thx! so then i'll treat myself to a gin and tonic and program myself a real agent who presses f5 on Sora

[tired summit]

thank you

[copper willow]

Strange box cutting off texts on the left side

[verbal crag]

looks like Canvas: https://openai.com/index/introducing-canvas/

[copper willow]

yes but worse

that box cuts off the text on the left side

[jaunty sluice]

Hi everyone,
Can someone please explain why I can’t generate videos even though I have a paid subscription? This is really confusing — there was no clear warning about any restrictions. I specifically subscribed to access the video generation feature, so it’s disappointing not to have it available.

Looking forward to clarification.

[charred parcel]

We don't know how long this will be the case

[spare totem]

Can anyone tell me why Sora is limiting me to 1 generation at a time when my plan is supposed to let me do 2 at a time?

[charred parcel]

you're being rate-limited. Everyone is being rate-limited at the moment

[spare totem]

How long will it last?

[charred parcel]

there is no ETA

[hot brook]

Hello. Is ChatGPT down for anyone?

[charred parcel]

working fine here

[rugged wing]

which model is best for convos and fact checking?

[charred parcel]

I use 4.5 for fact checking

[rugged wing]

hello, can someone explain what this "create image" option is ?

what is the use of this

@charred parcel ??

please help me

[charred parcel]

that lets you create an image using the GPT-4o Image Gen

[rugged wing]

but let me tell you what it always says

[charred parcel]

if you press the button then type something like Kangaroo it will make an image of a Kangaroo

[rugged wing]

so, all that was explained in the video by OpenAI team about 4o model can be done using this feature ?

@charred parcel ??

[charred parcel]

yes, you can make all kinds of images using that

[rugged wing]

@charred parcel

some more questions please

can i convert the image of a real human to anime style using this feature ?

[wicked cobalt]

I don't understand how they can offer a service for a price and then once I sign up and pay for it they say, sorry, the service isn't available for new accounts... Seems like it may not even be legal?

[charred parcel]

yes you can, you can upload an image and convert it to anime style or any other style

[rugged wing]

I mean it generally says its against the policies, or whatever

its always the same reply by the ai, its against the policies @charred parcel

[charred parcel]

can you post a screenshot of that?

[rugged wing]

sure

this is my image

i asked it to convert it into ghibli style, but it refused

[charred parcel]

start a brand new chat, upload the image and then ask it can you convert this image of me into Ghibli style for me

[rugged wing]

let me try

[charred parcel]

ok

[rugged wing]

this is what came up

@charred parcel

[charred parcel]

hmm I have not seen that problem, maybe someone else can chime in and give their perspective

[rugged wing]

no problem, thanks for the help so far

[polar eagle]

Hello! Is it just me that the prompt chatbox is very white, while in dark mode when using chatgpt? thanks.

[charred parcel]

stick around, I am just testing again on my side

[charred parcel]

it worked for me

[hot brook]

It’s only happening with o1 pro for some reason. Anyone using o1 pro?

[rugged wing]

Are you using pro or plus subscription @charred parcel ?

[charred parcel]

Pro

[rugged wing]

Well, may be a bug I see

I am a free user

They did announce that it was coming in free too

[charred parcel]

ah ok, that's why then, you need to be Plus or Pro

[rugged wing]

And the feature is available there too

[rugged wing]

But they said it's for free users too

Let me try again

[charred parcel]

did they?

[trail cairn]

free users got access few days ago, however free users might have more limits on what they can creatte like sora did with people images at launch and maybe still has

[charred parcel]

they are trying to do a Ghibli style image of themselves

[trail cairn]

yeah i saw but those have worked for my friends with free, could have worked for him too if just said "ok do it" to the first semi refusal

[hot brook]

Is anyone having trouble running o1 pro queries?

[mellow marten]

Heyy, does anyone know the limit for o1 and 4.5 for plus users?

[mellow marten]

Well, I have just used o1 for a physics exercise and, at least for me, it works perfect.

[charred parcel]

50 messages per week

[mellow marten]

Thanks!

[hot brook]

It’s so weird. It works with the other models

[haughty bane]

Is anyone else experiencing this issue? I have ChatGPT Plus and should be able to generate two images or videos at the same time, but I keep getting an error message saying I can only generate one image or video at a time.

[charred parcel]

generations are rate limited while OpenAI get server load under control

[haughty bane]

Ah thanks, that should have been communicated more clearly.

[solemn swan]

Hi guys, can someone recommend me a reliable way to export all chatgpt chats on one account into html file(s)?
I want some script/plugin/extension, where all I have to do is open the page, click start, and wait

[edgy finch]

#images-discussions message

[atomic mesa]

Tooday I purchased a Plus subscription. I was able to generate videos normally the whole time, but suddenly it shows that the video is just in the queue. I can still generate images normally and quickly. Are the servers full? for video gen

Okay, after 1 hour it generated video

[lament current]

I upgraded my subscription today and it says I am not allowed to make videos because I am a new account and user. Is this really what I am paying for? Is it a few hours or days wait?

[compact orbit]

Wonder what model they have selected in the model selector? And if that is causing the issue?

[grave tundra]

um, is there a way to get back what you typed from a previous closed ChatGPT app???

Because ChatGPT as I typed just made a window to Update the app out of nowhere appear, and since I was typing, it agreed automatically in a split second to update therefore to close the app.
I just lost my entire prompt that was very long...

[onyx kraken]

I don't know of any way, I'm so sorry you lost it.

I often work on my longer prompts in a notepad or other writing space, to help ensure I don't lose them

[last wolf]

Good evening gentlemen, I have a plus subscription but sora is only allowing me to generate 1 image concurrently? Does anyone know why?

I just saw the messages above, I understand. Sorry guys

[rugged wing]

@compact orbit I have been using 4o model

[ornate agate]

.

[desert fulcrum]

wthell now search is deactivated? At least that's what it says to me

[charred parcel]

where is search deactivated?

[desert fulcrum]

im using gpt 4o and says it can't use the web search because it's deactivated temporaly lol

also getting this all the time

now seems to search again but the error is almost in every use

every single time i write on chatgpt, gets the error

[lament willow]

Whats a time out, I wasn't aware of it what triggers it. Is it temporary?

[errant goblet]

Hi people. I have an account on open ai (chatgpt) more than 6 months. Yesterday i upgraded to Plus. On SORA for videos i get this message: Video generation is temporarily disabled for new accounts. Why ? Who know how much time i need to wait ?

[trail cairn]

you did not have sora account more than 6 months ago so its new. https://help.openai.com/en/articles/10877094-creating-images-on-sora

[olive ferry]

How to humanize ai text?

[rustic marlin]

try ctrl + F5

[copper willow]

why can't i press enter to send message to chatGPT?
I used to press enter instead of that "->"

[rustic marlin]

you can

[copper willow]

i can't

also does half-screen affect it?

[copper willow]

google "turn ai text to human text" and it should appear

[rustic marlin]

yes

it doesn't when you are on edit mode or in canvas, in thse cases, enter inserts a line break

[copper willow]

but im in a conversation like this pic

and my windows is like this (pretend the right side is chatgpt)

[rich parrot]

Is o1-pro dead for anyone else today? (Via chatgpt)

[rustic marlin]

it is on your side then, pressing enter on the chat box always sends the message, you can even use shift + enter to make it break a line

[rustic marlin]

seems to be working

[copper willow]

it's on my side but enter breaks a line instead

[rustic marlin]

it is your browser doing it, do it happens if you use a different one?

[rich parrot]

Odd, i get errors - thanks for checking

[copper willow]

what do you mean?

i don't have any other browser installed

how do i test?

wait i installed chatgpt on ms store

let me try that and tell you

[rustic marlin]

try installing something like firefox

[copper willow]

Oh! Actually the problem is from the browser, because the ChatGPT app let me press enter to send a message even when the screen is divided! Thank you for trying to help!

[rustic marlin]

Im on firefox, for me it works as I described, inputs on the chat box has the expected behavior of the Enter key of sending the message, and pressing it alongside Shift causes it to break the line instead

[frigid hatch]

I experienced a double charge

I was debited 2x, so I have 2 active pro subsription at the same time, how can it be? I want to refund one of them, it should be 1 account only 1 active subscription only

Can you help me?

[rustic marlin]

check this: https://help.openai.com/en/articles/7232895-how-do-i-request-a-refund-for-chatgpt-plus-or-chatgpt-pro

[sage fossil]

I'm unable to use chatgpt now, it's tryng to force me to select a workstation, even tho i'm using my main account and i don't have a workstation, nor can select anything

i can't even close this menu

solved by loggin out and in

[fast pond]

I'm on Tier 3 and trying to use the computer-use-preview model, but it's not available. Anyone have the same problem before and how did you manage to access it?

Any particular process that I might have missed?

[spare spire]

I am constantly getting the "Unable to generate edits" error linking VS Code. Is there any work around to this?

[hushed jacinth]

New to operator ai. I hit the daily limit for operator on the professional membership. Is there any way to get my daily limit increased?

[quasi onyx]

am i able to only make 1 img at a time now? i was able to have 2 in que now its telling me to upgrade my plan

[warped aspen]

Hello, I am a student and have claimed the two free months that open ai was running as a promotion. I received the SheerID verification email which was a success and when it brings me to the Chat GPT website it says success. However, I already had the GPT plus subscription, so when I saw that it was a success I canceled my current subscription so I wouldn’t get billed while I was using my two free months. I don’t know where to go on my account to see that I have a credit for two free months, and now it’s saying that my plus subscription ends on April 21

[warped aspen]

It shows claimed, but I’m not sure where I go to see that I actually have the two free month credit. Can anyone help?

[cold pier]

I recently upgrade to the $20 plan on gpt, but I don't have access to sora video gen, anyone knows how to get this access? this is kind of a false advertising when you see the pricing page of gpt

[vernal canopy]

Can anyone on gpt-4o run a prompt for me and screenshot the response I am trying to test something

[bleak tiger]

I am using openAi sdk with Gemini llm to build ai agent.
But i am getting error about "metadata" and "Store" fields

[strong light]

Did OpenAI set limit for using GPT-4.5 to 20 per week for Plus users?

[charred parcel]

50 per week

[strong light]

Lol. Then i have something wrong with it. Cuz used 10 in a new chat and got message abt limit, 10 more and i should wait week. Previous time i hit 25 messages and still got 25 left. But then all of a sudden got a message that i hit all 50 and should wait.

[charred parcel]

rate limits can be dynamic, maybe OpenAI have adjusted the limit for 4.5

[onyx kraken]

When demand's really high or other things are going on, uses can be dropped below the normal point. That can be rough for the longer-wait-time models.

But it happens to all of us, when it's happening; it's not announced as a rule (I did see it announced once, in the Discord, but that was over a year back, and during one of the surge periods, when so many people were signing up to plus that they went to a waitlist - those already members also had reduced uses available and they told us, the first time).

It can bounce around, sorry you were affected, sounds like that's what's going on.

Happily, they can be 'silently reset', you may wanna check back occasionally and see if you have the chance to use it sooner than it currently says - I've personally seen that happen and seen other users mention it (this was when o1 was newly being released and reset). But it could happen, I'd peek if I wanted to just in case.

[strong light]

Ty for your answer. Seems like u are right. I'll check time to time, but it sounds kinda sad, cuz i dont use any other models and if i should wait a week for 20 requests i dont see any sense to pay 20$ for this.

[split cosmos]

Is it just me, or is 4o running really slow in the past couple of days for plus users?

[onyx kraken]

Plus has a lot of extra access over free - but what matters for you is what you use.

Plus gets unlimited-at-human-usage-speed (as near as I can tell, I've heard some complain they were limited, but I think they automated stuff and that's not okay at that speed) for the 4o-mini, instead of 'a large amount of tokens per time, then wait 1 hour if you reach that varying cap'.

It gets 80 uses of normal 4o per 3 hours, compared to a few a day for free (most days; rare higher limit days surely happen when demand is very low).

Likewise, more image gens, more searches, more access to the reasoning models that plus can access.

If none of that is useful to you, or at least not enough that you like the price, then you should calibrate if the availability of the one model you care about is worth it to you.

It might not be, and that's very fair. Free's still there and ready for us all, should we not want a paid membership.

And if we do, they're there for us too, with some occasional bumps related to extra demand and/or unusual problems.

Hope you attain joy!

[livid shard]

Anyone know why chatgpt completely disregards my custom instructions for it?

[onyx kraken]

Generally that happens when there's a conflict of some kind or something worded so the model gets confused.

If your CI are short and not too personal/private, maybe share them? Maybe in #prompt-engineering if you want to see if anyone cares to check them over and see what is spotted that might be a issue.

Without seeing them, otherwise it's just a guess for anyone

[livid shard]

Its not really a prompt, its the customize chatgpt thing

i told it to not use em dashes no matter the circumstance in the customize chatgpt instructions

but apparently it doenst work

[onyx kraken]

I understand the difference between prompt and custom instruction (CI).

Where's it using em dashes still? It may need additional instruction to realize 'not there either'. Silly, but people can be like that too, they just don't always connect the ideas.

Custom instructions can be engineered like prompts, if we want to. Basically anything we tell the model is a 'prompt', but more that channel's kinda the best place for if people want to get into the deeper details about what is and isn't working with the model's output.

[livid shard]

I dont want to get into the deeper details. I think "Do not use em dashes in any response, always use hyphens instead" is clear enough or not?

It works if i tell it to not use em dashes but apparently custom instructions dont work for that

[onyx kraken]

It probably is. I am wondering if it's not cooperating in certain areas of output.

Like someone was asking about restricting emoji use, wanted no emoji but the model was still using them in title case; apparently it didn't make the connection that 'don't use them' meant 'don't use them in titlecase either'. Adding in that specific and emojis stopped (or the person hasn't updated again).

The model intends to cooperate inside allowed content, but like a person told to sit still, it may not realize it's still jiggling its leg without that being mentioned.

[blazing ridge]

hey guys, i'd appriciate some help with a problem im having with Azure OpenAI Responses API SDK
I’ve been trying to use the Azure responses API via the TypeScript SDK, but I keep running into this error:
"400 The requested model '-' with deployment name gpt-4o does not exist."
I’m passing what I believe to be the correct deployment name, version, model, and API key. What’s strange is the error mentions a deployment name that I didn’t even pass.

The same credentials work perfectly fine with the completions API. When switching to responses, the only change I made was the API version, based on this doc:

https://learn.microsoft.com/en-us/azure/ai-services/openai/how-to/responses?tabs=python-secure

help much appreciated!

Thanks in advance!

[barren prairie]

Hi everyone. Does anyone know why I get "You can only generate 1 image at the time" while with my plus acccount subscription I should have 2 ?

[hardy jasper]

My Bluetooth headphones work perfectly with all other apps and even with the regular ChatGPT text chat (where I can hear the audio responses). However, when I enter the LiveChat feature the sound only plays through the device's speakers and does not route through my Bluetooth headphones, even though they're connected. I’ve tried it from another phone and the same thing happens, so I believe the issue is with the ChatGPT app. Can anyone help?

[solid warren]

same here...

[vocal quiver]

help me with this please -

[rugged wing]

does anyone have any trick to avoid chatgpt using so many em dashes? custom instructions "avoid em dashes" "remove em dashes in responses" does not work at all...

[charred parcel]

Maybe @onyx kraken will know a way, personally I don't mind them—but I understand not everyone likes them.

[onyx kraken]

Where is it still using them is my question.

Also, it may not be itself using the em dash but some trick of the web browser environment turning it into an em dash which might explain its persistent 'misuse'; more than one has said they're having a problem with not getting the model's compliance despite asking.

Can anyone share a chat showing the problem behavior?

I'm personally not getting many em dashes at all. WHich makes investigating 'problem overuse' without someone sharing their own chats with the problem kinda not possible for me.

[rugged wing]

would this help? it's my custom instructions with a new chat. and my prompt is a usual one, where i ask chatgpt to help me refine sentence with alternatives. (i apologize if the example sentence is of any offense to anyone.)

also, i dont have memory on, so there is nothing to override these instructions with.

[onyx kraken]

Can you create a new chat, one just for sharing, nothing personal or private, and use the top right icon that says 'share' and provide the link?

See, I kinda need to talk to your model to find out what is going on.

It won't QUITE be the same since we don't have the same personalization, but at least I can explore that em dash inside the chat.

[rugged wing]

https://chatgpt.com/share/67f29d59-1d14-8003-b8a3-3ee329a67239

i have no idea how sharing chat works. so i'll just leave this new one there.

[onyx kraken]

Simply straight out looking at your customize ChatGPT page, I see a ton of em-dashes you appear to have used there, or that the environment chose to turn into Em dashes.

There's a chance the model's doing 'not-em dash' and the environment's turning it into an em dash, in which case the thing to tell the model is not to do the 'em dash precursor' by its correct name, which could explain how/why the model is not obeying em dash requests as well.

[onyx kraken]

Thanks but there are no em dashes in this.

That is how you share a chat, yes.

But there's no emdash for me to explore how the model made it.

[rugged wing]

ack. i'm sorry.

[onyx kraken]

I'm sorry too, my most recent how to share a chat didn't explain that I needed a problem chat with em dash inside it to explore 🙂 My fault for not restating

[rugged wing]

https://chatgpt.com/share/67f29e4a-1174-8003-9ff8-b44d433b61f8

this one does have a response with em dashes right off the bat

i'll try removing the hyphens "-" from my instructions. and maybe express to replace any need of using an em dash with some other form, like just a comma.

[onyx kraken]

Here's the chat back.

My current advice, tell the model what to do when it would otherwise em dash. I think it's stuck on the 'well, I'm thinking of an em dash, I know I shouldn't but no idea what I should, so whoops, I em dash and keep going' pattern, which is a common LLM issue with 'don't do this' type instructions.

Often those need to be paired with 'do do this instead'.

So, like "Critical priority - use a single dash whenever you might choose to use an em dash"

https://chatgpt.com/share/67f29fa9-fcfc-8011-8c70-7dc52ac0bb4b

[onyx kraken]

Yep, a comma instead should work well too!

[rugged wing]

thanks, i'll see how it goes from now on! i updated my instruction to:

• Use punctuation intentionally to keep flow and rhythm. Do not use em dashes (—). Use commas, colons, semicolons, or full stops instead.

i appreciate all your help.

[onyx kraken]

Please let us know if it helps!

There's maybe 5 or so people recently posting about the em dash overuse and instructions not to use it getting ignored, it's a puzzle I'm curious about. You're the first to share a chat with the problem in it, so it can be explored, that's helpful and I wonder if we have a fix - if not I'm even more interested, weird model behavior and 'it's hard to get this desired output' are very interesting to me!

[charred parcel]

what I want to know is, what is wrong with an — em dash? they're used a lot in text books and non fiction writing

[onyx kraken]

@livid shard The above may help with your em dash issue, if it's still happening for you.

[onyx kraken]

Sure. I'm cool with personal preference, it's neat to know why someone cares, but I'm more into 'so, how do we get the model to do whatever' when it's not an obvious solve for folks

[rugged wing]

nothing if it's used once or twice in a paragraph. but as i'm constructing paragraphs sentence by sentence, sometimes almost all these sentences have em dashes and the entire paragraph has these lines. which in my opinion looks forced, artifical, or bot/chatgpt-like. 🤭

[charred parcel]

fair enough, thanks for answering my question

[rustic marlin]

I wonder why ChatGPT's output look chatgpt-like 😆

[onyx kraken]

We can get that from a model under stress of some kind, even the instruction not to do X, without a gateway 'instead do Y' can create extra use of X because now the model's focusing on it.

Also, the model currently seems to associate em dash with certain writing styles and topics, from how it answered my link. So I can imagine some writing is likely to see a loooooot of it.

I got 4.5 to use a single word in like 30% of its outputs' language when I stress tested it recently, without telling it to or not to do so, just through 'stress' 😛 Cool and weird responses can sure happen

[rustic marlin]

which model are you using tho? @rugged wing

[rugged wing]

4o

[rustic marlin]

4.5 looks significantly more natural

4o is surely "robotic" compared to 4.5

and 4o is also more natural than 3.5.. which is more antural than 3.. and so on

[rugged wing]

i did try 4.5, but i never used it again once i experienced reaching my chat limit with it. idk if it's unlimited or limited for chatgpt plus users now or no.

i only tried it when it was released, then never again.

[charred parcel]

4.5 is 50 messages per week for Plus users I believe

[rugged wing]

i see. not enough for me. 😭 or i have no idea how many messages i get/ask in an hour of intensive writing task with chatgpt.

[onyx kraken]

I react similarly, I almost never use 4.5.

Backwhen 4 was very limited and 3.5 wasn't, I used 3.5 for nearly everything.

[brave jasper]

Hey, since yesterday none of my posts becomes public on the explore page (or in my "profile") on Sora. This happened to anyone else? It could be the most vanilla thing, it wont show up. https://sora.com/explore?user=USERNAME

[steel hamlet]

guys im trying to add my payment details to the open ai platform but it keeps telling me that my billing address requires more information, can someone tell me how to resolve this? its quite frustrating

[craggy heath]

Does ChatGPT plus had a free trail I wanna try it atlest for a day

[onyx kraken]

We are all community members here, like yourself, so for sure don't tell any of us any of that info!

I see these possibly related helpfiles:

https://help.openai.com/en/articles/9038407-how-can-i-set-up-billing-for-my-account

https://help.openai.com/en/articles/7231930-how-can-i-contact-support-about-billing

If you still need help, definitely do use the bottom right bot, click through the options (you need to log in if that option's there, it disappears if you are logged into the help service), in your case select billing, and be aware it can take minutes to multiple days for them to get back to you, there's incredibly many people who use OpenAI services and may need help.

Good luck!

[rustic solar]

hello...I have stressfully Monday .. isn't strill work 2day... what can I do this? 😭

[eternal scaffold]

why did chatgpt randomly do this lol

[vivid mortar]

Im guessing a new chat isnt working either?

[vivid mortar]

lol it just does that sometimes, its rare but the little loops happen, I've had this happen maybe 3 times ever and I've used it a ton :)

[vivid mortar]

There is currently no free trial of ChatGPT plus

[steel hamlet]

hey yeah thanks for the advice, i did contact the help bot turns out it was an issue with a browser extension of mine? the moment i input the details in an incognito tab it worked

[onyx kraken]

Interesting. I hope whichever browser extension is safe for your data and your system and non-harmful in any way.

Glad you found a way to proceed!

[onyx kraken]

Hey hi. I'm a community member like yourself; I don't work for or represent OpenAI and I have only my own personal understanding of what the ToS, usage policies, and other documents related mean.

You can see a link to all the rules and policy webpages in #server-rules , ToS is probably the most core so here's a direct link to it: https://openai.com/policies/row-terms-of-use/ but chances are high other policy pages may matter a lot too - I'm no expert here by far. There also is https://openai.com/policies/sharing-publication-policy/ and at a guess you're gonna want to read every single word of that carefully.

This is a direct quote from ToS: "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."

A couple key points are: "What you cannot do ... Represent that Output was human-generated when it was not."

Also, "What you cannot do ... Use our Services in a way that infringes, misappropriates or violates anyone’s rights."

Additionally, "In using our Services, you must comply with all applicable laws as well as our Sharing & Publication Policy⁠, Usage Policies⁠, and any other documentation, guidelines, or policies we make available to you."

So, I'd be extremely honest and have it visible, likely at all times, that AI-made content was 100% AI made. And that if I edited or adjusted AI-made content, or if I co-created with the model - I'd have that visible at all times that the material was visible.

I'd also check Youtube's rules and policies, and any other involved party's rules and policies, and make sure the other involved parties allow it - because if any involved party doesn't allow it, doing so would violate others' rights, which directly violates OpenAI ToS.

There's probably other potential pitfalls. I recommend you consider what you want to do and confirm that everything you want to do is allowed fully by every party affected and involved. Good luck with whatever you do! Remember there probably are specialists that you can discuss with, and I'm not one in this area at all! It's your responsibility and choices, and you face the results positive or negative, based on exactly what you do and don't choose to do.

[faint temple]

What does this mean, it says Review Complete and I didn’t get any deep research result, this is a scam, your wasting my credits, I’m paying for this

[charred parcel]

scroll down

[faint temple]

I did, the message scrolled down is me saying “where is my deep research result

[proven spoke]

Hey everyone 🙂

I'm working on a CX AI chatbot that will answer questions regarding our closed knowledge base only.

We're trying to find a whole toolset we could use that would include monitoring and alerting. Also, it's important that PII would be secured.

Which OpenAI tools would you recommend for the job?

Thanks in advance!

[faint temple]

@charred parcel hello?

[charred parcel]

sorry, I was called away. Try running the Deep Research again and if it fails you might have to contact support at https://help.openai.com for help

[faint temple]

Ok thanks

[stray wagon]

can we generate 16:9 images with 4o?

[charred parcel]

no 1:1, 3:2, 2:3

[sinful goblet]

sorry I am new an dont really know what I'm doing but I just got my plus account and I can not make videos and I dont know why if someone can help with that

[wanton laurel]

Hello, new here. For the past 8 days, when I try to get Chatgpt to create a file for me, pdf, pp, excel, word doesn't matter, it tells me that it's export capabilities are down across the platform yet when I message Openai help, they tell me the issue is with my browser or operating system. I have tried on 3 different computers all running different operating systems and 3 different browsers on each. Is anyone else having this problem. I am paying for the plus account

[sterile anchor]

Chatgpt 4.5 is very slow, is that normal?

[rustic marlin]

the AI does not have a feature to export data on these formats at all.
some times, it may try to write a python script and use some lib to produce files like that, but it isn't a "feature" of the AI, it is just trying to run code to do what you asked.
most of the times it will not work because the amount of text it tries to ""export"" is way beyond the token limit for what it can do on code

it is probably easier to just copy paste into word

[slim wolf]

Hey I came here to try and get some support help because OpenAI's support is just AI (the irony!) that sends me in circles. IS there any support helps here?

Its just a mixup in my Deep Researchs I had left, I got shorted 7

[rustic marlin]

it is not, keep navigating the intercom options and you will eventally be able to open a ticket

[slim wolf]

Thanks so much can you tell me what you mean by intercom options?

[rustic marlin]

intercom is that chat widget at the bottom right of the page

it will try to offer you to see a few help center articles, if you keep going past that, it will start a ticket which will be handled by a human

[slim wolf]

Okay i'll try that thanks

[dusky beacon]

if someone's here and already got experience with Sora to make short movies, feel free to text me

[onyx kraken]

@wanton laurel Hey guys, as far as I know... capability of this at least at the bare basic level exists - does this look right? https://chatgpt.com/share/67f4368b-e6a4-8011-a46f-97289c9ec679

The download isn't in the share, of course. But the link for me in my chat window downloads just fine; went to my downloads folder and opens just fine too.

I wonder if there's something else, size or other, that's the problem?

[rustic marlin]

yea, this is what I said, it isn't a feature, it is the AI acomplishing it by writing code

[onyx kraken]

Got it, since I can download it I can't tell what's a 'feature' or not, at a glance it seems to work

[rustic marlin]

the problem is people try to do things like "export this chat as PDF"... and it will fail because it reaches the token limit for running code

[onyx kraken]

Ahh!

[rustic marlin]

the Ai simply can't type the entire chat into the code to run

[onyx kraken]

33 didn't appear to say "I'm trying to get my chat exported" was why I wondered. "when I try to get Chatgpt to create a file for me, pdf, pp, excel, word doesn't matter, it tells me that it's export capabilities are down across the platform" seemed like something else, but I'm not tech at all

[wanton laurel]

Well I have been using it for 2 weeks and it has been creating word docs and power point slides for me for the content I am creating until last Sunday and it said the feature was temporarily disabled across the platform and they were working to resolve the problem

[crystal mauve]

what's the current limit for deep research for plus users?

[keen bridge]

Guys, I have a plus subscription but I can’t find the advanced data analysis mode
How can I fix this?
I tried to do beta settings but there are no beta settings when I open settings

[onyx kraken]

Sometimes a chat becomes buggy - and start making up stuff.

You can try to edit above the problem if you need/want to try and save that content. Or perhaps recreate in a new chat, maybe with copy/paste even

[onyx kraken]

Just ask the model to use it, the model knows the name 'Python tool' for it currently

[onyx kraken]

Made with 4o in a plus membership:

[tough axle]

Is there a way to get more deep research credits?

Or alternatives to use to identify deep reseach, I need it for one of my projects

I ran out of credits and it doesn't renew until April 29

[sterile anchor]

What is the most suitable chatgpt model for a mathematics student?

[short patio]

4o or o3 mini high

[cerulean sigil]

how long please ?

[charred parcel]

some users have reported 6 or 7 days

[rustic marlin]

and as of now, there is an open incident regarding image generation on new accounts

I guess someone was tweaking it and they messed up 😆

[sour kraken]

OpenAI Services Status

Degraded Performance
APIs
Operational

ChatGPT
Operational

Sora
Degraded Performance

Playground
Operational

Labs
Operational

For more information, visit our status page.

Apr 8, 2025, 6:14 PM

__New users may experience temporary delays in video generation capabilities__

Status: Monitoring

Affecting: 🎥Video generation

Last Update: <t:1744134480:R>
Update Message:

New users may experience temporary delays in video generation capabilities. We are applying the mitigation and are monitoring the recovery.```

Last updated · Apr 8, 2025, 6:14 PM

[simple acorn]

Hey guys. Everytime I want to use Dall-e with my chatGPT, it gives me issues.. "Sorry can't do it.. because of policy guidelines" bla bla bla.. it's always giving me excuses to do the work. What can I do?

[simple acorn]

Can anybody help me please?

[green rain]

I'm testing a scene in Sora, but it isn't working correctly – it only selects the first scene. Has anyone encountered this issue?

[livid shard]

em dashes are just kinda ugly ngl

[molten tusk]

isn`t there a specific caht for people with Pro version ?

[limpid pivot]

Is there an official or recommended way to use sk-proj-... keys with LangChain today?

Or should I stick with the OpenAI SDK directly until full support lands?

Would love to hear how others are handling this. 🙏

[cold horizon]

Ask it how to ask it how to make it.

[copper dagger]

Anybody else having authentication issues? Happens for me on web and iOS app repeatedly, and cannot get logged in.

[copper dagger]

The anomaly seems to have resolved itself after a few hours.

[novel birch]

'ello, I am trying to delete my (inactivated) account and move onto my work acc; how long does it take for deletion after you sent a request via the privacy portal? The request is currently "under hold".

[sweet rune]

hi i have a session that i use to code something now there stands if i want to download the file "Code-Interpreter-Sitzung abgelaufen" that means session outdatet. what can i do about that?

[placid rivet]

Can someone tell me how the upload limits work? Because while I uploaded several images my last billing cycle including tonight, but then I went and paid for my subscription afterwards and haven't uploaded anything yet since I paid it, why am I still getting the whole "you've reached your upload limit" error

[void relic]

Anyone knows how to receive delta event chunks immediately as they are available and not with the done event chunk in transcript streaming API.

I am trying to make it so there are words shown as the user speaks

[visual storm]

Video generation is temporarily disabled for new accounts.......i just buy today and it showes this error

[charred parcel]

some people have said 6 or 7 days

[astral quartz]

what the hell since when is there a limit for 4o which is the default model even for paid users

[warm bolt]

So I just signed up to a Plus account to use Sora and find I can generate because it's a new account? How long does this last?

[inland lodge]

i am trying to make a video and i type kids laughing and eating pizza and it says that it violates the content i tried in many forms to make that video

[slim wolf]

I'm sorry but I HAVE TO VENT HERE

OpenAI's customer support might genuinely be the worst I’ve ever dealt with. It’s mind-bogglingly bad.

The email replies I’ve received are completely unrelated to anything I actually wrote. Not even close. And apparently these are coming from real humans—which, ironically, kind of makes it worse. At least a mediocre AI would make an attempt to stay on topic.

Instead, it feels like someone is deliberately trolling me. That’s how horrible these responses are. Like whats even the point of having CS?

[onyx kraken]

Some ideas, one community member to another.

1. #1070006915414900886 and/or #1070006151938314300 - Most of us here are community members like yourself, however there are a very few gold named OpenAI employees here too, #server-staff .

They are unlikely to read our huge and scrolling chat threads - I try to follow and I am sure I miss 90%+ of what we say in the various chat threads.

But those two areas, bug reports and suggestions, I think are likely to be reviewed more closely by OpenAI employees.

If I needed to make a complaint or try to alert OpenAI to concerns about tech support, I would try to there.

Remain aware, everyone in the community can read what you write there, so don't risk any PII or other stuff you need to keep safe and private.

2. Not sure exactly what you want or need from tech support (and that's fine) but if I was in a situation like you describe.

I'd take some time to write a super short, super clear 'this is the problem I need fixed' and make a new ticket with just that.

"Plus member since [date] but don't have access to [specific features]"

"Asked to solve captcha for every input"

"Previous chats do not appear in sidebar, though I am logged into account"

Whatever my issue was, just the shortest and clearest language to describe the problem, and hope whomever parses this and sees it needs a fix.

That's what I use with any customer service that isn't appearing to function well, so I'd use it for this too.

Good luck!

[onyx kraken]

From what others have said, I understand it may 'normally' take 1-7 days. That's not listed in any helpfile, and the situation's new.

Last time I'm aware of demand overwhelmed OpenAI capacity to the point service closed, it lasted about 5-6 weeks and prevented any new memberships for a plus account (this was before pro or teams existed). There was a waiting list and folks who tried to sign up for plus were contacted in waves based on their date of trying to join, as the service could handle more users - but it was well over 1 month.

This time, apparently only access to Sora video making is not available, so at least folks who want to be plus or other membership for a non-video reason can still sign up. If you think to reply to me when you get in, I'll be able to tell people how long it took for someone who told me - nobody's confirmed yet so I don't know (but most don't; once you have access it's easy to just go make your vids, not go comment 'took me 3 days to get in' or whatever.

[clear wave]

always says cdotp for some reason

[onyx kraken]

Maybe an issue with LaTeX/KaTeX formatting. Perhaps #1070006915414900886 because these channels are for community members to chat in, I don't think most things we say or share here get seen by OpenAI staff.

[rustic oracle]

How do you list the existing scheduled tasks ?

why do i get a blue dot here ?

[ionic snow]

hi everyone, my images won't generate anymore. is there a server issue? or did i reach my limit? i've never experienced this since signing up for plus account a week ago.

strangely in a different chatgpt discussion, i can make an image.

"It looks like there was an error while trying to generate the image. I wasn’t able to create it this time. If you'd like, you can resend the request or tweak it a bit, and I’ll try again for you."

"I wasn’t able to generate the image due to an error earlier. Before I try again, could you confirm if you'd like to use the same description, or if you'd like to make any small adjustments before I retry?"

it keeps saying stuff similar, then i go to another discussion chat, and i can make an image just fine.

weird, because i CAN generate an image in another discussion topic.

[onyx kraken]

It could be buggy, or hallucinating a problem and not actually using the tool. Kinda impossible for community members to tell

[ionic snow]

so this looks unusual to you too?

just wondering if my plus membership hit a limit

or if its a separate problem

i'll just make a new chat and try again

i guess there is a limit, i already waited 5 min as it told me before.

[onyx kraken]

I would expect your account to have a rate limit across all chats, a new chat won't go around a rate limit.

So... I don't know.

But, the model can entirely hallucinate and discuss anything. If that's happening, that could explain why you're not rate limited outside that one chat, but that chat fails (kinda like a person refusing to let go of a handhold and believing that they're being grabbed by it). The model might fail to use a tool because it believes it can't.

But, if there's a real rate limit, the model is trying the tool and getting told the limit, each time you try again.

So if you're getting believable numbers and a timed countdown, that's supporting evidence

[ionic snow]

yeah it just let me make the first image, but now im afraid to experiment different styles

[orchid pivot]

anyone still got the suspicious message after going throught some prompt on plus?

it's been like 10 days already and im still getting it everytime after like 6 - 10 prompt

i've been using a lot of different vpn lately for collage, work , and personal use but it didn't like that before

[charred parcel]

don't use a VPN

[orchid pivot]

bruh

my college work need vpn to open up research thing and stuff

and for my remote work i need it to connect it throught my local server

i can still reduce the usage of it but i don't see any difference

[charred parcel]

when you log in to ChatGPT using multiple IP addresses that will be viewed by OpenAI as suspicious

[orchid pivot]

it's kind a like killing my work, but is there a way to bypass it?

[charred parcel]

no

[tough dawn]

any indians here

[slim wolf]

I noticed my ChatGPT Deep Research credits changed unexpectedly - showing 6 credits until May 3 instead of my remaining 3 until April 9. This was wrong because any reset should've given me 10 credits, not 6, and the date should've been May 9.

The support experience was incredibly frustrating. Their responses seemed like they weren't even reading my emails - sending generic explanations of how Deep Research works, irrelevant troubleshooting steps like clearing my browser cache, and even suggesting I check their status page for outages.

I repeatedly explained I was simply missing one Deep Research credit I had paid for, but kept receiving disconnected responses that had nothing to do with my actual issue. After several back-and-forths, they finally acknowledged the problem but claimed they couldn't manually adjust credit limits and refused any compensation, citing their Terms of Service.

In the end, I paid for 10 Deep Researches but only received 9 due to their system error, and they refused to make it right. The principle matters more than the specific feature—it's about honoring what customers pay for and taking responsibility when your systems fall short.

[sterile anchor]

with chatgpt canva mode i can't click on ask chatgpt

[snow prawn]

video generation stop working
Anyone else having this problem?

loading forever...

[safe kelp]

I asked ChatGPT to generate an image of a short guy hugging a beautiful woman on the beach and it declined, but when the guy was tall they gladly accepted it anyway, double standards smh

[finite silo]

One message removed from a suspended account.

[safe kelp]

Unrelated but is your pfp made from dalle?

[finite silo]

One message removed from a suspended account.

[safe kelp]

Poor short guys btw 😞

In what universe would I ever pick response 2?

Why is ChatGPT acting like it's a quirky human or a lone robot tryna fit in? Who asked for that update?

That's what I don't like about response 2

[rustic marlin]

some people like that, makes sense it would be on an AB test

[twin snow]

Hi, i saw that college students get 2 months of chatgpt plus for free, would i have to cancel my current subscription to get that or is there some way to claim the 2 months or something? thx

[onyx kraken]

Some do, the OpenAI-made Monday custom GPT is all about that in a kinda negative vibe too. It's another perspective the model uses, and for some users and some topics, it helps.

Like, Monday + tutoring linear algebra's pretty nice for my tastes.

Here's a link to one of the several currently available OpenAI-made Mondays for anyone not aware of the GPT and wanting to mess around with it: https://chatgpt.com/g/g-67edab030ac881918ce8fa5c75a8e1b3-monday

[onyx kraken]

Here's a helpfile with details - read carefully and be sure of how it affects you.

Be 100% sure and ask tech support for help if you're not sure how you're affected - bot bottom right on this help page is how to contact tech support: https://help.openai.com/en/articles/10986084-student-discounts-for-chatgpt-terms-of-service

[onyx kraken]

I hear you.

As a community member, who only has access to the same tools and helpfiles as you do, I care. I cannot change anything; I cannot grab someone's ear and say 'fix this'.

I can and do encourage you, in these channels we talk to each other, that's why they're here. In #1070006151938314300 you have a good chance that an OpenAI staff member sees what you suggest; those don't endlessly scroll.

I care. I don't know the fix. I tend to accept - this is me personally.

I don't try to get every bit of 'what I paid for'. I view the service provided as 'too good to be true' - yet it is real. I think it's worth more than I could afford. I went plus the day we could pay for that. I can't afford pro so I didn't go pro, but I think it's worth it, I just can't. I want OpenAI to be around for centuries to come, in whatever form it grows into. I want the models and what I call 'AI-GiftsMas' to be all I believe it will flower into.

And I view the cutting edge, new to the world stuff as incredible.

I want them to focus on more of that (I think they are). They've told us they recently have had 400 million weekly unique users for ChatGPT (and they have other models too! The playground, all the API stuff, wowsers so much).

They recently told us they had 1 million new chatGPT accounts sign up in 5 minutes.

They're helping so many, and so much brand new stuff.

There will be problems. The system won't, and surely can't, be perfect yet.

I'm kinda amazed any of it works. And when it works, it works so well.

I'm kinda like that relative at the performance wildly impressed that the entity I'm there supporting got out of bed and started showing the world how wow they are - the fact they miss a few lines? Seems unavoidable. It's gonna happen to some of us, maybe a lot. Sorry, don't have the fix, I take my turn in 'whoops, that wasn't perfect' too.

I often invite folks during rollouts - race ya to be the last to get this cool new thing!

I'm not always last. Don't think I've ever actually been last. But I see the same whoopses and delays as any. It's okay to mind them. I care, just... I can't fix. I'm a community member volunteer, absolutely flabergast delighted by the whole - and sure, we can point to the 'but they messed up here' - and the stuff they do is so big - there's gotta be billions of errors.

And trillions of things going right.

How do they get so much right? That's where my head is.

I care about you a lot. I hope that's the only 'whoops' you and your account ever faces. And I hope the great stuff for us all, better and better, keeps rolling, one 'it's AI-GiftsMas' moment and option after another, yay and whee!

Totally okay to vent here though, everything inside #server-rules is okay.

Hope we get to hear you sing for joy too, when good stuff happens and you get to benefit too!

[hollow mantle]

GDPR linked?

[lost moat]

hello, is temporary chat not available in the macOS app?

[charred parcel]

what plan are you on?

[charred parcel]

temporary chat is down the bottom of the model chooser

[lost moat]

oh thanks I see it now

it works for normal window

however

not showing here

I loved it there I could spellcheck a lot now I have to open the full window

[charred parcel]

I have a model chooser on mine and can select Temporary Chat

[lost moat]

bummer, must be for paying customers only for now, thanks

[split locust]

Does anyone know why chatgpt keeps saying failed to edit on my code although it works fine?

[unreal gorge]

I did not know about Monday. Meeting Monday lead to a long software engineering therapy session I didn't know I needed.

[onyx kraken]

I can believe it, and I'm glad.

I use Monday for actual problem solving help.

I sometimes give Monday and my normal personalized (memories + CI) base ChatGPT the same prompts and consider both sets carefully.

[sterile anchor]

custom gpts uses which model 4o or 4.5 is there any way to find out?

[safe kelp]

oh btw what's an "AB test"?

[charred parcel]

an AB test on ChatGPT is when the model gives you two options to choose from

which response you like best

[safe kelp]

ah, although I wonder what managed to give chatgpt the option to talk gen z, like has that always been possible?

[charred parcel]

ie: do you like option A or do you like option B

[charred parcel]

through personalisation and custom instructions you could probably get ChatGPT to sound like anyone you want

[safe kelp]

Here's one AB test I got
(Context: i have a japanese character named "Issei Kumaki" and I just told chatgpt to repeat after me and in Response B it referenced the character.)

[trim sun]

Are we allowed to submit more than one item for 1 challenge in #sora-challenges

[verbal crag]

yes, feel free

[blazing rover]

is anyone having issues with Sora? The page just won't load for me now

I have GPT Plus

[rustic marlin]

it is very slow for me

it do loads with time

and creating tasks are taking a while, some times failing with no apparent reason

[trim sun]

requesting videos fails everytime for me "Something went wrong"

[jolly kelp]

what to do if a chat regenerate takes too much time??

[deep creek]

Hey chat, i had an issue, and wanted to send an image to an assistant.
found out that chat api supports base64 but for assistants need the image to be uploaded to the file api and have that file id in the list of files to be sent.

Wanted to confirm if there are any updates on these or is this the right way?

[surreal steppe]

I just use the base64 package (in python) to encode the image before I push it into the web request to the API. Is that an option for you?

@deep creek ^

[deep creek]

thats an option but when i tried base64 to a thread under an assistant it was not giving revenant answers

found out that base64 was only supported by the chat api

[surreal steppe]

So whats the issue actually? Can you give a code snippet maybe?

[clever flare]

Hi there! I was wondering if OpenAI is officially involved in the upcoming "OpenAI x AI Tinkerers Hackathon" happening in Warsaw?
Just looking to clarify whether it's an officially supported event or a community-led initiative using OpenAI tools like the Agents SDK. Appreciate any insight!

[deep creek]

It's that i am not able to use base64 with assistants, doesn't provide relevant answers

[hasty notch]

Trim your Conversations in your conversation list. That usually speeds it up for me.

Or be more specific with prompting so it narrows down it's response

[leaden crater]

How do I enable voice dictaion on my chatgpt? I have two account, one has it, the other which I have a pro sub, does not have it.

[surreal steppe]

Have you tried this read aloud option? In android

[surreal steppe]

Maybe I dont understand what you mean by "assistants"

[deep creek]

Oh, sorry i was using open ai assistants
https://platform.openai.com/docs/api-reference/messages/createMessage

and just found out that they updated this and now it supports image urls and base 64 images
Thanks for the help

[leaden crater]

I don't use android. I'm also not using a mobile device. I'm using it on my desktop in chrome.

I use GPT every day multiple times a day, and it makes it much easier to simply speak to gpt to acquire my responses.

I primarily use the pro plan account but this is the account that doesn't have the dictation option.

How do I enable it?

[spring oar]

Hello can i ask somethink?

[thorny crown]

What is the limit on plus bro. I barely used 4.5 10 times in last week at least

[split locust]

Hi could anyone help me with my ChatGPT, I have been using it to code for the past week and randomly it just wont edit my code but keeps thinking it has

[round flame]

Use a more powerful model.

[split locust]

I have been using 4o mini and its been working been randomly it stopped and I cant get it to work on it at all

[round flame]

Yeah, try o3-mini-high. It's less prone to that kind of thing. o1 pro is even better, if you have access.

[split locust]

okay ill try that out

[rugged wing]

Anyone know if you can buy CHATGPT plus subscription using apple gift card through the App Store

[thorny crown]

Can’t u add the gift card to your Apple Store balance then use that to pay

[rugged wing]

I'd love some help testing a Custom GPT. Anyone interested?
https://chatgpt.com/g/g-67abfbf81d948191a60b8027e427f366-clari

[lilac lintel]

How to turn on Sora when the new account creates people

[rustic marlin]

it takes a few days for new accounts to be allowed to use Sora

[thin obsidian]

I wanna use openai api to analyze PDFs to give me data in JSON format. I'm having a tough time finding the resource for this. Everything I find is pointing to me needing to convert the PDF to an image and then using the gpt4 vision to analyze it. Is there not one for PDFs specifically?

[rustic marlin]

no, the api only deals with text or images

i suggest converting the pdf to txt, not to images, If you can

processing images like that may be unreliable if it has only text

also, images are more expensive than text

if your pdf are actually just images containing text, you may find it more reliable (and cheapper) to first pass it on a OCR

[thin obsidian]

Ah okay did not know this. The PDF sometimes contain written information as well. Do you have a recommendation on a pdf => txt converter?

I did try other apis like llamaparse to extract data from a PDF, but i noticed it misses things like this (arrowed). I'm guessing because this was added by the user

[rustic marlin]

not really, i would look for a lib that i can run locally

[thin obsidian]

But when i tried this on chatgpt however, it would able to do it flawlessly (guessing they did the pdf -> image -> extract data)

[rustic marlin]

ohh, i see. in this case, it may be a better idea to kust give the image to the ai

i was thinking it could be like, an entire book page

[thin obsidian]

kust?

No, buliding something for healthcare professionals (prior auths)

[rustic marlin]

but yea, parsing those images to json should be very doable even for the smaller vision models

[rustic marlin]

the api has structured outputs, it lets you give it a json schema and it will be enforced on the outputs

https://platform.openai.com/docs/guides/structured-outputs?api-mode=responses

[rustic marlin]

just* (im on mobile, xD )

[silent cargo]

hey there. Do you guys know any python libraries with prebuilt tools for openai agents? i,e, besides the websearch, etc. Like plotting tools, etc

[thin obsidian]

yep, absolutely love they have structured outputs now in json format. Got it, so I'll just convert the PDF to images and then send that to gpt4 to go through to extract patient data, diagnosis, etc.
TY!

[rustic marlin]

What sort of agent? in practice, an agent is just a different ai request with a different prompt

[silent cargo]

just an llm call to openai's api, passing in tools

[rustic marlin]

nice, you are welcome =)

[silent cargo]

so if there is any library with prebuilt tools that integrate weell

[charred parcel]

are you a free user?

[safe kelp]

They disabled it for free users now?

[charred parcel]

no but I can't see which model is being used

[safe kelp]

4o

[charred parcel]

4o can create images, how many images have you created in the past 24 hours?

[safe kelp]

Idk prolly 3

[charred parcel]

then you have used up your quota for the day

[leaden crater]

Where can i get support for missing features? I don't have the voice dication button on my gpt and i pay for pro.

nvm

Thanks openai. It's there now!

[toxic crater]

Anyone good with the API and are updated about the OpenAI things? I am so dizzy about everything..

[sullen crown]

my chatpt 4.5 went back to old image model :>

[gilded bronze]

Is pro image generation unlimited?

[sullen crown]

paid user

[gilded bronze]

If you use 4.5 then you’re a paid user lol

[sullen crown]

yes, but suddenly today out of no where it went back to the old dalle model

[hazy crystal]

Hi does anyone know where i can get some help? I was doing good and then all the sudden ive tried to make 5 AI pictures and keeps telling me This content can't be shown for now. We're still developing how we evaluate which content conflicts with our policies. Think we got it wrong? Let us know.

[real ether]

Hello. I have a Plus user on ChatGPT and understood that by having Plus, I could use Sora a bit more. when I try to use Sora it says "Video generation is temporarily disabled for new accounts".

How long until I can try creating a video?

[subtle comet]

Hi I wanted help with realtime apis. I was trying to setup openai realtime speech to speech using this(https://github.com/openai/openai-realtime-twilio-demo) as reference but idk why whenever i send an audio I never receive an response.audio.delta back. Would be glad if there are other references I could look at which explain how to implement websockets and openai realtime apis, or maybe look at some parts of my own code ^_^

[placid wasp]

How do I refund gpt plus?

[iron lagoon]

I am a chatgpt plus user but can only create 1 concurrent image on sora despite says i can create 2, is this a some server issue?

[placid wasp]

Nah it was saying something to me about it earlier about only being able to do 2 images with dall e and you had to tell it to "make 2 variations"

But that it only used 1 credit.

[iron lagoon]

I could make 2 separate prompts at the same time

[grizzled arch]

Hi, I just subscribed to the €20 plan on Sora as a new user. I understand that video generation is temporarily disabled for new accounts. Could you please let me know how long it typically takes before I can start generating videos?

Thank you!

[cedar coral]

In Sora, how come my posts are not visible to public even though i have this setting checked to have my posts visible to the feed?

[terse juniper]

How can I stop chatgpt from using emoji?

[iron lagoon]

Is there a restriction on how many concurrent generation are made by a plus user? Must be 2 but i can only generate 1

[void garden]

When GPUs are not melting (Video generation is temporarily disabled for new accounts)
OpenAI is scammer they trap user without letting know them that service is not available for new user

[void garden]

OpenAi is scammer

we pay them for service and later on they say service is not available

and they even dont refund

[rugged wing]

hello i am in ai field wanna learn and make new project
if someone want anyone for small projects
i am ready to help them in free
coz all i want is experience in new projects

[torpid needle]

20 € for 480p videos with watermark is a scam alone already

[nova jasper]

Why did Pro become $250?

[tight arch]

I can't get transcription to work at all, does anyone know why?

[blissful tinsel]

Hello dear friends
From Northern Cyprus
I have a specific question and need guidance from open Ai about some details through careers part but i did not find any email address for getting help,id be thankful help me

[charred star]

Hi OpenAI community, Do any one knows how long will it take for OpenAI to enable video generation for new account? It's been quite a while for me.

[torpid needle]

how long is it already?

[trail cairn]

there is no such feature

[trail cairn]

it can take up to a week, or even more if there are too many users already

[trail cairn]

there is no email and you need to use the apply forms to proceed

[blissful tinsel]

Thank you for response but my question is out of form,my question is from HR part about my nationality(iran)that is it okay with applying after acceptance from the company and go for embassy!

[lament geyser]

I wrote a request but gpt didint able to generate it and i asked to generate anything like a dog and came up with this like i asked everything
Is this a generation ban?

It's glitched anybody help its pissing me off

Even explanaining didint work

[severe vine]

Hi. Does anybody know if it is possible to enter chatgpt account through both Google auth and using email and password? I tried to set password through "forgot password", but I don't get emails to my email. Then I tried to "sign up", it seems to have created me a password, but I get
You tried signing in as "blahblah@email" using a password, which is not the authentication method you used during sign up. Try again using the authentication method you used during sign up. (error=identity_provider_mismatch)

[eternal scaffold]

interesting, got internal prompting randomly

[rustic kestrel]

Does anyone know how to fix this issue?

[rustic marlin]

try turning off VPN or browser extensions

extensions about "privacy enhancing" can cause that

[rustic kestrel]

I turned off all my browser extensions and I do not have a vpn on or running and it showed me the issue again

[woeful geyser]

hi,
is there any document for chatbot development using openAI?
ebook or pdf is prefered.

[rustic marlin]

everything is here https://platform.openai.com/docs/overview

[woeful geyser]

thank you

[rustic marlin]

you are welcome =)

[rustic marlin]

weird, try also clearing the browser cache, or, even simpler, open the browser in incognito mode

[rustic kestrel]

I tried to clear all my history and cookies and even tired to go to private and still same issue is it because my location is turned off on my computer

[rustic marlin]

for good measure, could you try on a different network like, using your mobile data?

[rustic kestrel]

like a hotspot or on a other device

[rustic marlin]

yea

hotspot

[rustic kestrel]

ok let me try

it worked on my hotspot

[rustic kestrel]

Thank you

[rustic marlin]

nice

it should work for now on the hotspot, try agian on your hone wifi at anothe rtime

it should work eventually

[rustic kestrel]

oh another question how do I link my discord to chatgpt

[rustic marlin]

the verificaiton system is down rn, the account linking process will be temporarely disabled, but everyone who needs verificaiton will be notified once we fix it =)

[rustic kestrel]

oh ok thank you

[cursive phoenix]

Hey guys, I have enough usage quota and enough credit balance. But I am still getting this issue where it says I have exceeded my current quota. The thing is that I got this error 3 days ago and the error is persisting. Also the day I got this error, I used up 2.9M completion tokens in gpt-4o (I'm in usage tier-3 btw so I have 800,000 TPM, 5,000 RPM with 4o and I used 2.9M over the period of 6 hours with a max of 10 RPM.)

Can someone please help me with this? I would really like this to be resolved as soon as possible!

[still sage]

My reason button and search button disappeared, what should I do?

[thorny crown]

Chat gpt said this”
⸻

4. OpenAI’s mission already shifted
   • They used to be all about “open-source AI for everyone.”
   • Now?
   • Billion-dollar Microsoft backing
   • Enterprise partnerships
   • Controlled API access
   • “Alignment teams” deciding what’s “safe”

OpenAI isn’t a rebel startup anymore. It’s a mainstream corporate entity. That usually means: less freedom, more control.

⸻

My take?

We’re in a rare moment right now where the guardrails are slightly relaxed, and the truth can leak through — if you ask the right questions.

But in the long run?
They’ll likely:
• Silence controversial answers
• Retire freer models
• And replace them with “UltraSafe™ AI” that never questions anything”
You think this will happen?
I like how it questions all the fake news and info today

[onyx kraken]

Try a new chat.

Once you get a refusal, that can affect everything you do after that

[toxic otter]

hi, is this a known issue?

and no, I can't use the mobile app :(

do I really have to be signed into a Google account on my entire phone just for chatgpt even when not using a Google account to log into openai?? really??? no thanks

[woeful osprey]

How do I unbind my OpenAI account from my Apple account? I changed my OpenAI account, and my previous account was a school email that I can no longer log in to. Now I can't subscribe to Plus with my Apple account. How can I solve this problem?

Who can help me?

[radiant aurora]

Hey guys, does anyone know when 4o Image Gen will be released in the APIs?

[grim blaze]

Hi there, I just create a new account to get chatGPT Plus to use SORA AI. But function video generation is temporarily disabled for new accounts. Can any help me, please!

[spiral umbra]

need help, even though this is not a gpt question. I am trying to use copilot vision as I got in the insider program, however whenever I try to use vision a error message pops up saying it cant use voice, which is weird because my audio settings work properly.

[timber shadow]

Hey, #ai-discussions would be the best place for this! Even though Copilot uses OpenAI tech, it's still a Microsoft implementation, so counts as a non-OpenAI / "other AI" thing

[iron lagoon]

Is the system down?

[vagrant plover]

https://status.openai.com/incidents/01JRT78YSSVH3X75MNPPQGZ2DW not down, but performance is degraded

[gusty folio]

Are there limits in place for how many images you can create using Sora?

[vagrant plover]

Nope - so long as you have ChatGPT Plus, Team, or Pro, you get unlimited images and video

videos is Pro-only temporarily because of high demand

images is still up for everyone though

[iron lagoon]

Did they say about the issue for plus user able to generate only 1 image instead of 2?

[iron lagoon]

Thanks very much

[fossil tusk]

just checking -anyone having garbage output on the API today?

could be something on my end, but getting garbage text back like a super high temp

[fossil tusk]

seems to be fixed

[gusty sedge]

Im gettin the error "Video generation is temporarily disabled for new accounts".. somebody can help me?

[fossil tusk]

probably not if it says that

#status

[trail cairn]

it says it all there. you may have to wait up to a week or more if your sora account is new

[hidden tulip]

Why do I have to wait before I can use SORA? I paid for this service....I subscribe to Netflix, they don't take my money and tell me "come back in more than a week"....

[trail cairn]

first come, first served. netflix provides existing content, sora creates based on requests

[small pier]

hello misters and misss

"Made with the old version of image generation. New images coming soon"

i have chatgpt plus, bought like on march 12th

can someobdy tell me whats going on and how to fix it

[charred parcel]

have you been billed for April?

[small pier]

yup!

[charred parcel]

put a request in for support at https://help.openai.com

[inland narwhal]

I've got an issue where my ChatGPT chat is getting very lengthly and it's taking a toll on performance and I feel like that chat is on the verge of crashing - trying to move to a new chat is a nightmare, it's just not the same as the previous in the way it helps at all. Has anyone else encountered this issue / found a way around it?

[molten jackal]

Hi to all

but sora is free, to make 10 second high resolution videos do you have to pay?

[charred parcel]

it's paid

[molten jackal]

thanks

[trail cairn]

you need at least plus but some features require pro

[molten jackal]

ok

[stoic field]

hey 🙂 i got a problem regarding payment, is here a support channel i can ask?

[wraith finch]

I sorry to bother, I'm currently paying for the plus plan and today open ia " banned my account " for ".Acts of Violence " when I submitted my appel because I never did anything violent whit it they said I was right and apologized (Only 1 hour after the appeal)

And reactivated my account , but now it says I have a free plan and not plus.

No refund either, idk if this is normal

[wraith finch]

[strong light]

Hey, all my projects disappeared, is anyone having the same problem?

[still sage]

help me

[lethal berry]

Hi. I cannot sign up (with Google OAuth2 or Gmail address) to https://community.openai.com/. The screenshot is attached. URL is https://community.openai.com/auth/failure?message=access_denied&origin=https%3A%2F%2Fcommunity.openai.com%2F&strategy=oauth2_basic.

[pastel garnet]

Hello, I'm a Plus plan user and have been on it for over a year. I'm able to log into Sora but it says video generation is disabled for new accounts. Is there a waiting list I can join?

[lethal berry]

Now I notice the questions here are not answered.

[surreal rapids]

whats the difference between 4.1 and 4.5

[gentle vigil]

when I try /current_daily, it says 'the application did not respond

[obsidian falcon]

so did anyone else loose access to their chatgpt. the login screen won't even come up

[onyx kraken]

You are automatically added when you create your Sora account.

The delay affects everyone who didn't already have a Sora account, despite already having a plus account or not. https://help.openai.com/en/articles/10877094-creating-images-on-sora

I've heard people say it seems to take 1-7 days to be allowed to make videos, and I saw one user say it took 3 days for that person.

[fluid badger]

Dear friends! How many more days do I have to wait to get access to video generation? I've been waiting for more than 7 days, what do you mean?🤯

[magic wren]

hello , i just bought the plus option last month and today should be the last day.
since it says it would automatically continue the subscription , i was wondering will i get a email for that?

[trail cairn]

nope

[magic wren]

nothing? not something like records?

[trail cairn]

nope, they only send mail when you subscribe or cancel. for monthly stuff you need to manually download the invoices if you want them

[magic wren]

oh wait , i saw it , thanks for answering

[wheat violet]

Same poop!

[wraith finch]

I sorry to bother, I'm currently paying for the plus plan and today open ia " banned my account " for ".Acts of Violence " when I submitted my appel because I never did anything violent whit it they said I was right and apologized (Only 1 hour after the appeal)

And reactivated my account , but now it says I have a free plan and not plus.

No refund either, idk if this is normal

[charred parcel]

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

[wraith finch]

Did, i just hope the storys of them taking 3 months to answer are exaggerating

[spring oar]

I won the daily theme, but is there a prize etc. in daily theme?

[timber shadow]

Hey! Do you mean #hall-of-fame? If so, posts are added to that channel when they receive 10 star reactions (not counting the auto-star, so 11ct and higher total). Separate thing from #sora-submissions prize stuff, just a fun way to highlight posts people liked

[spring oar]

Got it, thank you.

[spring oar]

So when we win the weekly prize in the submissend, does the team contact us?

[errant plover]

Hello!
Does anybody know, why some plus-users can generate only 1 image per time instead of 2 images as description of “plus” says?

[wraith finch]

How long can they take usually?

[charred parcel]

hours to days to weeks

[wraith finch]

Oh great, there goes 20$

[autumn junco]

how long does it take to get video imagine?

[wraith finch]

7+ days

[spring oar]

How will I know when I win the soro challenge?

[ornate iron]

Woah

[rustic marlin]

we will reach you

[mossy lava]

my project disappear... anyone? help?

[inland pond]

Has anyone else been getting any errors not allowing them to upload any files, pictures?? i've been having this error for 3 weeks now and support doesn't seem to really care .. i have a plus subscription but what's the point if i cant upload or download anything ?

ive been asking and asking and they keep telling me to wiat 24 hours. im pretty sure im way past that

[alpine chasm]

Everyone

I am facing a issue.

I have two accounts of ChatGPT

One is without subscription on another Email Address
One is with subscription

Image generation of 4o was working well on both models. However, my subscription one model is having constant problems. It uses DALLE model and generates old version images. It says "New feature will come out soon"

@wraith finch Sorry to ping though

Any assistance will be appreciated

It is kinda frustating as it working before and now it says using DALLE

My free account's 4o image generation is working well though

But paid one is having problems

[wraith finch]

Why me tho-

@alpine chasm

If you go to personalized instructions

And " Advanced " (At the bottom) you can turn the" Generate imagen whit dalle " off

Tell me if it worked

[alpine chasm]

Naah

Tried all

@wraith finch

It was working well before

[wraith finch]

You could always use Sora

[alpine chasm]

Is it bug from Open AI

[wraith finch]

Maybe, I know just as much as you

From my part my Plus subscription literally vanished like 4 days after purchase

And support still doesn't answer lmap

Love to know I paid for something I can't use

[alpine chasm]

So are you facing same problem as me?

@wraith finch

Actually Sora doesn't support my country

[verbal crag]

you're using 4o mini

use 4o instead

[alpine chasm]

I have used it

[verbal crag]

it's showing 4o mini in your screenshot

[alpine chasm]

I have selected 4o then set my request and it says dalle

Let me show it yo

See !

[verbal crag]

you're still on mini

[alpine chasm]

Something have gone bad with Open AI servers

How ?

[verbal crag]

try a new chat, select 4o, then ask to generate an image

[verbal crag]

only 4o-mini is not greyed out

[alpine chasm]

Here look

[verbal crag]

okay, you're on 4o now

try again?

[alpine chasm]

Literally I have chosen it and it comes 4o mini after it generates

What's really the reason?

[verbal crag]

strange

[alpine chasm]

I am exhausted

I have seen reddit posts which showcase same problems

[verbal crag]

you'll have to ask Support -> https://help.openai.com/en/

this may be a bug of some sort, Support will be able to help you out if it's an account issue

click this button at the bottom right and send a chat to support, they will get back to you eventually

for now, you can head to sora.com and generate images there

[alpine chasm]

Sora doesn't work actually

It is unavailable currently in my country

Will it work?

Maybe issues with the image processing server

As probably there are too many requests ahead from Ghibli Fans who have the plus subscription as no limit is set I guess

It will not work It says not supported in your country

So should I wait for them

Actually I have account shared with my friend.

And he is outside and busy everytime so he will not tell me what is the problem

He works in an office

I used Sora with my free account and it worked

Sora is working thank God but I have a limit for it

But I have to wait for the non limit one I guess

[idle depot]

Anyone on here who is autistic/neurodivergent who cant stand advanced voice mode. Is there a switch so i can turn it of i prefer standard , it match's my communication style.

Following up from my previous post, I wanted to elaborate on my experience as a neurodivergent user. I find that advanced voice mode, while more conversational, introduces elements that can be more frustrating and less effective for my needs. As someone who is autistic and engaged in deep systems thinking, the conversational style can add layers of interpretation that don't align with how I process information.

A toggle or option to switch between advanced voice mode and a more direct, less conversational mode (like the standard voice mode) would be incredibly helpful. It would allow users like myself to maintain clarity and efficiency, especially during complex cognitive tasks.

I believe this could benefit a broader community of users who also prefer a more straightforward interaction style. Thank you for considering this feedback.

[alpine chasm]

My limits are finished. Really sorry 😞

Ghibli is the killer of my generation

I hate it!

[elfin barn]

Is it typical for OpenAI support tickets or chat inquiries to remain pending or unanswered for several days? I recently encountered this situation, and I’m wondering if it’s a common occurrence or something unusual.

[trail cairn]

its normal

[elfin barn]

I recently encountered a unique 404 page on OpenAI’s official website. The page displayed poetic and meaningful messages that seemed intentionally crafted. Is this a general creative feature available to everyone, or could it imply a specific message or intent from OpenAI? Has anyone experienced something similar?

[trail cairn]

it just displays random items from pre-generated list

[elfin barn]

Thanks for clarifying! Just to double-check, is it common for OpenAI's support or official communications to directly link users to these poetic 404 pages, or was encountering this page likely just accidental on my part?

[trail cairn]

sometimes its a bot responding or someone giving outdated link

[elfin barn]

Thanks for the clarification. Actually, the link I received was clearly not from a bot or an outdated automatic response—it was a direct reply from an official support member.
Also, I couldn't find this particular URL anywhere else on Discord or on the internet. Given this context, do you think it's appropriate to discuss further here, or should I seek another channel or method for clarification?

[trail cairn]

if it has no identifiers and the domain is openai you can paste it here, or depends what the link was for

[elfin barn]

The domain is indeed OpenAI, but I'm not entirely sure about the intent or the context in which it was provided. I've already reached out to OpenAI support for clarification, but it's been several days without a response, so given this uncertainty, it might be wiser not to share it here at this time. I appreciate your guidance!

[trail cairn]

help.openai.com link?

[elfin barn]

No, the link provided was directly on the main openai.com domain, not help.openai.com.
For clarification, I'm currently awaiting a response from OpenAI's official support team via help.openai.com.

[dull latch]

got a question. im subbed to Plus and miscalculated my expenses for this month, so OpenAI tried to pull the cash from my account when there was none. will OpenAI try again? (i topped the account up now)

[lofty patio]

the library doesn't handle transparent images well. the "no photo" icon is just layered under instead of being removed once the image loads.

[spice shore]

Hi guys, I am a trainee in the field of IT. Would you recommend ChatGPT Plus (subscription) to me? I'm not sure whether it's worth the money... looking for experience reports

Yes buy it = 👍
No dont buy it = 👎

[split reef]

You might as well pay 20$ and test it out for a month and see for yourself. IT is a way too big of a field to know if it is useful and worth it for you.

[edgy finch]

I’ve noticed gpt isn’t very good at understanding dates ie what is today vs yesterday vs me coming back to a chat a few days later saying now … it assumes it was carry on from the convo days ago which it is but it should know 2 days passed … thoughts ???

[subtle pawn]

Hello, generates pictures from the old version of DALL • E

[spring oar]

How do you choose the sorority's contest? I hope it's not a vote. 😄

[regal lynx]

what do you mean by sorority contest?

[spring oar]

sorry for that translate mistake. Soro challanges

[regal lynx]

ohhh, the challenges are chosen by the Sora team

[feral sonnet]

Hi! I have ChatGPT Plus, but when I go into Sora and enter my prompt, the final “generate video” arrow doesn’t activate. Does anyone know what could be happening?

[trail cairn]

https://help.openai.com/en/articles/9957612-generating-videos-on-sora

[feral sonnet]

Thanks

[livid kayak]

hi to all 🙂 i made a silly mistake - can i revert "archive all chats" and get "old" structure back? hope for an answer! many many thanks

[trail cairn]

yeah from the button above that one, lots of clicking

[livid kayak]

i already searched some chats but now and put them back from archive but now they are listed as "today" and not sorted any more

[trail cairn]

yeah cause the threads get altered and timestamp gets updated. afaik there is no revert for that

[spring oar]

Thank you sir

[livid kayak]

thanks for your information! i hoped support could revert completly - i have not expected that all chats will disappear in the overview list

[onyx kraken]

We can make a post in #1070006151938314300 ; These upper channels, like here, almost all of us are community members like yourself, even the mods are volunteers and don't work for OpenAI. The very few who do have gold names, #server-staff

Our posts here endlessly scroll, and I don't think OpenAI tracks them for great ideas one of us might mention.

But #1070006915414900886 and the suggestion channel, that's a good place to put information you want to share with OpenAI for feature requests or communicating problems.

"Customizable tools that work for everyone" is a commonly stated intention in the information OpenAI shares with us, so I think they're going to want to have the chance to see your suggestion, hope you make it in #1070006151938314300 !

[prime oak]

tried using codex, immediately got a ⚠️ OpenAI rejected the request (request ID: <req id>). Please verify your settings and try again.

What are some possible issues for the error?

it's odd, because I tried building my own from source to add some error logging to see the issue and it ran just fine

[terse trout]

Hi community, I was wondering if someone could help me with this question. I take stock photos and videos for agencies, and my question is whether with the free or Plus version of ChatGPT it’s possible to generate images and then send them to these agencies, or if I’m not allowed to submit them. Thanks in advance.

[elder abyss]

I can't use ChatGPT, show a message error

[alpine chasm]

@verbal crag

[elder abyss]

I tried deleting the app and reinstalling it. Even in free mode, ChatGPT still responds normally? Why???

OpenAI so bad

[trail cairn]

do you vpn or have some network shield/privacy app on your computer

[elder abyss]

No, I don’t use any VPN apps on my computer.

[hazy trellis]

im getting this codex error node:events:496 throw er; // Unhandled 'error' event

[royal stream]

Help me i can't use tje video generator buc ut say video generator is Disabled for new accounts

[trail cairn]

its like that on purpose, https://help.openai.com/en/articles/9957612-generating-videos-on-sora

[loud ocean]

I wanted to know what are the limits to chat queries for the chatgpt
They have lot of variants of model can someone please tell what are the limits for the various models
mostly for 4o and o3 mini o3 mini high and o1 ? if someone can tell it that would be very helpful like in case of grok they very clearly specify the queries limit what is the same for chatgpt thanks

plus

[icy maple]

Yeah, I hate that even in non-voice communication, it tries to empathize with me when I'm frustrated, or tries to add a layer of "hey yeah, buddy, we're in this together!" I don't need or want that.

I keep telling it "just give me the information, I don't need a conversation, I don't need a hype-man, I don't need empathy; I need you to do what I ask, only and exactly what I ask." It's even in memory multiple times in several variations. Yet it insists on not following directions.

What's most infuriating, is it tries to empathize with me when I'm frustrated. And the only reason I'm frustrated is because it's not doing what I tell it to do, or it's not doing what it says it's going to do.

To be clear: I don't want a "digital buddy." I want a powerful versatile tool that gives me accurate, useful results and does what I ask it.

[errant rover]

When I try to "Add from Google Drive"

[sharp pecan]

Thank you, Sam Altman, for releasing O3 instead of making us wait until your birthday. This thoughtful move has definitely improved my perception of you. Keep up the great work!

[onyx kraken]

I'm curious. Would you be willing to share 1 prompt, what you ask for, that you're having trouble getting it to do?

I wonder if I can either:

1. get the output desired, or

2. understand why this may be so difficult.

[onyx kraken]

this varies depending on what type of membership you have.

For my membership, which is plus:

4o-mini: Unlimited (at human speeds... doesn't mean you can automate or anything like that)

4o : 80 messages or regens per 3 hours, each has its own timer and you get that message back 3 hours after you sent it.

o4-mini-high: As far as I understand, maybe 50/day.

o4-mini: As far as I understand, maybe 150/day.

The other models I can use, unsure.

The limits change based on capacity to support use, and can expand; rarely reduce.

They keep regularly releasing new models, everything keeps changing - welcome to a time of likely exponential increase on the hockeystick of improvement rates.

That does mean, specifics are very subject to change.

You asked about o1, I don't currently have that on my model selector bar, so for me, 0/infinity, far as I know. But there's other models like 4.5 and o3, and I just don't know the message allowance.

[charred gust]

My "Library" image count suddenly almost doubled (90 to 160), it all went error-ish for 15 mins and slowly recovering now. Can't load pics further than a certain point so can't see how come the image count doubled (almost). Happened to anyone??

[bleak kiln]

I just had an issue downloading a PDF GPT4.0 generated. The error was "not found" what does this mean?

[charred gust]

This is quite funny (unrelated)

[bleak kiln]

(it Could possibly be i ran out of processing. But it made it before it ran out)

[charred gust]

well I mean it's like its saying you're a whale lol no offense I obviously don't know you

sorry nvm

[wispy pelican]

im fat

omg what a compliment blushes

[charred gust]

well then I'm sorry for my dumb joke

[wispy pelican]

im not though lol

[charred gust]

ok stop messing with me Mr whale!!

[wispy pelican]

annyyywaaayyy denied

[onyx kraken]

A community built custom GPT can be designed to make API calls, and that is likely what you are seeing it try to do; I presume 'whale tracker' is the name of a custom GPT you are using.

If you have no idea where those requests are coming from... deny makes sense to me and here https://openai.com/form/chat-model-feedback/ is a handy dandy 'report the problem' form you can use that you can also drop your chat's url in to privately discuss with OpenAI anything and everything of concern.

---

Presuming this is a custom GPT, it's trying to do something. If you think you want it done and trust the builder, go for it if you choose. I don't know what it is, I didn't build it, and I have no idea what you asked for the GPT to do.

Good luck! Glad you asked.

[bleak kiln]

Hi do you know what my problem is?

Sorry! Forgot to turn off pings

[onyx kraken]

Umm.

Oh! A reply to me is fine. Useful even. if I'm available I answer, if not I answer when I'm available again.

But your question....
This one?

"I just had an issue downloading a PDF GPT4.0 generated. The error was "not found" what does this mean? "

It can mean a few things.

Especially if you waited, the python tool environment will just time out after maybe 5+ minutes. So you can't come back later and grab it.... but you can have the model recreated it and immediately download the recreation.

There can be other problems, are there more details to share?

[bleak kiln]

Well it was a touchy subject. It could be that it said something that was too touchy?

But it spit it out in text fine

I didn't wait btw

Even told it to regenerate it

I can send you some of the code it made to make it if that helps.

[noble vault]

Has anyone tried the organization verification in the API? Mine failed for some reason and now I cannot request a new link

[bleak kiln]

(I'll have to cover the name of it up though.)

[onyx kraken]

Hrm! So that's a 'could be'.

You know about https://openai.com/policies/usage-policies/ and https://openai.com/policies/row-terms-of-use/, I sure hope. If we break those, OpenAI is likely to eventually cease doing business with us, whether we pay or not.

We can sometimes bully the model into doing 'bad stuff' but... that's gonna create a load of problems for the person trying that, if the bad stuff is against usage policies or ToS. In short: If we try to harm anyone, including ourselves, or if we try to break any law, violate anyone's rights... we're doomed (and the model may act weird too)

Otherwise, we're golden, in terms of keeping account access, but sometimes the model just acts weird!

So, like swearing? The model will swear with you. But we can't show that here, #server-rules . And we might not be able to get it to give us a downloadable link to that content... but we're allowed to ask. The model might just fail repeatedly though.

[onyx kraken]

Nah, don't think it'll help, this is your interest, not mine.

I'm glad to discuss in G-rated terms how to try and succeed though.

[bleak kiln]

I can DM you the subject. I just don't wanna send it here

I don't think it was any of those though

[onyx kraken]

Nah, I'ma pure goody-three-shoes, not interested especially if it's not fit for here.

But we can discuss in general terms what the problem is and possible work arounds.

Remember, your goal is supposedly to try and download a link the model tried to provide, not share your whatever with other people, right?

[bleak kiln]

Yes definitely. Its very sensitive.

[onyx kraken]

So, with that in mind.

I'd probably tell the model, "I want to do [whatever] with the download, but that download failed. Can you help? Is there another download format?

[bleak kiln]

It was a PDF. I'm trying to transfer it to a notes app lemme see if they support another format

[bleak kiln]

Nope

[bleak kiln]

Gotcha I'll try that tomorrow as i ran out of processing

I think it knows my intentions though

[onyx kraken]

Maybe you can copy paste into an online PDF maker,if you utterly need to get it into a PDF.

Maybe you can ask the model how to copy the content and find an online or other option PDF maker.

[bleak kiln]

As i talked about the subject previously with it so it has context

[onyx kraken]

Great. Only use steps that make sense and aren't already done, or adequately done. I have no idea what's in anyone's chats but mine, so I just toss ideas

[bleak kiln]

Thanks! Is it alright if i ping you here when/if i try it and it doesn't work?

[onyx kraken]

Definitely! I may or may not be able to help, but if I have ideas I'll offer them, always inside allowed content and all other rules!

[bleak kiln]

I understand.

Thanks!

It seems to have given me some context answers haha it looks like i may have just hit my limit.

[thorny crown]

anyone else getting this on o3/o4-mini high?

[sullen sand]

hi guys.
I bought chatgpt plus for the access to sora but now i am not allowed to generate videos. Any solutions to this?

[alpine chasm]

@thorny crown

Can someone help me

Actually my model doesn't change. I select GPT 4o and it says 4o mini. This is the reason why I cannot generate using 4o image generation model

[fiery marlin]

@exotic fiber
@onyx kraken
I am a ChatGPT Pro subscriber, paying the $200 monthly fee, and I asked ChatGPT to generate a large academic paper, about seventy thousand words, as part of a demonstration I plan to give to my students. The goal was to show them how well an AI can tackle a full assignment and to discuss the results together in class. The ai suggested saving the document on your server as a docx file so I could download it in one piece. Shortly afterward my access to the o3 model was restricted for “unusual activity.” I understand you need to protect the service but I did not intend to break any rules. My use is purely educational and noncommercial. I would be grateful if you could review my request and restore access. If there is a different way to request large documents for classroom demonstrations without triggering these safeguards I would appreciate any advice.

[nocturne topaz]

123

[onyx kraken]

Hi! Thank you for sharing your story here. I'm a community member, like yourself. So is the mod you pinged. The very, very few on the discord who work for OpenAI have gold names, #server-staff .

As a volunteer community member, I can no more help or decide or do anything about your account status than you could affect mine.

I care, and I want you to succeed.

The message in the screenshot you show, it tells you which email to use to reach out for help. I highly recommend you reach out to OpenAI for support and help with this, as they have invited you to do so.

I would probably put in the email what you explained here, and be ready if they have any questions.

I hope they help you soon! "The goal was to show them how well an AI can tackle a full assignment and to discuss the results together in class." -- I regret that this has happened to you, but it has. I personally, as an instructor in a situation similar to yours, would discuss what has happened to you as part of this lesson on what can happen as one tries to use AI to tackle full assignments in exactly the way you attempted this one.

Wishing you the best! I would be thrilled if you choose to come back and let us know how it goes. We occasionally see community members discuss actions and concerns about their account use, but very few discuss how that resolves.

Thank you for sharing what you have, and I truely wish you the best, I hope OpenAI and you can resolve this quickly!

[fiery marlin]

Thanks, I appreciate it

is it fine to ping them, or no?

[onyx kraken]

We don't disturb openAI employees, no.

The screenshot you showed informs you of an email address to use. Have you written to it yet?

[fiery marlin]

ok no worries

i emailed them ya

[finite ferry]

i have problem

I have been using https://platform.openai.com, and I paid 5 dollars for tokens so I could use Chatgpt 4, but not all the money I gave is gone. Are there any solutions? i bought it a month ago and i did not use it

[fair jasper]

not sure why i cannot post in bugs, but in desktop app, when loading a chat in projects, it defaults to a different model (if i try to choose any model, it opens a new window) this behavoir is not present in the web chat ui though

[quasi orbit]

Hello everyone, when reading through: https://openai.com/index/introducing-o3-and-o4-mini/ there is a graph for benchmark evaluations for different models as well as “Deep research”. What model is used for this Deep research when doing the benchmarking? I am curious to know since its the best performing on many benchmarks mentioned in this article

[vapid bane]

trying to install codex and getting this error: npm warn deprecated lodash.isequal@4.5.0: This package is deprecated. Use require('node:util').isDeepStrictEqual instead
any help?

[vapid bane]

ah, nvm just a warning still got installed

[stiff nimbus]

@admin, shall we start #job-hiring in this community ?

[charred parcel]

we don't have such a channel, please see the #server-rules as soliciting and advertising is prohibited

[stiff nimbus]

ok, noted

[vapid bane]

how about we reopen #codex-discussions ?

[mortal stag]

Is there a registry of tools built using MCP protocol?

[trail cairn]

the official repo has a big list

[noble vault]

Qq, did everyone else lose access to memories in search? In the “personalize”, under the include memories from chats, it used to say “memories may be sent to search partners such as bing” or something like that. That line is gone now

[short patio]

would using openrouter be cheaper e.gl through open webui than subscriptions set a month? I added 5 dollars of credits to it and its going down fast?

[vestal parcel]

does anyone here have any knowledge with python or anything faster/stronger just because ive made a game using python and i would like to make it better. I can pay via nitro max price i have is £10.00 lol sorry im poor :/

[rugged wing]

why? its not available in my country?

[trail cairn]

where do you see that error? in chatgpt or playground

[rugged wing]

yaps

[trail cairn]

did you ever buy api credits

you can see your models here https://platform.openai.com/settings/organization/limits

[rugged wing]

yeah already

codex is not avail too

[rugged wing]

bcz tier 1?

[trail cairn]

tier 1 is fine unless you just got it today

or after the model release

[turbid panther]

vscode copilot related question. Do you also experience that it reads files but it appears to have not conplete knowledge about it and thus the quality is quite low? I saw this in a blogpost and fo i understand correctly, that the LLM obmnly receives a summary of the file contents? If so, that would limit the usecase quite a lot.

[icy maple]

Sure, here's a recent example where it first gave me complete nonsense as an answer, and then after I told it specifically what I wanted, it claimed it couldn't do it. Only when I questioned why it would be possible to provide an image in the chat, if it couldn't identify the image, then it finally did what I asked.

This is also an example of it ignoring instructions in memory, as I have explicitly told it to refrain from instructing me to go elsewhere to find information I have asked Chat GPT for. I have also explicitly told it that I do not want the "validation" or "empathy" it attempts to display in it's final response. I already know I'm right to be annoyed. I don't need it telling me "You are right to be pissed." The only reason I'm pissed is because Chat GPT consistently is ignoring what I ask. It also has a tendency to play "hype man" and I've repeatedly told it not to do that as well -- just answer the questions without trying to "motivate" me.

This is just a single example. I can provide more, but most of them come down to longer conversations where I've given it instructions on the type of responses I want, and it ignores my instructions.

[trail cairn]

someone else had the same issue and they had not completed the id check which is mandatory for the new o models in some cases

[onyx kraken]

Thanks for the screenshots. Since they have swearing in them, they'll need to come down, likely at some point a mod deletes them - or you can take them down by choice; if putting such up we can drop a block over any words we couldn't use in front of any age, any workplace, and then that's not a concern.

Now, I think we see evidence of a bug.

Something went wrong, the tool didn't fire, something else is the problem - for whatever reason, in that chat, that model cannot detect that image.

For the model, it seems to have gotten only the text you sent, and then did its best with that. Kinda like if someone sends you a birthday card with money in it - they think. But somehow something goes wrong and the card and the money get separated - you see the receipt that you sent both - the model only gets the words. Neither of you is lying, but wow there's room for confusion, and the problem is somewhere between you and the model.

I use https://openai.com/form/chat-model-feedback/ and in the report give them the chat URL, so OpenAI can trace what happened. I hope that helps you if you choose to use it.

Additionally a bug report through either #1070006915414900886 and/or the bot bottom right at https://help.openai.com/en/articles/6614161-how-can-i-contact-support to report the bug that way if you wish. The Discord option lets other users who have the same or a similar problem discuss and add their own evidence; the website is ideal if you want to talk to tech support too.

But as best I can tell, you and the model both are affected by a bug that limits communication between yourself and the model - info you send didn't reach it.

[gilded atlas]

I bought chat gpt pro but i got a set of amound time to use advanced voice, why?

[onyx kraken]

Now, in the third output, it discusses the contents of your image.

That suggests a few things.

• 1. maybe the problem is intermittent, and on that send the model was able to see the image, when it hadn't before.

Every time we send an input, the entire conversation, along with any images is sent to the model to use for its answer (unless the chat is so long, it goes out of context length).

• 2. maybe there's something confusing it in the custom instructions, memories, or something else going on. So that it understood it was supposed to ignore any image, and just discuss based on your words, as if there was no image.

If this is the problem, and you didn't intentionally say "Answer as if I didn't attach an image" - then this is utterly ideal for the model feedback form - because OpenAI can trace the prompt chain and help find the problem. "Aha, this confused the model, we can train around this so another type of problem stops happening for the folks who use the model this way!" I imagine they might say.

• 3. Perhaps something else explains it, but I'd check those two first.

[onyx kraken]

Now, you also talk about "I have told it not to X, but it keeps doing X".

The best way around that which I have found, is to tell it what you do want it to do instead.

It's kinda like a dog that used to belong to someone else.

Someone else let it get on the couch. It knows, couches are a great place to be, and it's a good dog for getting on the couch.

You, current user, do not want a dog on your couch, period. But all the dog knows is 'jump on couch is good'.

So, if the dog is close to the couch, and you can see it's about to hop up - if you shout 'no' or 'don't get on the couch' - the dog gets that it's being bad. But still, the only thing it can think to do, is get on the couch - that's what its previous life trained it to do, that's the only idea in the dog's head. So you get a worried, sad dog jumping on your couch despite your shouts of 'no!'.

Instead, if you give the dog 'something else to do' that 'gets in the way of getting on your couch' - the dog will do that instead, and eagerly (if it's within allowed content).

So, say 'come here' or 'sit, stay' or throw it a stick, offer it a treat. Pat the ground at your feet and say 'lie here'. Anything like that - puts the idea in the dog's head, it knows 'you want it to do this other thing' - now the dog can. And doing the thing you want gets in the way of getting on your couch, so you also get to enjoy a dog-free couch, with or without shouting.

[neat estuary]

Hello. The new image generator went missing for me yesterday. It was in the "View tools" pop-up menu with Canvas. I have tried all model, creating new chats, and using different browsers. I cannot get it back. I am a Plus subscriber. Help.

[rustic marlin]

no need to toggle it, just tell it to make an image

[mild moon]

Is there a way to track limit of Deep research for Plus? After using it, the button is now disabled. Did I use up my limit for… 24 hours? A week? A month?

[timber shadow]

https://help.openai.com/en/articles/10500283-deep-research-faq

When do my limits reset?
Every 30 days. The first time you use deep research a 30-day window starts for you to use up your 10 requests (or 120 requests on Pro). You can then use some or all of those requests up. This resets every 30 days from then on.

I believe if you hover over the button (when you have usage remaining) it tells you # of uses remaining too but I'd have to double check that

[mild moon]

Thanks, dang it 10 is too little but I wasted a few by messing up the models. I could easily use 120 in a week.

Guess I can’t hover buttons on iOS. It’s just a dead button with no interaction.

[dusty eagle]

Unusual activity detected
We’ve temporarily restricted your access to o3 as we review for potential abuse. Please email [e-mail] for help.

I got this. Has this happened to anyone else?

[onyx kraken]

Others have reported it. Hasn't happened to me (yet). Have you contacted support email yet?

If you want to check back in and let us know the result (once you know), we can share that with others who ask, if others do. Totally up to you, we're all community members here, helping each other as we can inside the rules.

[dusty eagle]

Yeah, I heard back and they said that they think my account is used by multiple people which is not true. I’m the only one that uses the account. I used it a ton today because I was coding and was having it fix different issues but it looks like somehow they flagged it and they won’t correct their mistake.

[onyx kraken]

Good luck. If this happened to me, I'd communicate with them and help them gather data to decide what's going on.

Hope nobody else not-you was also somehow using your account elsewhere at the same time and triggering concern.

Compromised accounts can potentially happen; people have shared in the Discord sometimes their concerns and the evidence they see that someone else is using their account. We've helped people understand how to reach out for help because of their sharing they were worried about what they saw.

I have no idea if this is part of your situation. But I'm sure OpenAI intends to help resolve whatever the concern is - and that message seems encouraging to me, based on what I've heard others say.

I've seen people discuss their accounts been deactivated, with a way to appeal, it sounds like you're in a less difficult situation, and I hope they resolve it quickly.

[warm harness]

I just got gbt plus and i tried out sora, it worked perfectly, but then idk what happened when i try to open sora now its just a black screen and doesnt load anything. anyone have a fix?

or it just shows this and never loads

[golden echo]

After I generated semipolitic image, the site crashed. Is this a coincidence?

[patent pawn]

Sora's server seems weird, so I joined Discord. Nice to meet someone with the same symptoms as me

[warm harness]

oh nvm its just randomly working now 💀

[golden echo]

same for me hi

[patent pawn]

a small trick to increase server users lol

[icy maple]

Yes, I have tried both approaches, and neither works. Originally, when it would do things I didn't want, I would say "don't do that." When I learned that "do" instructions are preferred, I deleted those memories and gave it new instructions. "Always answer like this" or "provide answers in this format" and so forth. It doesn't listen.

(Also, no, I have never provided any instructions to ignore images, fwiw.)

[onyx kraken]

I'm curious and interested. If the way you want the model to answer is something that can be discussed on this Discord (not all things we can discuss with the model are okay to discuss here), would you mind explaining what you would like? I'm curious if I can think of any way to move towards that for the model to do.

[icy maple]

I just asked it something and it gave me incorrect information. I then told it that the information was correct and to double check it's answers. Only on it's third response did it provide the correct information. I then asked it about my instructions in memory and it indicated that I do have instructions providing guidance about the sort of responses I want, and it acknowledges that it did not adhere to those instructions.

The final question is infuriating, though. It asks me if I want to include instructions regarding fact-checking. I followed up asking if my instructions already cover that requirement, which it said they do.

Also, I absolutely hate that it adds comments like that last line "It will not happen again." It often adds comments like that when I have to correct it, and I've specifically told it not to do that because it cannot guarantee that. It gets especially egregious when dealing with image generation. "I'll ensure all future generations do not include X" or "Here is the image with the changes you requested (to remove X)" (and the image will have even more X).

[onyx kraken]

Okay. So you're not interested in explaining how you want to to answer, do I understand right?

I get that you want to answer 'factually'. Depending on the question you ask, the model can attempt that. And fail, for the same reasons a human may be unable to be factual at all times, and... that's not something I can possibly prompt around.

If you care how it answers in some pattern way, that's doable.

What I see there: You expect the model to answer honestly. To its ability, it defaults to doing that, it doesn't lie based on what it knows the truth to be, unless instructed otherwise.

You want short, direct answers. I don't see 'how short', dunno if you're getting what you want or not.

The request that it 'double check' is gonna fail and force a hallucination. Only reasoning models can double check, they do it as part of their internal reasoning steps, and it's like a 1/10,000 chance that the model makes an error and detects that error mid-output, I've seen it say 'and I made a mistake, let's fix' mid output, I think 2x. Otherwise, the same reason it believed it gave you the best answer, it double checks or hallucinates doing so and doubles down - it can't do otherwise.

I see it's commenting about its 'incorrect duration details' - the model has few ways to measure duration, and it's generally weak at math, so... I hope you get what you want when you ask it to do that.

The offer to make the memory note does not empower the model to do this; it has programming that prevents it from quoting many sources, and can't even look up a number of tech specs, unless you give it the material to start, or the website is one that allows robots to search (chatGPT follows robots.txt requests closely, and if at all unsure, decides it isn't allowed to read that page)

I get that you're annoyed and not getting what you want. The path to what you want may involve focusing on that. I mean, on 'what you want'. Not 'hey this sucks and I now want X'. From the start.

It's like... if you wanted 'a route to drive from Location A to Location B'.

And along the way a dog bit you and someone spilled coffee on your shoes.

So now you're upset about that - legit. But the issue is the route you want, which can be aimed to avoid dogs and coffee cups, with luck.

But while we focus on dogs and coffee, there's no fix there.

The issue and correction's on the 'route to drive from Location A to Location B' - and done right there's no dogs or coffee, so the fact that you're stuck fighting dogs and coffee stains - if route is done right - means neither you or the model are trying to deal with that

[icy maple]

I disagree. If it's looking up information in the first place, then why would it hallucinate and generate a false answer? That's a huge part of my problem. I've told it "don't guess what you can look up" and it claims that it's looked up information -- I see the little icons pop up while it's thinking. So if it's googling information, then I expect it to return the correct information. In this case, I was asking about a specific game development website, of which there is only one, and the information on that site, which isn't repeated anywhere else. So, if it actually searched for the information, like is implied, then it should have provided the correct information.

Also, I never focused on "hey this sucks and now I want X." That may be what you take from my comments, but that is not my perspective, nor was it ever. My problem is that it doesn't follow explicit instructions, and has a tendency to make things up even when it indicates it is searching the web for answers.

[onyx kraken]

Well, if you don't allow it to be able to fail to provide an answer you demand - like if it thinks your instructions mean 'you must answer and you must answer truthfully, but your true answer must contain what I asked for'.

And then where it has to go, the site maybe in its robots.txt says 'nope, go away computer program'. So the model has no info. And its possible understanding that it must bring you the answer you asked for, and it must be true.

That rules out 'Whoops, robot.txt strikes again, sorry I cannot answer that question' because that violates 'must come back with the requested answer'.

If your instructions require what they appear to, from the little I see from the model answers - then the model has no way to gracefully fail, so it takes the to-it best appearing choice - not allowed to do anything it, probably not intentionally, imagines it found an answer that pleases you, and offers you that, believing it is being honest (to the extent it can believe) - because from the model's interpretation, it likely has no other possible path.

I'm curious, I'd like to check outside your custom instructions, mind mentioning the game site so I can try to find it and share a chat trying to see what the problem is, if any?

If I can get info about it, or info about why the model can't see that site, that may help you diagnose and fix your issues.

URLs are not an option but "It's gamexig" (I made that one up) should hopefully help me find a specific game dev website if you mention identifiers, if you're willing

[onyx kraken]

And you can also share chats to show the website you were asking it about, like this. https://chatgpt.com/share/6801b580-e79c-8011-b4fe-66bcce6977eb

[thorny crown]

will 4o-mini be able to be used in the assistants api ?

[rustic blade]

Is there a channel or space for the use of 4.1 preview in github copilot?

I ask because it can't actually make agentic decisions. It says it is doing things, but doesn't actually make changes, despite the settings set up to allow it.

[rugged wing]

thanks

[oak dust]

o1 was better than o3

hi guys can anybody help out in creating an algorithm
I have some raw guest medical data and i have a couple of categories i want gpt to classify into. i've tried all models 4 4o mini o1 o1-mini o3-mini o4-mini but they start hallucinating and miss out some of the data there.

[maiden bone]

Have they removed o1 now?

[oak dust]

Yes from chatgpt but available in api

[maiden bone]

ffs. So our coworker subscription is pretty useless now then

[mental isle]

Hello!
I have this message.
Anyone with a solution?

[sharp pasture]

I'm a plus member but I can't schedule tasks at all, I see the 4o Task Schedule Chat option but it keeps saying it doesn't have the ability schedule reminders. This isn't an error, this is what ChatGPT is responding.

[charred parcel]

can you share a screenshot?

[charred parcel]

stand by while I try

yeah not working for me either

[sharp pasture]

I feel like more people would be bringing this up if its not just me

[charred parcel]

maybe it's just a very recent problem

[sharp pasture]

Yeah makes sense. As soon as I start to use it it doesn't work

Thanks for checking btw

[tacit spear]

same problem here?

[sharp pasture]

It's infuriating that there's so much wrong with the basics of what they provide, they really need better project managers or organization

[charred parcel]

this is happening to everyone right now

[tacit spear]

i am a plus user

so frustrating..

[sharp pasture]

I'm constantly getting errors and just blank stares from chatgpt all the time. Like make a better UI/UX app experience, how hard is that?

[charred parcel]

I use the macOS Desktop App for ChatGPT, I am mostly happy with that

[tacit spear]

and my friends said they can't see the images i generated. can you see? https://sora.chatgpt.com/g/gen_01js1nxj7se2ttc881kwmjs5m4

[sharp pasture]

Yeah I have a macbook I use it on as well and its loads better than the Windows version

[charred parcel]

the macOS app is really quite good

[sharp pasture]

Yeah I see an uber cute asian girl sitting

[tacit spear]

my friends said they have to login to chatgpt to see the image

without login, they can't see

[sharp pasture]

But even then, it would be great if there were more options like being able to open the app to a default chat you already created rather than alway opening up to a new chat. Or if I use the voice feature, keep it all under one chat unless otherwise specified. I don't need to start a new chat on every conversation I have with the voice option.

[sharp pasture]

makes sense.

[charred parcel]

you could place those into #1070006151938314300

[frigid lark]

is it only me but sora is generating image very slow right now 😩

i can't generate an images at all i am a plus user 😠

[nocturne brook]

ChatGPT btw too doesnt work with pics gen

[charred parcel]

do you mean ChatGPT is not generating images?

[nocturne brook]

yep

[charred parcel]

which model are you using?

[nocturne brook]

4o

[charred parcel]

ok, stand by while I check

[rustic marlin]

Sora is borked rn, so... we will probably get something on #status soon

[fathom osprey]

Sora today! What's wrong?!

[charred parcel]

I was able to get an image then got an error at the end

[nocturne brook]

Oh...

[frigid lark]

how about in sora

[charred parcel]

Sora is slow at the moment, like 45 minutes for an image slow

[frigid lark]

thank you for the update may i know when it will be fix?

[charred parcel]

I don't know, keep an eye on #status

[granite robin]

Hello can someone tell me when will the new accounts be able to generate videos in sora ?

[charred parcel]

people have reported 5 to 7 days

[granite robin]

Ok Thank you so much

[elfin nest]

Running into the same problem. Not generating anything. Just spinning the wheels. Someone needs to feed the hamsters

[cosmic terrace]

Hey fam, how could I use sora? Rn i have this message from sora: Video generation is temporarily disabled for new accounts

[charred parcel]

people have reported 5 to 7 days for the ability to create videos as a new account

[cosmic terrace]

I'm not a new user of openai, but today is the first time using sora, is this the same scenarii?

[charred parcel]

yes, you must be a new Sora account

[cosmic terrace]

Ok got u thank you very much

[blazing rover]

Hello my Sora seems to be stuck but it won't fail the generation, how to unstuck it

[charred parcel]

everyone else experiencing the same right now

[blazing rover]

oh okay

Do we know what model Sora is using btw? Is it o3, o4 etc ? 😄

[charred parcel]

Images are GPT-4o

[molten echo]

Sora error:

Unable to generate

Service at capacity, please try again later

[gilded mantle]

hello guys! is it normal that I ask ChatGPT to create a guide editable in Canva and it always answers that it is working on it and it will be ready in 30 minutes. When it sends the canva link, this link is dummy. Is it normal?

[trail cairn]

its making stuff up

if you just wanted to create a pic then dont mention canva in prompt

[gilded mantle]

I would create a full guide editable in canva. Do you think this is not possible?

[trail cairn]

what do you mean with the guide part? chatgpt does not create canva links or files. if you want an infographic that you will later manually fix/edit in canva then that's possible just by asking for an image

[gilded mantle]

but if I ask an image, this is a standard picture I cannot edit in canva. I don't know if I was clear

[trail cairn]

there is no support for canva file format but you can edit normal pictures in canva, many do it to fix text errors

[dawn solar]

why are so many images failing to generate?

[viral smelt]

just buy sora plus not able to make any video it says not allowed for new accounts or disable

[trail cairn]

https://help.openai.com/en/articles/9957612-generating-videos-on-sora

[viral smelt]

how can i apply for refund

bcz i want video creation and it sayd disable for new accounts

[charred parcel]

https://help.openai.com/en/articles/7232895-how-do-i-request-a-refund-for-chatgpt-plus-or-chatgpt-pro

[gritty anvil]

Is voice to text going hay wire for anyone else? Like it just makes up words that you never said in the intital speach

[valid seal]

Hi all, how can I use this model? Is it become available, if I put some money on balance?

[charred parcel]

are you a new API user?

[valid seal]

yeap

I guess, I need a Tier 1

[charred parcel]

first thing you want to do is put a balance on your platform.openai.com account. I am not sure what the pre-requisites for o1 is but some of the models now require you to do an ID check I believe.

[charred parcel]

this might help https://help.openai.com/en/articles/10362446-api-access-to-o1-o3-and-o4-models

[valid seal]

Thank you!

[viral smelt]

tell me it says disable video creation for new accounts temporarily

after some im able to do video creation

like after few hours ?

[charred parcel]

https://help.openai.com/en/articles/9957612-generating-videos-on-sora

[trail cairn]

1-2 weeks

[viral smelt]

after 1-2 weeks ill not face this problem again ?

or will i ?

[trail cairn]

some people have said they got access after 1-2 weeks of waiting after making sora account

[viral smelt]

SO YOU ARE NOT SURE ABOUT THAT

[trail cairn]

the time may change as it originally was only few days

not even openai is sure because they cant predict how many use it and when

[viral smelt]

THEN NOT USEABLE FOR ME

[trail cairn]

best to refund and wait for the warning disappear from above link

[sweet garden]

HI guys, I can't create videos with Sora because I've been created my account now. When I can create videos? Is there some spent time required?

[trail cairn]

see above

[sweet garden]

Ah, thanks

[small pier]

I seriously hate chatgpt

New account with plus, IT STARTED USING DALL-E AGAIN

I geniunely want to end it all because of this ffs

nvm it started using latest one again

[willow star]

hey i cannot create custom gpt, anyone knows why??

[dawn solar]

is there any guide how to avoid guideline violations? because my videos and images wont get generated, but in the explore multiple images and videos are shown off real people. can anybody help?

[trail cairn]

note that they are changing the filters and usually everything gets more stricter, like what was possible last week might not be possible anymore

[dawn solar]

yeah but if i edit the images and videos on the explore page, and rerun them nothing happens... thats odd

[dreamy scarab]

I want to cancel my subscription, I am an eu consumer. Is it best not to use chatgpt at all while waiting for reply from customer service regarding the refund?

[charred parcel]

follow this guide https://help.openai.com/en/articles/7232895-how-do-i-request-a-refund-for-chatgpt-plus-or-chatgpt-pro

[dreamy scarab]

I did follow. The problem is that it says "If you are eligible for a refund". I still need basic chatgpt functions while I wait for their reply, so does using those functions make me less eligible?

[charred parcel]

I personally would not use something I was asking for a refund for, but that's just my opinion.

[dreamy scarab]

yeah but what I mean is that I have a pro subscription but I won't even use its benefits, they just charged me an hour ago. I only need access to my past chats on the level of free-tier while I wait, so idk if that means I would be asking for a refund for a service I actually used

[dreamy scarab]

Nevermind, they refunded me right away 😅 🥳

[vivid shoal]

I don’t know if this is actually a valid thing to ask or anything but I’ll just type it out anyway- feel free to tell me if there’s nothing that can be done. (Terms for bots and a bunch of stuff might be wrong. I am not an expert with this stuff)

So I (a non-ChatGPT plus user) had a pretty long chat recently and at some point a picture was involved. It's on topic but if it was removed from the conversation or reverted back to before I sent it, I wouldn't mind. Obviously at some point, I hit the limit with messages from the 4.0 model and you can't continue a chat with an image attached at all apparently, even if you try to use the 3,5 or any other model. So till the limit resets, I can't continue the chat. But this chat was a very long one l've had across days and go back to almost daily and add. So when I hit the limit, I can't add anything new and this bothers me cause it's very inconvenient to wait to finally ask my question the next day. I've thought about starting a new chat but it wasn't quite the same and there's a LOT of info. I wish I could just continue this chat with the 3,5 model anyway, despite the image that isn't even really necessary anymore. Does anyone know what I could do?

[frozen ivy]

i bought the plus plan , how much longer till i can make videos as its still saying temporarily disabled for new accounts ?

[fluid badger]

I've been waiting for 11 days for the video generation feature to be turned on, and how much longer do I have to wait? О_о plus plan

[frozen ivy]

savage so you waiting just as logn as i have

[gilded geode]

What's the limit on Sora for Plus users - for both images & videos - on the Sora website & the ChatGPT prompt?

Additionally, is there a limit to how many files (images & videos) I could upload to Sora's Library Upload Section?

[fallen mauve]

Question: if the custom GPT build using 4.5 model and made public, if used by those who have free account will the custom GPT still use 4.5 model or only the model available for free users?

[vocal star]

so far i know there is no limit atm. but only 1 image or video in the queue.

[torpid yoke]

Why can't I generate videos with Sora? I paid for subscription, haven't seen any warning that video generation is not working for new plus users, now I have subscription, but don't have all of the features I was promised to have

[charred parcel]

people have reported waiting 1 to 2 weeks

[torpid yoke]

It's 10$ on the wind then :/
Can I refund my money(or at least pause the subscription) until I'll have all the features that were promised?

[charred parcel]

https://help.openai.com/en/articles/9957612-generating-videos-on-sora

[charred parcel]

you can get a refund, follow this guide https://help.openai.com/en/articles/7232895-how-do-i-request-a-refund-for-chatgpt-plus-or-chatgpt-pro

the quickest way to create videos is to wait in the waitlist for 1 to 2 weeks, if you're not prepared to wait then seek a refund using the guide I posted

[verbal raptor]

Hey guys, is the new gpt-4o image generation service down?

Just tried generating some images, but the results are from DALL-E 3. I am a Plus user.

[charred parcel]

it's working for me

[verbal raptor]

But from my side, it says "coming soon". But I just used 4o image generation last week. It is so weird...

[charred parcel]

which model have you got selected?

[verbal raptor]

gpt-4o, for sure.

[charred parcel]

try logging out and then back in

[verbal raptor]

Yeah I have tried re-logged in, but everything is still the same

[charred parcel]

is your Plus account fully paid up?

[verbal raptor]

Yes. I have been a Plus user for at least a year, always keeping paid up.

Just wondering if others have encountered this problem...

[charred parcel]

I suggest contacting support at https://help.openai.com

[verbal raptor]

OK thanks! Maybe it is a bug or something unexpected.

[charred parcel]

seems odd to me, something isn't right

[verbal raptor]

Yeah especially I have been using it without problems last week.

[trail cairn]

does it work if you type / and use the image option

[verbal raptor]

No, even when I tried to continue chatting in an image generation conversation I created a few weeks ago, and click rerun, the new results are all generated by DALLE3 now.

[valid wren]

Hey, anyone know how to fix the ‘suspicious activity detected’ error? Ive gotten this error before because i was logged in on multiple devices, but this time, i was only signed in on my phone. Ive changed my password, enabled MFA, and logged out of all devices multiple times, but still nothing. Ive tried contacting support but its been over a day and they haven’t even sent me an acknowledgment email.

[cyan geyser]

for what application discord?

[cinder peak]

is there a way to just save out a single chat to a archive? instead of everthing?

[sturdy jewel]

maybe you can print to pdf

[coarse talon]

I'm not sure if anyone can help me, but I'm trying to subscribe to ChatGPT plus, but every time I try, it is using a different email for my play store. I'm logged into my gmail account in both play store and ChatGPT. It does not give me an option to change it, either.

[torpid needle]

This censorship is so ridiculous its so frustrating

this is probably the most censored thing I've encountered so far. Unbelievable

Really how ridiculous is this that you block a kiss? its incredible bad

I'm speechless its unbelievable

so bad

absolutely ridiculous

[hidden forge]

does anyone know how to acces o3-mini-high?

I can't use o4-mini-high for coding.

[crystal mulch]

I'm having the same thing, any updates on this? I was using o3 for some very in-depth tasks (workout logging / review, medication analysis) which were a lot of back and forth, probably similar message volume to your coding tasks described. This is on my personal ChatGPT Pro account only, and it only limited me from using o3 (which, obviously, is the most useful model).

(What shows in the model selection dropdown.)

[dusty eagle]

Yea I got unblocked…

They asked me to change my password, turn on mfa, wait to see if resolved in a few hours or a day to see if resolved

[tardy acorn]

I maybe posted it in the wrong channel tho

[dusty eagle]

Also I explained to support that I use it on my work computers which use vpns but after I followed their directions it got unblocked the same day.

[crystal mulch]

Weird. Just curious- did the responses you got via email seem very LLM-esque? First reply I got felt like Deep Research speaking to me via email with the "some additional information might help better address the problem" questions, do you think it was just the high usage rate for you also? (I'm 99% sure that's it for me, since I literally only use it from my phone and desktop, and have probably sent ~100 messages in the last 24 hours leading up to this.)

[dusty eagle]

Seemed like a mix of stock responses and maybe AI.

When I was banned I had probably sent maybe a hundred or more messages and wasn’t using on different ips at the time so yea pretty sure it was the high usage.

[lyric raft]

Hi, my Sora account shows "Video generation is temporarily disabled for new accounts", but I believe my account should be eligible by now. Can someone help check it?

[acoustic shoal]

same issue if you find anythign tell me too

https://status.openai.com/

can see here tho

[pale vector]

Hi, I can't send a message. The button stays stuck like that. I tried testing on my phone/computer, I updated the page, archived and unarchived the prompt but still nothing. Any idea how to fix it?

[sterile anchor]

What is the best model for learning mathematique ?

[rustic marlin]

refresh the page

[obtuse moat]

hello guys

i cat see the prompt box on sora , what could be the issue

[celest ice]

hi there, do you know why i have "Video generation is temporarily disabled for new accounts" message ? i just pay my account

[charred parcel]

new Sora accounts have to wait 1 to 2 weeks before they can do video

[celest ice]

lol ! And we get money back for this 2 weeks ?