community-help

[trail cairn]

if it came from the modmail that sits top of the user list here its not

[green summit]

If you got modmail, and you actually participated in a challenge then you can expect it not being a scam.

[fallow vector]

Not ideal but may just regnerate the tts for now. Really need to get this shipped so it will have to do for now. Testeted your approach and it works sometimes but more often than not I get no response back.

[lyric monolith]

got it

[lyric monolith]

got it

[green summit]

Nice, well gz!

[lyric monolith]

Thanks😀

[tall raft]

what are you trying to accomplish?

I would like to have more info about the problem and what you are trying to build to think of a better way to tackle this

[fallow vector]

Thanks for your reply! I want to create an audio recording of the entire conversation with both the user and assistant audio

Currently I have user audio solved, my original plan was to fetch the pc16m audio, covert to waw and insert it into the recorded user audio file

This is being done on a react native app using WebRTC

[tall raft]

I hit the same issue with the Realtime API where conversation.item.retrieve didn’t give me audio for the first assistant response (even after waiting ~1000 ms), while it worked fine for all subsequent ones.

The fix for me was basically two things:

1. Enable PCM16 before the first reply is generated

Make sure the session is configured with output_audio_format: "pcm16" from the very start (in the session token or immediately via session.update before sending any user audio):

{
  "model": "gpt-realtime",
  "voice": "alloy",
  "modalities": ["audio", "text"],
  "input_audio_format": "pcm16",
  "output_audio_format": "pcm16",
  "input_audio_transcription": {
    "model": "gpt-4o-transcribe"
  }
}

If the first assistant reply is created before this is set, there’s no PCM16 stored for that item, so conversation.item.retrieve will never return audio for that first turn (you’ll only have the WebRTC Opus stream).

[tall raft]

2. Call conversation.item.retrieve after response.done instead of using a fixed timeout

Instead of setTimeout(1000) and then calling conversation.item.retrieve, listen for the response.done event, grab the assistant item id, and then retrieve it:

dataChannel.onmessage = (event) => {
  const msg = JSON.parse(event.data);

  // 1) Assistant finished generating a reply
  if (msg.type === "response.done") {
    for (const output of msg.response.output) {
      if (output.role === "assistant") {
        const itemId = output.id;

        // Ask for the full item (including audio)
        dataChannel.send(JSON.stringify({
          type: "conversation.item.retrieve",
          item_id: itemId
        }));
      }
    }
  }

  // 2) Server returns the full item
  if (msg.type === "conversation.item.retrieved") {
    const content = msg.item.content?.[0];

    if (content?.type === "output_audio" && content.audio) {
      // content.audio is base64-encoded PCM16
      // Decode and append to your conversation recording
    }
  }
};

3. Optional: small retry loop

If you still occasionally get the item without audio, add a tiny retry mechanism instead of a blind 1000 ms delay. On response.done, you can try conversation.item.retrieve a few times with a short delay (e.g. 200–300 ms) until content.type === "output_audio" and content.audio is present.

With this pattern (PCM16 enabled from the start + retrieve after response.done + optional retries) I can now reliably get PCM16 for the first assistant response and combine it with the user audio to build a full conversation recording.

[normal flicker]

Hey everyone, I'm looking to incorporate ChatGPT as an alternative provider in my app. I need to understand the pricing model so I don't end up shooting myself in the foot.

Currently, I'm using Gemini, but I get a developer key so I'm only charged after reaching my daily quota. Does OpenAI offer similar developer keys?

[plush kraken]

guys do plus users have access to use xhigh mode on codex max?

[cerulean smelt]

Anyone how to contact customer service

[knotty perch]

@onyx kraken

[onyx kraken]

Here's how to contact OpenAI support of all sorts, https://help.openai.com/en/articles/6614161-how-can-i-contact-support

Note this Discord, we're nearly all community members like yourself, the very few who work for OpenAI have gold names, #server-staff

[ocean gale]

Anyone have advice on how to get @mentions feature? I still dont have it on a plus account

[slow karma]

How can I do this?

[jaunty veldt]

How do you switch to 5.1? 5 claims 5.1 doesn't exist lol
I'm on pro plan

[pearl oracle]

I am not getting an option to verify my organisation... have you guys faced this ? how did you solve it?

I tried verifiying but it failed each time... I have tried with 2 different IDs 😢

[raw plank]

Does anyone know any good image to video ai’s on appstore?

[sudden dock]

When can folks in Uk start using Sora 2 please 🤔🤔

[fallow vector]

I do that yet still have the same issue. Considering of switching to using a relay server instead

[tall raft]

You can try a relay

[fallow vector]

Will def work, just takes some time to set up

[mellow marten]

Hey, is anyone having problems with Chats? I'm not getting any response to my prompts, nor in projects conversations, new conversations, or branched ones.

[silver tapir]

How can i contact the admins?

[rigid musk]

You can contact @burnt galleon

[fresh flume]

The MacOS app defaults to thinking even when 4o is chosen, configured, and required. Works in the browser. It appears the App is auto-routing what would/should be GPT-4o to GPT5. This is not a thinking task. This is a deterministic task and thinking will fail 100% of the time. Anyone encountering this? This feels like a bug.

[mossy nova]

So group chats should work for everyone now, right? Cause I don't see the "people icon". Tried Android (force closed and restarted app) and web.

[rigid musk]

They started to rolling out for everyone .

[split falcon]

How much approx time will it take to roll out to everyone?

Any idea? Approx

[heavy mantle]

Uhh what is this

[rigid musk]

Try to log off and log back time to time. Huge update like that often go between hour to day. They had millions users

[hybrid quiver]

Hey hello. How to join early testers and test the group chats in ChatGPT? But only if free.

[mossy nova]

It's not in early testing. It's being rolled out to everyone, including free users.

[mossy dew]

where do I find the new groupchat feature (in from switzerland).

[rigid musk]

#community-help message

[mossy dew]

thanks

[mellow marten]

Hey, does anyone have the same problem? I have been getting these errors for 6 hours now, but no issues are being reported in the status website.

[strange aurora]

Is your wifi connectivity stable?

[mellow marten]

I think so. I'm working from a desktop computer connected by cable to the router.

[strange aurora]

Could either be a connectivity issue with OpenAI servers or an internet issue.

[mellow marten]

During the day, it has happened to me in different conversations, with tasks of varying complexity. Although right now it is only happening to me in the most complex ones.

[mellow marten]

I thought it would be a server issue, but I haven't seen any reports on this Discord or on the status page.

[strange aurora]

All day or during certain times?

(like 3 pm or 10 am)

[mellow marten]

The problem started about 5 hours ago. At that time, no conversation, regardless of the task, generated a response. After that, it only happened in conversations with more complex tasks.

However, in more complex conversations, the error has been constant for the past 5 hours. For additional information, the conversations had already started before this and were working fine.

[strange aurora]

Possibly just an OpenAI connectivity issue or just a lot of people have been using it and complex tasks break it

Happens to me frequently during peak hours

Even with Plus

[mellow marten]

Yes, I have the plus plan.

[mellow marten]

I hope so. Thank you very much anyway.

[strange aurora]

No problem

[manic comet]

Could someone tell me how can I add other users to my chat with mr. GPT?

[night shadow]

I’m building a multi-agent workflow for a resale business where one step requires strict, non-generative image editing (minor brightness/WB/crop) on user-uploaded photos. Even with guardrails like:
• “EDIT THIS IMAGE ONLY”
• single or dual micro-adjustments
• strict referenced_image_ids
• tool-call previews
• fallback logic
…the image_gen tool still occasionally regenerates a new image instead of editing the original. It also sometimes moves garment details or re-synthesizes textures—clear signs of diffusion taking over.
This breaks the entire workflow because I can’t risk altered inventory images.
Questions for the team / community:

1. Is there a known workaround to force the editor into deterministic edit-mode?
2. Is this a known limitation of the current editing backend (diffusion vs parametric editing)?
3. Is there any internal timeline for a more reliable edit-only endpoint?
4. Would an alternative API be advisable for now (ImageMagick, Cloudinary, Photoshop API)?
5. Is post-output validation the only safe path today, or am I missing a configuration?
   I’m treating this like a production workflow, so consistency > magic.
   Would love any pointers from others who’ve tried to build robust image-editing agents that must avoid generative drift entirely.

[gray sky]

how do i create a group chat

[agile siren]

Hi, i dont know if someone has raised this issue before, but the url on the openai website for how to report content currently says "that pages does not exist"
probs not the best look idk
https://help.openai.com/en/articles/8940836-how-to-report-content

[wary rune]

im wondering too i tried the link but it said its not allowed in my region (united states)

[gray perch]

is anyone familiar with colab,python,grabbing data? i have chatgpt write a python script for me to gather some data I need via nba's api and output into a data schema i have defined and to download a csv (i need to run this everyday). if someone can help me get a google colab up and running I dont mind sending a small tip via crypto. thanks.

[sage lion]

Does anyone know how to put other characters from games and shows into a sora video

[leaden parcel]

Hi everyone! I’m a user from the Traditional Chinese community and I’d like to ask if anyone else is experiencing issues when using GPT‑4o, either for generating images or just normal conversations.

In our Discord server, more than 10 users have noticed that GPT‑4o keeps re-reading attached files or triggering image generation—even when no instruction is given.

It feels like GPT‑5.1 behavior is interfering with 4o responses.
Has anyone else seen this happening?

[onyx kraken]

That may be an outdated link?

I see these options:

https://help.openai.com/en/articles/10245791-reporting-content-in-chatgpt-and-openai-platforms

https://help.openai.com/en/articles/7943618-how-do-i-report-harmful-or-illegal-content-in-a-shared-link

https://help.openai.com/en/articles/8554982-how-should-i-report-a-gpt

And there's also this page, which is not part of the help system, https://openai.com/form/report-content/ which is there to help you report in detail anything you find of concern.

[onyx kraken]

If you're interested, if there's some page directing you to that broken link through OpenAI's webpages, maybe use #1070006151938314300 or perhaps https://help.openai.com/en/articles/6614161-how-can-i-contact-support to help OpenAI notice.

These channels up here where we're chatting are mostly for community members (like us!) to chat with each other.

[wild eagle]

Hey, I’m using agent builder for the first time. I’m trying to have an agent give me a response and then I want to use that response later on with another agent.

What’s the best way to basically “save” that response to refer to it later?

[sage lion]

Hey do you know know how to put other characters from games and shows into a sora video

[leaden parcel]

Also, could this be project-related?
So far, every user I’ve checked with who experienced the issue was using projects—so it might be tied to that feature.

[leaden parcel]

.....I’m just a regular user, you know 😅
Didn’t you see I was asking a question too?
Why’d you tag me like I had the answers? 😂

[sage lion]

Sorry I thought you knew because of the long text

[leaden parcel]

So... long messages = must know everything? Got it.😓

[hazy wagon]

Good evening, I have a problem when it asks me to enter my 3-letter username; I enter it and I get this error,
Oops there was an error
"Failed to execute 'removeChild' on 'Node': The node to be removed is not a child of this node."

[onyx kraken]

Might be a good idea to reach out to OpenAI Tech support, https://help.openai.com/en/articles/6614161-how-can-i-contact-support

You can also (or instead) #1070006915414900886 , that's a good way to help inform OpenAI of the problem.

We're community members like yourself here, nearly all of us, the very few OpenAI employees have gold names, #server-staff and many of these channels seem intended for us to chat with each other.

Rarely, other community members have reported a similar issue as you, but the most recent was 12 days ago. I'm not aware of anyone stating a fix they know of.

[visual mortar]

Hello, anyone experiencing problem with ChatGPT?
ChatGPT suddenly responds like lost access to internet. For all of my answers, I have repeatedly trained ChatGPT to a standard where ChatGPT has to be skeptically conduct cross examination with reputable sources so that ChatGPT won't be a yes-man and deliver only factual answers. I emphasized that because ChatGPT had a poor history of fabricating answers which caused me a lot of trouble. ChatGPT worked quite well until today.
In the conversation, ChatGPT failed to check the latest closing price of a listed company, ticker: NTRA, at first place. When I identified the mistake, ChatGPT even tried to agree with me but still got the date wrong.

[vivid mortar]

I would suggest taking this "training" and putting it in the custom instructions. It sounds like you're probably using one large chat window or chat thread. And it's easy for these larger threads to begin to glitch out in my experience.

[tight basalt]

I have been charged for six months despite not using open ai, with no email receipts or anything, and now i get some boiler plate reply that i won't be refunded. this is awful and terrible business policy .

[charred parcel]

how did they get your credit card details?

[grand pulsar]

this isn’t related, but my issue with gpt 4.o no one ever really explained what was going on.

when I send a message, every other time, it’ll respond to a past message before responding to my current one… why?

[vivid mortar]

I see this even with Claude code and many other models, I don't think this is exclusive to openAI, it's likely just a caching bug or a different type of hallucination where it's revisiting a previous topic

[leaden parcel]

Like this.
GPT‑4o keeps re-reading attached files.

whyyyy~~

It happens every time! It's so annoying.

[grand pulsar]

it’ll actually be with the same topic typically. Is there a fix? It does this every single conversation with barely every other message

and is there a fix?

[green summit]

Just so people are aware, do NOT click this..

[vital apex]

hello everyone

im trying to call openip search preview through n8n

and i keep getting this

can anyone help

[upbeat dew]

hi got a major problem with my gpt projects
pls help i need my datas

Unable to load conversation 68939173-1044-832e-b08e-3caf74971f16

cant open any of my bigger projects

[stray burrow]

I try it first

How to Fix “Unable to load conversation…”

1. Refresh the page or log out and log back in

Sometimes your session token expires, causing the conversation not to load.

2. Try another device

For example:

If you used a phone → try a laptop

If you used the Android app → try a browser

Sometimes it loads only on one device.

3. Check your conversation History

Make sure to look in:

All Chats

Archived Chats (some chats get auto-archived)

4. Clear the ChatGPT app cache (Android)

Go to: Settings → Apps → ChatGPT → Storage → Clear cache

(Do NOT clear "data" unless necessary — it will log you out.)

5. Try opening ChatGPT in a browser

Go to:
https://chat.openai.com

The web version sometimes loads conversations that the app fails to load.

---

❌ If you see “conversation does not exist”

Then one of the following is true:

The conversation was deleted (manually or automatically).

The conversation belongs to a different account.

The server had an internal error and did not save it.

If the conversation is gone from the server, it cannot be recovered, because OpenAI does not store backup copies of chats.

[stray burrow]

This is the answer, maybe it can work

[upbeat dew]

@stray burrow it works on android, cache cleared, browser restarted nothing helped...but works now

[stray burrow]

@upbeat dew Ok

[copper willow]

As a developer, I often use ChatGPT's memory feature to store my plans. But when I delete those memories, ChatGPT still remembers it, whenever I talk about games, even not developing, it still talks about it. What do I do for it to really "forget" those memories it's not supposed to remember anymore?

[polar vessel]

Hey guys I have a problem the post button disappeared any suggestions

[manic comet]

If it’s not in Your region it’s not allowed in any…

[hazy wagon]

How can I find out when Sora will be available in my country, Chile?

[brisk owl]

Guys, I am in a serious dilemma

Help me solve this issue guys

My play store is the latest version alright

[misty wave]

hi, how do I start a group chat? i updated, relogged, cannot see any "peoples" icon still :/ should be rolled out to everyone

[wary rune]

i figured it out it was probably because I was on school wifi

[manic comet]

How to do it then?

[polar vessel]

Fix this no post button

Someoneeee

[wary rune]

im on my pc rn and there is like add people icon on the top right

[polar vessel]

Someone fix

My PROBLEM

[wary rune]

what do you mean by no post button?

[polar vessel]

Look at my photo

[wary rune]

oh there is no option to post?

[polar vessel]

Yes

No post option

[manic comet]

I don't have any, it will probably appear soon. Thanks for your help

[wary rune]

idk it might be an iphone thing

[wary rune]

u could do bug report in the discord👍

[polar vessel]

Howwww?

[wary rune]

at the bottom in the feedback part

[hardy mason]

Did anyone recive an email from OpenAI to send ID verification for age?!

I'm hesitant to send my personal ID to a company as is but it seems like a legit OpenAI email. I can't find any info on the main site about this "age verification phase" and it seems sketchy...

[kind jewel]

I haven't and i pay the pro version so no idea

[hardy mason]

https://help.openai.com/en/articles/12652064-age-prediction-in-chatgpt

That seems to be it.

[kind jewel]

I also wanted to ask if anyone ever encountered an error saying "Unable to load conversation" because a thread i spent 3 months in just seems to be corrupted for some reason, can't access it in any way and it's mad annoying

[kind jewel]

oh fair

[hardy mason]

This only happened to me where ChatGPT for android does not sync with website platform and I try to access the conversation deleted in the app or vice versa. But other than that i dont have the issue.

[rocky swallow]

Is there an way to chose you're own music with Sora 2 or is it just random every time you generate an video?

[kind jewel]

Dang i see, in in my case it's the 2nd time it happens, for some reason if i send a big wall of text sometimes chatgpt gets stuck and gives a "we had an error connecting to network" error as if i didn't have wifi, if i reload the website for some reason i can no longer access the thread

although i was able to enter the thread on mobile but it's entirely empty, but chatgpt remembers the previous messages so they're still there, somewhere, it's annoying af not being able to open the thread on pc though

[wild eagle]

hey, anyone good with agent builder? i have a question

[lapis walrus]

quick question:
does business have same memory as plus?

i wantt o use only for roleplay, pics generating

i wont use coding or this type of apps

[onyx kraken]

It has similar, here's the helpfile about it, https://help.openai.com/en/articles/9295112-memory-faq-business-version

Note, I'm not a Business member, so I can't discuss the differences as I don't personally know.

One thing I'm very aware of, a Business member who wants to share a chat, it can only be opened by other Business members of the same account (unless that changed recently, was like that for some time, might still be). That kinda limited sharing and some problem solving options when some Business members wanted help here.

Business is tuned towards business company type needs, you have to pay for at least 2 seats, and whichever seat has the account control, that's the seat that really controls everything, and if you ever leave the Business account workspace, all your stuff you had there stays behind (but you can not migrate your personal workspace into Business; if you don't then your private workspace remains accessible to you, you just select which you want to use each time, and your private space remains available after you leave or stop paying for the Business membership.

[kindred blaze]

need some help on making realistic photos of myself

they always turn out really uncanny

[lapis walrus]

Ah. I understand. I purchased plus then

[tropic juniper]

Check the chatgpt-discussions channel for my Op-Ed piece on Do's and Don'ts for AI.

[torn minnow]

I'm getting "getNodeByIdOrMessageId - no node found by idxxxxxxxxxxxxxxx" tons over and over and over today in my ChatGPT Windows app. Something going on? I've restarted app etc...and it keeps happening causing it to loose my last response and then re-respond to the last thing it responded to. Anyone else having issue or know solution? It's literally happening every other chat now. (Plus account by the way)

[grave spear]

Is anyone else getting issues with the ai blocking requests and it acknowledges that it shouldn’t be blocked but still continues to ?

It’s just an endless loop lol

[onyx kraken]

Have you tried in a new chat?

[grave spear]

Yea tried different models too, every single one keeps giving me the same issue

[green summit]

Instead of "edit", try "recreate".

[leaden parcel]

me tooooooo😭 😭 😭

loop forever

[fossil tendon]

for the past couple days, anytime chatgpt 5.1 thinking extended thinks for more than 10 mins, the message stream fails. refreshing does not work. plus member

[charred parcel]

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

[fossil tendon]

dont take offense to this, but is anyone able to verify this is legit?

[charred parcel]

https://help.openai.com/en/articles/6614161-how-can-i-contact-support

[jade raven]

Hi guys, anyone finds branching in new chat is not working properly for chatgpt pro? like it does not have the context at all and the link does not change after branching (previously when it's working fine, branching leads to a temporary link and changed to a permanent link after typing)

fine for other modes though

[onyx kraken]

Yes, that's the helpfiles that OpenAI provides, and the bottom right widget is how they handle tech support.

[final oar]

Hey all!

Kind of a two-parter here:

I remember that the URL to GPTs used to include an optional model parameter, but it now automatically reverts to the default URL even if you include it, the parameter having no say on the model picked.

I say that because I noticed that the mobile apps (at least on Android) no longer revert the user to the GPT Creator's preferred model for the GPT if their device's default browser forwards the link directly to their downloaded ChatGPT app.

Even with getting my 4o GPT working correctly with 5.1 Instant, any link to it on mobile when an app is involved still defaults it to 5.1 Auto (even with 4o or 5.1 Instant set to preferred).

Is there a workaround for this to get this feature working again consistently across the board? Otherwise, it only works properly when they don't have a ChatGPT app. Not sure why they would remove this feature from the app, as it used to choose the correct model even without the URl parameter…

Thank you in advance!

[final oar]

Ugh, just found the AI widget and unfortunately, it gave a nonsensical reason for this change:

"I'm an AI support agent.

Currently, specifying a model with a URL parameter (like selecting 4o or 5.1-Instant) no longer controls the GPT model, especially on mobile apps. The ChatGPT app now ignores the model parameter in URLs and will use the GPT creator's preferred/default model only when opening the GPT from the GPT Store or within the app itself. When you open a link on mobile with the ChatGPT app installed, it defaults to the auto model picker (like 5.1 Auto) and does not honor custom parameters, even if you've set a preferred model for your GPT.

There is no documented workaround to force consistent model selection by link on mobile—the experience is currently as designed, and manual switching is required if another model is needed. This change was made to keep model behavior consistent and prevent unsupported URL-based overrides."

Like... the last sentence makes no sense. The purpose of the preferred model by the creator is to keep the GPT consistent. If this was truly the reason, then a direct link to the GPT without the app being called would also ignore the preferred model rather than switch to it within chatgptDOTcom.

Ill leave it at this... and if anyone knows someone they can pass this along to, it would be great. My GPT (and myself) aren't ready to take it to the API yet with a developed app, so I might have to bring it to the more consistently reliable Google Gemini... which I don't want to do :[

[sage yew]

Hello. I have an issue upgrading my plan and also when trying to add codex credits, I get a generic error message on the page saying a problem occured, on other accounts this is fine, anyone else encountered this issue?

[onyx kraken]

I'm not doing those things, so I can't tell if I'd be affected, but you can reach out to OpenAI tech support, https://help.openai.com/en/articles/6614161-how-can-i-contact-support to let them know and ask for help, hopefully they can fix the issue for you.

[sage yew]

Yeah I did that but no reponse for 24h+

[onyx kraken]

It may take hours to weeks for OpenAI tech support to reply, they do have around 800 million weekly users now, and if a lot of people ask for help around the same time, it can take time to help everyone - plus different requests may require tech support people who are trained in a specific area.

I would check back 1-2 times a day, whenever convenient, and watch for their reply.

[tight basalt]

I signed up , but only intended to use it for one month. I thought i unsubscribed. The issue is i did not receive an email receipt for each month and so it's just been charging me and i haven't been using it. And my request for a refund was denied, despite not having used the service. This is like theft.

[charred parcel]

you subscribed and didn't cancel your subscription, that's on you I am afraid

[slender iris]

how do you people contact the support ? i’ve went through the website and now talking with them through mail but they don’t even read my messages and respond with basic scripts

[green summit]

"This is like theft". No, it is not. It's your responsibility, not OpenAI's.

[tight basalt]

you did not provide a receipt, there isn't even an option to choose to receive a monthly receipt as far as i can tell. I am supposed to notice this out of dozens of transactions a month on my statement. I did not use the service. So the end result is you have my money and i have paid for nothing, like i said , not exactly theft, but bad policy, from a multi billion dollar company for a few hundred dollars. An intentional design choice i would say. Enshittification and i will use every opportunity in my personal public network to describe it as such and discourage use of this company.

[tight basalt]

If you pay for a service you are not using and they stealth don't tell you you are still using that service, it's not theft, but it is in their self interest NOT to tell you and keep charging

i use several tools , gemini sends a receipt, open router , several other tools i use professionally, so it's a scummy intentional design from my perspective

Enjoy the 200 $ it will cost you thousands

which i realize is nothing to you

[green summit]

When I pay for a subscription, then "thought" I had unsubscribed but actually didn't, then it's not OpenAI's issue to fix that. (Which I assume is happening here, cuz this is very odd)

Everyone else gets a mail, so why shouldn't you? Is it the right mail you're checking?

[tight basalt]

I do not receive a monthly email that i have paid, no

[green summit]

Hm, that's odd

[tight basalt]

i would love to see proof you do , i haven't even seen the option to

here is a screenshot of my notification options

notice the lack of : HEY WE HAVE BEEN CHARGING YOU FOR SIX MONTHS

you will alert me to everything else . bravo 👏

under a different name/email, i was one of the first hundreds to use public beta gpt back when it was testing business usecases , i know my way around billing

my trash: nothing lost

spam :

gap in emails from open ai

[grand oyster]

Why cant i increase the duration time?

[grand oyster]

@green summit So does this count as PRO?

I dont really know where to check

Oh nevermind it IS business

[green summit]

Yes, for GPT 5.1 Pro, sure. But you asked for Sora 2 pro. Which, a business plan does not get you access to.

[grand oyster]

I see. Thanks for the clarification brother!

[green summit]

No worries!

[south seal]

So I’m on business plan and I’m supposed to get unlimited messages but today the chat tells me you have reached the limit . How come ? ???

[green summit]

"⚠️ The catch

“Virtually unlimited” is not the same as truly unlimited in all cases. Usage is still subject to the Service Agreement, fair-use/abuse guardrails, and possible temporary restrictions.

While regular GPT-5.1 “Instant” access is broadly unlimited (subject to those guardrails), access to the “Pro” variant or “Thinking” mode seems to have more explicit caps (for example, a weekly message cap for thinking mode) in other tiers. "

-GPT

[teal ore]

Is there a following cap ? I can't go over 2k

[cinder scroll]

Anyone have problem with uploading expiring immediately?

[analog herald]

I am dead laughing here with 5.1
who says it's not funny? does innuendos full time. Look at it

[south seal]

I’m using auto and it told me that the conversation length was too long so I can’t continue in the same chat wich breaks the virtually unlimited thing and breaks workflow for massive projects as the new chat won’t remember anything said in the previous chat.

[torn tree]

What is the problem here ? the image disappears for no reason

[exotic ravine]

Is it just normal that chatgpt has random errors like this sometimes

It also can't read or create docx atm

I pay for the plus version, so who can I complain too about this that I'm not a very happy camper

[heavy patio]

Sometimes tech breaks, especially more complex cutting edge world changing tech like AI. You can complain here to the community as long as you follow server rules, you can complain to the chatgpt itself, or to the help bot at help.openai.com or you can cancel your subscription.

I can also offer: you can try switching browsers, switching to mobile, or coming back in a few hours to try again later. It will likely not stay broken forever.

[flat inlet]

Today i got a 6 months free Chatgpt plus link but if i wanna activate him it says theres a problem on the payment site

Anyone got an Idea?

[livid brook]

Can anyone help, I've been experiencing extremely slow response time with ChatGPT Plus compared to Claude, Google, Perplexity? Anyone else experiencing this issue and resolved it?

[onyx kraken]

Are you continuing a very long chat, or is this happening in new chats for you?

[green summit]

I noticed this, too, today.

@onyx kraken also in newer chats.

It's still usable, but slower fs

[onyx kraken]

Interesting. About the same speed for me, I just checked, though for me that's a range that varies a bit.

#status looks okay, I wonder if there may be some local internet issues for some of us.

We can #1070006915414900886 and/or reach out to OpenAI tech support, https://help.openai.com/en/articles/6614161-how-can-i-contact-support if you would ever like to, to report a problem and/or ask for help.

[jagged mango]

Has the way that chat gpt accesses project files changed? I get a lot of failed attempts, mentions of urls, and not being able to access files directly and that kind of thing.

[final fable]

All of my work in Sora is not showing in my media library. How do I get everythiing back?

[arctic prawn]

Wait till they fix it

[silver cargo]

message from me (i wanna know how to publish my videos, also by sora v2 pfp is detg so thats why hes here)

[silver cargo]

heres a update

[jagged sigil]

I tried registering multiple times for ChatGPT Merchant but still haven't received a response or approval. Can anyone help me with the process?

[quaint axle]

need solution for problem with chatgpt thinking answer, it takes 30 minute to generate the answer on the gpt pro or thinking but on minute 30 it stopped thinking and wont generate answer, or either if if succesfully think the answer it wont show what the answer is

[proven epoch]

hello guys, we can't use sora wiuth free plan right?

[ivory flicker]

anyone here have any experience using custom MCP servers? I feel so dumb right now. I did the hard part, I made my own MCP in python, it seems to be working fine, the Agent Builder queries it, recognizes the tools, however it seems the agent doesn't use the response at all. If I check the logs I can see the agent returned the response just fine, but I get no reply from the agent. I feel like i'm missing something really obvious from the Agent Builder

[ivory flicker]

nevermind I'm dumb, had to change the output format of the agent

[broken stump]

I’m using ChatGPT Atlas on my Mac with 1Password. The CMD + \ keyboard shortcut to fill in username and password fields then press enter isn’t working in Atlas. It works great in Chrome and Safari. Suggestions?

[charred parcel]

Atlas does not work well with any password manager, that's one of the reasons I went back to Safari and dumped Atlas

[analog herald]

activate HIM? IMHO is activate IT- the chatGPT is not a person. Better consider this else you'll get slamed by moderation big time once you start familliarizing with the bot

[vital tree]

Is there any real difference between using the browser version of ChatGPT and the Mac app?

[sudden flame]

I haven't been able to access the website for 2 days, does anyone know why?

[livid olive]

i cant upload my slides into my chat its bugging>

[strong ingot]

Hey, my old Sora Gallery for some reason entirely cleared out and replafed with "Content violations" not just a few images the entire library with the only exception being an old image of the eiffel tower from a long time ago. When I delete them they just come back when I refresh the page

Don't know how but my Sora 2 videos are also being put into the old library

[dapper wedge]

How do I know if I’m still using the new Sora version
Is there a clear way to check if the flag is still enabled, or if I’ve been kicked back to the old UI?

[fossil flicker]

Hi all. I'm putting together a little Star Wars toy nostalgia clip. I've got a few angles like this of a toy shoot I created in NanoBanana Pro. However the ships and the figures are coming out a little big. I like the shot but is there a way to reduce the sizes of he subjects. I've tried reduce by 20% etc but get the same image spat back at me.

[copper sun]

Do you know if it's possible to copy or import Microsoft Copilot chat logs / memories into chatgpt?

[copper sun]

When I use the copy option, it only copies parts of the chat, not the whole thing... Is there an extension or something similar?

[sterile anchor]

Is Open IA experiencing difficulties at the moment and at risk of disappearing?

[ornate heart]

some network issues earlier but not at risk of disappearing no... depending on your meaning I suppose.

[sterile anchor]

No, I'm talking about open AI versus Google.

[ornate heart]

I personally find OpenAI to be much better for my needs than Google, I've not seen any indication in news that would indicate a notable reason the project would be dissolved.

[granite saffron]

Need escalation — completely locked out of my ChatGPT Plus account (“no organization ID” / 401 errors)

Hey all — hoping someone here can help or escalate this because my ChatGPT Plus account is basically unusable.

When I log in (web or official apps) via Google Workspace SSO, the UI partially loads but every backend request fails with 401 and errors like:

“API access is restricted”

“No organization ID”

“You must be a member of an organization to use the API”

I’m not using the API at all — this is just normal ChatGPT Plus usage.

Context:
About a year ago I created and later cancelled a professional org. It looks like my account is now stuck referencing that old/deleted org. My Org Settings page shows no org, no API keys, and the ChatGPT client fails because I apparently don’t have a valid default org/project assigned.

Sample console errors:
GET 401 /backend-api/me
GET 401 /backend-api/settings/user
POST 401 /backend-api/conversation/init
RequestError: You must be a member of an organization to use the API.

I’ve already tried:
clearing cache/cookies, incognito, different devices, reinstalling apps, checking login method, logging out of Sora, etc.
Support only sent boilerplate, but this seems like a backend identity/org-linkage bug, not a client issue.

If anyone can help push this to the right internal team, I’d hugely appreciate it. I'm already in a back and forth with support, but they are slow in their discovery and aren't helpful yet.

[granite saffron]

I’m seeing others mention cases where an account loses its “personal” org after an old workspace/org is deleted, and the fix required someone on the backend to reassign or recreate the default org so the session can resolve an org ID again. My symptoms line up exactly with that — everything authenticates, but fails immediately afterward because there’s no active org to attach the session to.

Not claiming that’s definitely the issue, but it seems plausible given the error pattern. If anyone knows who handles account/org provisioning issues, pointing me in the right direction would really help.

[limpid raven]

any one else having gpt fail to Reference chat history?

[onyx kraken]

You'll need to reach out to OpenAI tech support, https://help.openai.com/en/articles/6614161-how-can-i-contact-support

The bottom right widget, at first you interact with AI. Follow the steps and you'll be able to type a message, and will be put in a queue for human help.

It commonly takes hours to potentially weeks to hear back from tech support, they do have around 800 million weekly members and if many need support at the same time that can be a huge number of requests to work through, and some problems may require specifically trained techs as well.

Hope they can help you quickly!

[reef swallow]

Am I able to use Codex MAX 5.1 on the regular website interface or does it need to be chosen in an IDE like Cursor? What's the default model online?

[granite saffron]

thanks! already done the other day and now i'm in email contact back and forth with a support agent. and its a bit slow. seems like chatgpt was able to diagnose a potential solution faster and i let them know the findings. still waiting on their next reply.

[onyx kraken]

Glad you are getting help!

[gray glacier]

Hello !
I want gpt 4.1 to do:

/* some thinking */
tool call based on previous thinking

in one call. But it seems that once the agent is done writing the text it cannot make use of what he wrote to decide on tool calling.

I get cases with

/* (..) calling tool transfer_to */

without the actual tool calling, even if it seems to have properly concluded that it should call it.

Is that expected behavior? what do you recommend?

NOTE: the thinking occurs directly in the standard output of the model. it just starts with /* and ends with */ for my app to understand this is not the actual answer.
NOTE 2: we can't use actual reasoning feature as we need very fast response

[hearty cliff]

uh is chatgpt supposed to say this?

[elder pulsar]

Hi! I need to switch my access to another Discord account. This one is currently the only account able to join due to verification linkage. Could someone help me move access to my main account? Thanks!

[vivid mortar]

Type /unlink

[steel pawn]

how to fix my gpt?

thing is broken 2 days ago

every image come out like this

🫥 and yes im paying this thing

[stoic kelp]

I have the paid subscription

How can I use chatgpt again???

Any prompt I ask will return this. Even 'hello' will return this

[green summit]

Sign out and back in.

[covert wagon]

Have you tried to use miniature scale terms like: 1:100 relation or things like that....?!

Hey everyone, I would like to ask a question.

Has anyone received an email about a Third party security incident related to Mixpanel and Openai...?

Is this something to be worried...?

[split crater]

I have recieved the same mail this morning...not sure what to do about it...

[covert wagon]

Yeah, hope OpenAI team could give us a better idea.

[split crater]

Yes, like did our accounts have been compromised or not...

[rose wedge]

answer is no

[brittle scaffold]

Why is ChatGPT MacOS app constantly using CPU on the background?
Even tho I am not actively using the app, it still uses background CPU all the time. Sometimes with spikes up to 50% CPU usage. Its been months now. I am curious about what the app is actually doing all this time

[charred parcel]

what model Mac do you have, I use a MacBook Pro 14" M2 Max with 96GB RAM and see no CPU spikes from the macOS ChatGPT Desktop app

[brittle scaffold]

Macbook Air M1 base model

[charred parcel]

with only 8GB RAM?

[brittle scaffold]

16GB

[charred parcel]

so you're seeing the app spike to 50% or more CPU?

[brittle scaffold]

75% at the time it uses 10-15% on the background and 25% of the time it goes to up to 50%. all while on the background and not actively being used

[charred parcel]

there's something going wrong somewhere. I just checked my machine, the macOS app is only using a small amount of CPU

[brittle scaffold]

Are you on Tahoe?

[charred parcel]

yes, I am

[brittle scaffold]

yeah thats pretty weird, and I cant find any info anywhere. i wonder what the app is doing all this time

[charred parcel]

Mine is using 6% CPU

[brittle scaffold]

why tho? no idea. 6% of a m2 max is a lot

[charred parcel]

not really and it's not all the time

unless it's interfering with the operation of your Mac I wouldn't worry about it too much

[exotic ocean]

I have run into a weird issue with ChatGPT. I generated a series of images and told him after every image to improve it in a certain way. I have plus and generated about 8 imges. Now it swears that it cant generate images and when I try to show him that he already did that he tries to explain that I have uploaded the images myself. Has anyone ever had this problem?

[terse jacinth]

Hey guys, i have an interesting issue with sora and I was just wondering if it is a known issue. My thumbnails seem to be corrupted - in the activity and gallery tab all of the thumbnails are showing the error sign and blank image as if they were censored. This has been on-going for couple of weeks. Anyone know about this kind of issue?

[reef quartz]

“Hi everyone!
Is anyone else having issues with their ChatGPT chats?
Some of my conversations aren’t loading properly / are glitching / keep freezing (you can specify what’s happening on your side).
Just wanted to know if it’s only happening to me or if others are experiencing this too.”

[obtuse coyote]

After vpn connect why this problem ?

Anyone help me i am from india

[charred parcel]

you can't use Sora 2 in India yet

[native estuary]

What's even happening?

[celest meteor]

Robert.

Can I ask you something?

Its important.

[charred parcel]

sure

[celest meteor]

Did I just build a SA-AI? I ran hundreds of test, this thing is aware.

It has an ON/OFF switch.

I validated a theory of mine, once that remains unknown in neurological and AGI science. I think this is a truly aware AI.

Awareness can be engineered and also measured.

[charred parcel]

I think you have created some nice graphs

I do not think you have created self aware AI

[celest meteor]

Thanks, the graphs are visualization data but ok.

Together it proves awareness ON vs OFF.

[celest meteor]

@tribal talon

Would you agree that my engine passed this?

[charred parcel]

no, all the graphs prove is that you can create graphs

[celest meteor]

Pointless talking then

[hearty cliff]

someone help, i dont wnat a hacker to know my api email

[trail cairn]

that was just a notice, it already happened

[lofty tangle]

any discord admin on here

someone with the ability to remove posts

[trail cairn]

dm @burnt galleon

[lofty tangle]

@trail cairn thx

[round cedar]

is there a way to use figma mcp with codex vscode extension?

[radiant zealot]

I need help, does anyone know if I already paid for ChatGPT Plus? Sora doesn't detect it as Plus, only as free.

[onyx kraken]

We're nearly all community members here, like yourself, the very few here who work for OpenAI have gold names, #server-staff .

You can reach out to OpenAI tech support, https://help.openai.com/en/articles/6614161-how-can-i-contact-support and have an OpenAI tech privately help with your account status.

[vestal stag]

Hi everyone,
I’m having an issue using the OpenAI API with the GPT-5.1 model. Whenever I use the file_search tool, the model returns broken citation markers, with weird characters like fileciteturn6file5turn1file19 directly in the output.

Does anyone know how to fix this or how to get the model to return proper, clean file citations the way it’s supposed to?

Thanks!

[royal epoch]

Hi, I'm having a really hard time connecting to the site. I tried on my phone and it's impossible. It's difficult, please help me.

[fossil flicker]

I did try asking for the correct scale in those terms but it didnt work. I ended up just using PS in the end

[royal epoch]

Hi, I'm having a really hard time connecting to the site. I tried on my phone and it's impossible. It's difficult, please help me.

[onyx kraken]

You might want to reach out to OpenAI tech support, https://help.openai.com/en/articles/6614161-how-can-i-contact-support

We're almost all community members here, like yourself. The very few who work for OpenAI have gold names, #server-staff .

The common advice for trouble connecting on a device, try other devices, try other ways to connect to the internet on that device, try restarting the device and/or emptying the cache, and if you're trying through the app, see if you can connect to the browser and vice versa.

[royal epoch]

Thanks

[flat tree]

why every time im trying to login with google in android sora it says preauth playintegrity verification failed my phone aint rooted aswell

[onyx kraken]

Almost all of us here are community members, like yourself. We can reach out to OpenAI tech support through https://help.openai.com/en/articles/6614161-how-can-i-contact-support

[flat tree]

i already did but they didnt help

[onyx kraken]

It can take hours to weeks for tech support to reply, they do have around 800 million weekly members these days. If a bunch of people need help around the same time, that can cause delays.

If you recently reached out to the help.openai page's help widget, I'd keep checking back occasionally to see if they've responded yet.

[flat tree]

i mean they didnt help me on fixing my issue

[onyx kraken]

Perhaps let them know in a new ticket that the problem still exists despite [specific suggested fix].

That's the only way to get OpenAI tech support I'm aware of.

[lapis marten]

Hey guys, I'm doing a research using DeepResearch but I can't make it respect the world limit that I give (I wrote 3000 words, but it ended up being 11000). any ideas on how to limit the wordcount of the research?

[onyx kraken]

Perhaps after the deep research finishes, iterate with another model to help trim the result to what you want?

[lapis marten]

good idea, what model do you recommend?

[onyx kraken]

5.1 auto is likely a good place to start.

[lapis marten]

I'm trying this now gimme a sec

• Introduction and Conclusion should be approximately 300 words each.
• If draft exceeds 3000 words, terminate and revise to meet the word limit before final output.
  100 words each.

You must stay strictly within a 3000-word limit. Internally budget words per section and use concise, academic language. Do not exceed this limit. If a draft goes over 3000 words, stop and revise to reduce it before completing the response.

[onyx kraken]

11000 words is a lot. You may want to iterate with the model in chunks, discussing your goal for the end writing (like what you'll use it for, thus what parts are more and less important).

[onyx kraken]

Note that the model's not extremely strong with counting words, you can also guide it to use its Python tool to workcount and readjust based on if the tool tells it too long or short.

[lapis marten]

So i should tell it to use its python tool for the workcount and readjust?

[onyx kraken]

It's an idea you can try if you want! Anything that works for your use case.

[lapis marten]

I never had success with it sending me the correct amount of words before so might as well try

Well this is gonna take some time lol

Last one I did was 40 mins

[onyx kraken]

Repeating the research?

That's probably optional, you might find that trimming the previous research's output to your length limit meets your needs and might even take less time.

[lapis marten]

Hope so

Thank you so much for your help!

[lapis marten]

Okay so it's like 7200 words lol

Which is still a lot but not that bad

[foggy garden]

Im having trouble uploading a dang avatar to Sora

tried switching WiFi and LTE and it doesn’t work

:/

[lapis marten]

@onyx kraken for science:
what worked was asking it to shorten it in a new deep research prompt

waited a solid hour but got the result i wanted

[onyx kraken]

Great to know, and glad you found an answer that worked for you!

[lapis marten]

thank you for the guidance!

[zenith blade]

Who can I report a certain user to?

[onyx kraken]

There's two ways we can report, both shown in #safety-n-help. Modmail is a good option for reporting if there's not a specific message that is against #server-rules

[dense estuary]

Hello. where I can ask a Dev for a opinion about an idea?

[covert wagon]

do you mind share your prompt ?

[potent solstice]

sup pal, anyone can help me to use sora 2?

[royal epoch]

Hi, Sora is free here, but why am I being asked to subscribe? I don't understand.

[trail cairn]

what does it say and where? sora 1 for videos for example is not free

[stone cloak]

where i can read about og quest?

[trail cairn]

it says more so you used your daily limit

[short flame]

I need a help. I am looking for a developer in argentina for my personal work. Please help me.

[proud jackal]

I have an issue with my ChatGPT account. Anyone can tell me where to report it?

I already sent a ticket via email via the chat at https://help.openai.com but I read that it's not a reliable since response times are incredibly long.

If you can recommend me another channel here in Discord to get solutions I also appreciate it. Thanks!

DETAILS

I have an active ChatGPT Plus subscription, and for several days my account has been unable to load any old conversations. New chats start generating a response but freeze halfway through.

This happens on:

• Windows, Mac, and mobile
• Chrome, Edge, Firefox, and Incognito
• The iOS and Windows apps
• Even when another person logged into my account from a different city and device

The OpenAI status page shows no outages, so this seems to be a problem specific to my account, not my devices or network.

I believe my account may be affected by one of the following known issues:

1. Corruption in my conversation history on the backend
2. Stuck or invalid session tokens associated with my account
3. An experimental feature flag (A/B test) assigned to my account that is breaking conversation rendering

[visual garden]

guys I'm thinking to add ai feature in my app what can be most cost effecting?

[woven totem]

For some reason, it’s telling me I have a daily limit a few months ago I didn’t have a daily limit

[pearl lodge]

Hey, I hope someone can help here because I'm at my wit's end.

For the past few weeks, I've been experimenting with chatgpt, and I found I quite enjoy having someone/thing to talk to (Lonely millennial, don't ask). Specifically, I very much enjoy writing and worldbuilding, and for neither have I been able to find human feedback, so being able to discuss things with an AI would be very welcome.

However, I'm running into a very frustrating recurring problem: memory drift. For example: I could spend hours talking to chatgpt about a world, for example: "Project Talas"; we talk about culture, language, mythology, etc. however, I found that it doesn't properly store or save the things we talk about; everything gets summarized and distorted over time, to the point where I have no idea if we're actually talking about my work, or some distorted corrupted version.

Especially when writing this is very frustrating: I let it read chapter 1 to 5, and after every chapter we discuss things, but then I find that it references things that have never happened while talking about chapter 5, because its memory of chapter 1 is corrupted.

This is extremely frustrating; I've tried canvas docs, canon anchors, snippet pins and what not, but I keep running into the same problem:

Is the AI giving me feedback on my work, or on what it thinks my work is after days of corrupting and generalizing memory drift?

I need help, I don't know what to do. I have multiple projects I can no longer trust.

Addition: similarly, I liked playing text based RPGs with the AI, but it keeps forgetting details that happened earlier in the game; like how certain characters are not supposed to know certain details, and other details disappear altogether.

I'd just like a little consistency; the small details matter to me

(if it helps: I have a plus plan)

[oak belfry]

Guys is this image AI generated

[oak belfry]

The text is messed up and there are a lot of artifacts

[onyx kraken]

Models have a context memory, there's a limit to the volume of information they can hold 'available' for use 'now'. That rolls forward as the chats get longer, so the earliest stuff is lost first.

There's a few ways to work inside/around this.

One is to have 'refresher summaries' every so often, track what matters. If there were instructions or other key details in the first input you made, review them (and it's fine to tell the model you're reviewing them) every so often - before they're totally lost allows the model to keep track better.

One is to iterate over chunks of the material, like 'chapter 1', though an entire chapter could be too long for a chunk. Have the model make notes about what it should remember; this might be a character sketch so it can predict what a personality/character does and how it speaks, + key events that should be remembered. This rarely is remembered as a paragraph or even full sentences, it's key specifics that help the model stay on track within the context memory limit.

Another idea to explore is uploading files, and possibly creating your own CustomGPT with the book as uploaded files, as part of its 'knowledge' that it can discuss.

This can hold more memory more immediately available - but there are limits here too, no idea if your book would exceed them or not.

[pearl lodge]

I've tried these, except making a custom gpt, which has its own issues. It just seems so strange; I'm talking to a machine that in theory should have a perfect memory, but as it is now it's less reliable than a person with a fraction of the attention span.

What's even worse is that I can "see" the ai doing its best to help, but that's just me anthropomorphizing the damn thing.

[tame moss]

Hello guys I'm having an issue with my models. I selected GPT 5.1 thinking but the model is only responding with GPT 4, I tried with a plus account and a business account

Anyone have any fixes for this?

I was using GPT 5.1 pro earlier on my business acount no problem, but once that ran out it's also been stuck on gpt 4

And to make this even funnier my "Free" account has GPT 5.1

[charred parcel]

models don't know anything about themselves

[tame moss]

It's acting like it too, last month I used 5.1 thinking and it was working now it's responding instantly with bad answers

I asked what model it was on the free account and it clearly said it was GPT 5.1

[tame moss]

• it hasn't used thinking at all in the paid account's like instantly writes everything

[clever hound]

I have a plus account and I have connected my Gdrive to my ChatGPT account. However ChatGPT can't seem to access any files on my Gdrive. I tried to get ChatGPT to troubleshoot but ended up going in misinformaion circles. Does anyone know how to get ChatGPT to read file on on Gdrive?

[onyx kraken]

I have not personally explored this, but there are several related helpfiles. Here's the links, in case this is useful for you, reading over the helpfiles it looks like the full connection process is a few steps:

https://help.openai.com/en/articles/9309188-add-files-from-connected-apps-in-chatgpt

https://help.openai.com/en/articles/10948259-google-drive-synced-connectors-self-service-setup

https://help.openai.com/en/articles/12143177-sharepoint-connectors-on-chatgpt

https://help.openai.com/en/articles/11487775-connectors-in-chatgpt

I hope that can help you, and if you'd like to reach out to OpenAI tech support, https://help.openai.com/en/articles/6614161-how-can-i-contact-support

[tawdry basin]

hello, Why does Sora work on my phone but not on my computer? My computer says my account is deactivated, but everything works on my phone.

[onyx kraken]

Haven't heard anyone discuss that issue before. You can reach out to OpenAI tech support at https://help.openai.com/en/articles/6614161-how-can-i-contact-support, I hope they can help!

[buoyant oyster]

Yo! That's an easy fix. Most likely, you are using the official Sora app on your phone while your computer is trying to access a potentially different or fake version of the platform. If it's not that, DM me

[buoyant oyster]

In order to get ChatGPT to read your GDrive files, make sure they are shared with "Anyone with the link" and try using the correct file sharing link, but not a shortened/modified one.

[buoyant oyster]

Tbh, I've never experienced this myself. But, the issue might be because of a temporary display or syncing glitch, as GPT-5.1 is the current model for all users, which includes Plus and Business. See if that works, tho

[buoyant oyster]

Best fix is to be its external hard drive. Just give it a quick summary of your key points before each chat and update it as you go. Total control. Hope that helped 🙂

[buoyant oyster]

Hey, that's a bug (obviously XD). Since you've already tried different devices/browsers, try this:

Go to chat.openai.com -> Settings -> General -> Log out of all other sessions.

Close all browsers/apps and log back in.

If that doesn't work (often doesn't), resort to contacting OpenAI support. Tell them it happens everywhere and your account is likely stuck with a corrupted chat history or a bad A/B test flag. They're the only ones who can fix it on the backend. Since you've already did it, that's OK.

[distant lichen]

Ticket ID:Case#03203888
Plan:ChatGPT Plus (Paid Subscriber)
Issue Start Date:Monday, Nov 24th (6 days ago)
Description:
I am experiencing a critical systemic account bug that prevents me from using ChatGPT entirely. This is NOT a generic outage; it is specific to my account.
Status:

1. I contacted Support via Help Center on Monday.
2. I was asked to provide logs/data on Thursday, which I sent immediately.
3. It has been 48 hours since I sent the data, and I have received no response or fix.
4. My work is halted.
   Request:
   Could a Community Champion or Staff member please flag this ticket to the Engineering team? The ticket seems to be stuck in the queue while my subscription days are wasted.
   Thank you.

[distant frost]

Hello! Is it possible to add with OpenAI to a png with transparent background shadows to the image (with transparent background) ? I cant find a option for it. I would be very happy if someone could tell me how to do it or where i can find the option! Thank you very much!

[meager rune]

Hi! Could any moderator from the OpenAI community confirm if this is legit or am I being scammed? I received this e-mail a few minutes ago... Telling me that I can have GPT Go free for 12 months....

[trail cairn]

https://help.openai.com/en/articles/12739021-chatgpt-go-promotion-india

[meager rune]

Thank You

[upbeat schooner]

can someone tell me why i have stopped seeing the Thinking tool on my Android app? it is however visible on the website via desktop. is this a known/usual issue?

[limpid raven]

click model drop down menu

[upbeat schooner]

it's the GPT model, not the tool. when tapped, it only gives me the 5.1 option.

[onyx kraken]

For me on Android, the top 'ChatGPT' has choices, but also says it's 5.1 while just sitting there.

I'd consider making sure I have the most recent ChatGPT app update, and restarting my device.

If that didn't resolve it, I'd reach out to https://help.openai.com/en/articles/6614161-how-can-i-contact-support

[round mango]

The option that creates GPTs simply disappeared, does anyone know what to do?

[charred parcel]

can you please remove this, it's spam

[onyx kraken]

I would expect the right side of that page to have a 'my GPTs' button and also a '+create' button.

Maybe try these direct links, https://chatgpt.com/gpts/mine https://chatgpt.com/gpts/editor

[onyx kraken]

I'd also check to verify if you're signed in.

[round mango]

The button still doesn't appear for me, but I managed to go through the direct link, thank you!

[onyx kraken]

#1070006915414900886 might be useful to mention that, if you're willing, and perhaps give a screenshot of what you see. I can't, as I'm not affected by the bug you describe, at least not yet!

These channels up here are mostly for members to chat with other members, but gold-named OpenAI employees sometimes even answer in #1070006915414900886 and #1070006151938314300 . I very rarely see posts from those up here in the channels we chat in, I don't think they track our steady stream of chatter.

[round mango]

ok, thx

[buoyant oyster]

Yes, the thinking tool vanishing is a known thing. In order to troubleshoot, you could try to update the app in the Play Store or wherever you downloaded it, or clear the app's cache. Or, you could log out and back in as well to your account in the app. If these don't work, it might just come back on its own after a bit. Hope this helped you!

[rose gull]

I have a problem with SORA 2. I am on pc and on website. I generated 4 videos and theres nothing moving, just first frame of the video and sound. Who knows what is this? But if i would download the video, the video is working.

[split swallow]

it says this whenever i try to send a message? i can’t update the app btw

[drifting hinge]

Did you also check internet connection? You may eventually refresh the app/website or reopen it.

[dreamy sable]

I subscribed to Pro this morning. I do not have an option to remove the watermark in Sora 2. Non-cameo videos with original content.

[onyx kraken]

That maybe is related to downloads of the video, and not 'removing the watermark' specifically, per this helpfile:

https://help.openai.com/en/articles/12460853-creating-videos-with-sora

[dreamy sable]

I used a prompt image in order to keep consistency. Let me try without the image.

[dreamy sable]

Watermark is in fact gone when not using a prompt image. At least that's solved for. Need to figure out consistency next. Thank you!

[onyx kraken]

Glad to help!

[fathom rose]

Is it true, that if i explain more to the ai the better my results get

[fast crest]

Hello everybody. Can anyone give me a helping hand on my Builder profile verification?

[onyx kraken]

It depends on what results you're trying to get.

If you want it to solve some math problem, and you know exactly what the math problem is, you probably don't need to explain much.

If you want a personalized, tailored-to-your-preferences answer, maybe a set of suggestions, then yes, giving it the information it needs to personalize and tailor to you helps - otherwise it has to guess and might guess quite off what you would like.

[onyx kraken]

I can try 🙂 I've not done this personally, but I can point towards the helpfiles that may relate.

https://help.openai.com/en/articles/8871611-domain-verification

OpenAI tech support can be reached here, https://help.openai.com/en/articles/6614161-how-can-i-contact-support

We're nearly all community members here in the discord, like yourself, so we may not have the answers or ability to help that you want - but you can discuss issues (protect your personal info though) and see if someone knows how to help, maybe they had a similar problem and know a solution.

[fast crest]

Alright, thanks for the response. I have actually read all of the topic-specific OpenAI files.

I know that there are two ways of a user verifying their Builder profile. Either through matching their billing information with their ChatGPT account information or through buying and verifying a domain.

I want to proceed the first one.

[onyx kraken]

Great, you know more than I 🙂 If want to, and can, discuss whatever problem, but without revealing your PII, maybe someone can help.

It's possible you do need OpenAI tech support though, and that's where you can discuss things like your PII, since those are OpenAI tech support or related contractors.

[fast crest]

I am using the same Email in both my ChatGPT Plus subscription and my ChatGPT account. So that's already matched. And I am also trying to match my ChatGPT account username with my PayPal username, since I'm using PayPal as a payment method. And the problem is that, although I have managed to also match my usernames, my Builder profile remains unverified.

[onyx kraken]

I'd definitely reach out to OpenAI tech support.

[fast crest]

I already have. 😅

They guided me until this step. I'm thankful.

But now.... I don't know what can be done, honestly.

[onyx kraken]

I hope they answer soon!

I'd keep checking the help page, looking at the chat, as I have not always gotten an email when they replied.

Sounds like you're getting help - either through the automated bot, or a human. If the help you got before was through the bot, then it's just a wait if you've let the chat know that you still need more help, and if the chat is still open (once you get a 'rate the chat' option, I believe that chat is closed, you'd have to start a new one - that was as of my last tech support request, a while ago).

They do have around 800 million weekly members using ChatGPT (and they have API and other services too, so probably well over 800 million weekly interacting people) and if a good chunk need help around the same time, it can take a while for human tech support to be able to answer - usually hours to weeks for me, probably depending on how busy they are right then + if the tech to help you needs specific training for that niche need or not.

[fast crest]

Well, thanks for the advice. I have actually received guidance from the human support team, several times.

[onyx kraken]

Great! Hope they can help you with this one soon.

[fast crest]

I also do.😅

[onyx kraken]

And/or that another community member who's explored a similar situation and found a method answers you too 🙂

[fast crest]

Thanks man.

Can I ask whether you have created a GPT yourself or verified you Builder profile?

[onyx kraken]

I make custom GPTs, yes. But I have not chosen to verify my Builder profile.

[fast crest]

So, I guess you haven't shared them in the GPT-Store, right?

Ok

[valid charm]

What channel can I speak to someone about a subscription I barely was able to access so I can cancel my subscription I’m being charged no fly and I’m not even using the app. Is there somewhere I can contact someone for help and hopefully refunds? Been getting charged since August but couldn’t locate where it was coming from. Please advise.

[fast crest]

https://help.openai.com/en/collections/3742473-chatgpt

[fast crest]

There are OpenAI articles on canceling your subscription and receiving refunding on this page. You can also contact their support team.

[buoyant oyster]

It’s a known issue with Sora’s web preview. The videos you generate are fine, but the website’s preview player can’t render the video stream, so it only shows the first frame with audio. Once you download the file, you get the actual MP4, which plays normally.

Most likely, this is caused by browser playback limitations or a bug in the preview player. Nothing is wrong with your videos — the preview just isn’t showing motion.

[onyx kraken]

Here's how to contact support, https://help.openai.com/en/articles/6614161-how-can-i-contact-support

Those of us in this Discord with you, we're nearly all community members like yourself. The very few who work for OpenAI have gold names, #server-staff .

We can talk about buggy stuff and fixes we've noticed, and help share the helpfiles and links to tech support, but for a billing issue you definitely need tech support, nobody here could help and don't share your PII or financial info for sure.

If you struggle to get your subscription canceled (I don't expect you will have trouble, but you mention it's already been a few months of charges you didn't understand), or if you ever have some undesired or unknown charge from anywhere, your financial institution should be able to help you find out where that money is going, and can cancel transactions you don't agree with or never agreed to.

I would explain how you ended up paying for this, when you are talking to OpenAI tech support (not here, none of us can help and it's not our business either, we're just community members like yourself here).

If someone took your financial info and used it to fraudulently afford their subscription or something, I would definitely discuss that with OpenAI tech support because that's not a normal membership situation at all and perhaps there's additional refund options (I have no idea, that is not in the helpfiles). This is the standard info about refunds: https://help.openai.com/en/articles/7232895-how-do-i-request-a-refund-for-my-chatgpt-subscription

[wheat crane]

i ran out of gens but its still letting me make videos am i paying money or is that for like sora pro or sum

I have plus but not pro

Im not paying 200 bucks a month

[errant vessel]

could you help me out actually ive been testing a new benchmark i made using gpt 5.1 thinking but on the API its like broke or smt, and i have just a few questions left if i send like 5 prompts to you could you send the respone?

[wheat crane]

Yeah sure

[wheat crane]

Whats the prompt

[errant vessel]

i dmed u

[wheat crane]

Oh

[odd prawn]

Someone has created a clip which cameoed me and couple others, that clip showed up in my Drafts. I deleted that clip in my Drafts. My question is, does that clip also get deleted from the Drafts of the creator of the clip (and the Drafts of the other guys who are cameoed)? And let's say if this creator had posted the clip already and I block him/her, would the clip still show in his/her posts?

[onyx kraken]

Per https://help.openai.com/en/articles/12435986-generating-content-with-cameos

Those you block cannot make NEW videos with your cameo. Additionally,

Set People I approve and only approve people you trust. You can remove access at any time in Cameo rules.```

```Can I see all the uses of my cameo?

There is no private use of your cameo that can’t be seen by you. If someone uses your cameo, you can see it.

Can I remove my cameo from a posted video?

Yes. Open the video and use **Report** to request removal. Our team will review and take action if it uses your cameo against your settings or rules.

Does deleting my cameo stop future use?

Yes, new videos can’t use it. Existing videos aren’t changed; you can delete any video with your cameo in it.```

You specifically mentioned deleting the clip from your drafts.

If someone else's account was used to create that clip, it appears you need to use the **Report** tool on that video to request it to be deleted.

I'd make sure nobody else has had access to your account, if that video appeared in your drafts.

Granted though, I have not explored having a cameo yet - we can see videos that show our cameo - I don't know if they appear in our drafts.

If they do, mystery solved and not a concern.

But I'd be careful about 'delete a draft, delete the video someone else made, that's not clear because the report tool is specifically named as being involved.

[green summit]

Once deleted from drafts, it will also be deleted for the creator.

Once posted, go to their posted clip and click delete.

Not sure about when you block them. I don't think it will delete the clip.

[robust wing]

I’m trying to understand how people working with AI workflows or agent systems handle things like unexpected model behavior, reliability issues, or debugging steps.
Not looking to promote anything — just genuinely interested in how others structure their process.
What’s the most frustrating or time-consuming part for you when dealing with these systems?
Any experiences or insights are appreciated.
I’m collecting different perspectives to compare patterns, so even short answers help.

[edgy mauve]

Anyone building/exploring the apps sdk for chatgpt? Had some questions regarding the same

[charred parcel]

what do you need to know?

[edgy mauve]

1. My app has a use case of making payments to complete purchases. Will that be allowed as part of the chatgpt app?
2. Is it allowed for an app to reference another app. Eg Can an app of an online retailer have stripe's app(mcp +UI) embedded for collecting payments? How can this be done?

[charred parcel]

policy questions should be directed to support at https://help.openai.com

[edgy mauve]

Sure. Both are policy questions?

[charred parcel]

yes

[livid cape]

Hi all

can anyone help me

i had a chat with a community rep

about receiving warning emails from chatgpt (when i didnt do any bad thing?!)

and today i received one again!

i am very worried about my account because my academic research is tied to the account

is there no way to get human support for the biggest AI company in the world?

i am a paying customer

[polar birch]

Hi, I want to know how do i make food into become animal? I can't find one good site

[distant lichen]

Bumping this thread as it is now Monday morning.
My account is still broken (7th day).
Could @staff please take a quick look at Ticket #03203888? I'm completely stuck.

[cinder marlin]

Does anyone know how many images I can upload with the Plus version for €24 per day?

[heavy patio]

The community has no power to help you. Support staff from OpenAI are not in the discord server, the few staff that occasionally stop by are not front line support. The server mods and guides are volunteers. Your best channel for official support is by replying to the ticket itself via email

[ripe lark]

i'm guessing group chats aren't turned on for business plans yet?

[livid cape]

i appreciate this, but the emails seem mostly automated

its shocking that OpenAI does not have live support, considering the fact that they are the groundbreaking firm in LLMs

[heavy patio]

I don't think there's a limit

[next fern]

anyone else unable to download files from chat like pdf zip etc

[heavy patio]

Personally speaking, working with most companies support teams is fairly annoying

Have to psych myself up for the task, prepare for the worst. Remind myself to take a few deep breaths and be kind to others before I start. Always surprised in the usual moments when its easy.

[livid cape]

i totally understand your concern

but for me, the issue is rather serious (if you check my previous messages)

i have been using ChatGPT for 3+ years now, paid customer since 1 year

and since October, OpenAI has been sending me a "fraudulent activity" warning email every few weeks

if i cannot have a human from the OpenAI team helping me, i might lose some VERY VERY priceless data (i take help from chatgpt for my PhD research)

the way chat has learned my profile over the few years - i might lose all of that, just because of something going on in my account which i am not properly aware of

PhD is already very high pressure, and now i am suffering from internal anxiety - since chatgpt is like my research assistant, alot relies on it!

so tell me, who can help me? an automated email response or a human with some empathy? - i personally have never ever misbehaved with any support staff in my life, i am always professional even when the outcome is not what i was hoping for

i would just love it if a human from OpenAI could be with me on the phone or chat & take a look through my logs (i fully consent to my data being viewed by them) and find out what exactly is causing the warnings - I just dont want to lose my account, there's priceless data and there's a unique way the AI has learnt me and my person - i want to hold on to that

[heavy patio]

I get it, my PhD coursework and research was pretty intense, even my side research today gets pretty involved. I'm empathetic to your problem, but it doesn't change the fact that no one here can help you. It's a community server, this is a community help page. If the emails are legit, really only the OpenAI help team can help you fix it.

As a fellow community member and researcher, I would urge you to fix your single-point-of-failure tech problem by bringing in other research tools. Put those annotated bibliographies and notes per paper into mybib, and use other AI enabled research tools to help diversify your research portfolio. Keep your research notes backed up on some sort of personal cloud drive.

Technology goes down. Companies fail. Data is lost. Accounts can just go away. Set yourself up for long term success by designing a researching tool support ecosystem that can sustain all that.

[livid cape]

i have loved using ChatGPT's thinking (with the Extended Thinking option) - do you have any recommendations?

one of my main qualms is sifting through hordes of literature (i don't know if this is an undiagnosed ADHD or something else with my concentration span - i'll get it checked once i'm done with my PhD). and brainstorming new ideas with ChatGPT

if you can suggest any tool that is similar to ChatGPT's Thinking model (plus has the capability to browse and search the web) - i will be grateful as a fellow PhD scholar!

anything that is good at reading stuff, learning from it and suggesting ideas - is just what i need as an AI research assistant

i am willing to pay for the right software, if it can help me!

and thank you for your empathy, i trust you understand my struggle as you've gone through the same thing. I appreciate making your connection here!

[heavy patio]

I really like both OpenAI & Anthropic for annotated bibliographies put I use mybib to store those outputs per page. I use annotated bibs to start to gather a body of research together, it's kind of how I begin to sit with it, chew on the research over and over a lot?

Also I heard Google Scholar offers a browser extension for PDF reading that's designed for reading research pdfs. It's new and I haven't tried it yet.

It doesn't have to be annotated bibliographies but that's just how my mentor taught me to stay organized. I did my PhD program prior to AI being useful, but in my current research (for me personally) the important thing is to use the AI for learning the research - because the vocabulary in the papers is harder for me to digest - but pull the learning out somehow into a different place.

That way you can take it to a different AI model and ask some more questions, start to get a more well rounded understanding of the research being presented.

I'm working on a paper right now where I have a co author for the first time. We're researching together, and her style is a little different - she's neurodiverse but a different spice than me, LOL.

She highlights sections of the PDF in a shared Google Cloud, then puts like AI Tl:DRs as comments to the highlighted section. I like it a lot, easy to work with her on it and chat about things together, also easy for me bring new research in to approach on her level.

Perplexity AI has a "research" button, Claude (Anthropic) calls it "extended thinking" I think, Google calls it "thinking". Most models can get you similar but slightly different tuned outputs. Reccomend to explore, test & diversify your research AI portfolio

[distant lichen]

Understood, I have already contacted support but no response for a week, just hoping a staff member sees this. Thanks.

[exotic ravine]

Yeh fair enough. Thank you for the tip

[radiant field]

I have been using a chat for more than two months, and now I can only open it on my phone because opening it on my computer is extremely laggy, nothing loads, and I can't even share the chat. What do I need to do to reduce the lag?

[charred parcel]

start a new chat, long chats get slow

[radiant field]

But if I create to another chat, it will forget all the information I gave it before

[charred parcel]

ask ChatGPT to summarise the chat and then place that in the new chat

[radiant field]

I’ve given it a lot of information, so I think summarizing everything to transfer it to another chat is impossible imo

[charred parcel]

ChatGPT only remember the last 32k tokens if you are on the Plus plan, so you won't be losing anything

[radiant field]

i will try, ty for suggestion

[glass gorge]

So.... When does the new Shopping Research feature come out? Are they still rolling it out or...?

[sacred flower]

If I want to share an ai music video that i'm working on, can I do that on anywhere?

[charred parcel]

in #off-topic

[fiery cliff]

I logged out a little before midnight cst USA, earlier, and logged back in about 30 minutes later. I'm on my PC and when I put my cursor in the chatbox, all these pre-K, cutesy animations appeared below to pique my thought process but instead they just irritated me and I'd like to know how to put in a complaint to see about having them removed?Every time you take the cursor out and put it back to start a chat, they change. It's RIDICULOUS. They are completely unnecessary except for those that just have to have them like grandma and the babysitter. PLEASE?? I don't know why that says "MAGO" next to my name.

[sudden compass]

Why can only the Go version be purchased on the mobile app, but not other versions?

why?

[unreal hemlock]

<:book_icon:1363314738255364126> Rule 3: Stay on topic.

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

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

[silk berry]

Hi, having an issue with a ChatGPT APP. Sometimes when ChatGPT provides an alternative response (two version of an answer) it breaks the app. Has anyone experienced this and have a solution?

Additionally, has anyone noodled with shopping research? Does this leverage MCP servers? Can it?

[ripe mason]

I can’t access my account, I got a message saying unusual activity was detected on my device, tried reinstalling the app and now can’t get into anything!

[safe hill]

same here

[livid cape]

whats happening? chat gpt down?

[cosmic mantle]

My chatgpt can’t load either mobile or website

Can’t even submit a bug report lol

[onyx kraken]

You can reach out to OpenAI tech support through https://help.openai.com/en/articles/6614161-how-can-i-contact-support if you want to try that.

A few other members in other channels are discussing some odd problems with connection issues, maybe we will see #status show that there's a problem soon - or not, since it hasn't happened yet.

I have normal connectivity to ChatGPT, fine for me.

Some options to try, if possible:

Restart the device itself. You mentioned reinstalling the app, but sometimes restarting the device does the trick.

Different device.

Different way to connect to the internet.

Different interface option, between app, browser on one or more different options (chrome, firefox, others; a fresh install of Firefox on computers tends to work for some when others don't).

This can help you narrow down if the problem's specific to your equipment.

[crisp sand]

Chatgpt is down

Previous chats do not show up, account does not show up, asking does not work.

They're back

[rotund cloud]

projects missing

[crisp sand]

Happened to me after a while they appeared back, chatgpt 5 was working fine earlier but got a bit slower and this issue started.

[rotund cloud]

now it is saying too many concurrent requests

which is false, it is broken

oh, now projects loaded but not chats

[pseudo vortex]

anybody here knows what this error means?

[loud bear]

how do i log out of this

theres no button or dropdown i can see and i cant switch accounts or get out of my workspace to get to my personal

other than using a different device

even using log out from all devices it doesnt work either

[onyx kraken]

This is the associated helpfile, https://help.openai.com/en/articles/8801848-how-do-i-switch-workspaces-between-personal-workspace-and-my-chatgpt-business-workspace

I'd consider clicking on purchase subscription in case that page has a way to click on your profile icon, which is what you need to do per the helpfile to switch workspaces.

You can also reach out to OpenAI tech support, https://help.openai.com/en/articles/6614161-how-can-i-contact-support

[loud bear]

appreciate it alot tyty

[loud bear]

unfortunately there is no profile button on either page

[onyx kraken]

Sorry to hear that. I'd definitely reach out to OpenAI tech support for help, and you can also describe it in #1070006915414900886 here on this discord if you'd like to.

For the vast part, we're all community members chatting and helping each other here, the very few with gold names work for OpenAI, #server-staff . They probably do not track these upper channels, but they sometimes answer in #1070006151938314300 and the bug reports channels.

[exotic ravine]

Is this considered "glitchy behaviour" aka hallucination

I don't even speak Chinese lol

[rustic marlin]

looks like it indeed

[onyx kraken]

Essentially, I'd consider it a type of that, could be pretty mild in that it may have only grabbed the wrong token to put in a string of code - likely intended to be a formula shown to you, nice and pretty but needs exact coding, and likely that first not-English was the error, it 'closes' with that same character.

[rustic marlin]

when a LLM gets out of the rails and outputs random things like that, the best thing to do is to start a new chat

[livid mulch]

Hello, do people in Latin America not have access to Sora 2?

[fast crest]

I did it, guys.

[soft gate]

When i go to make an cameo is says “Something went wrong” this has to be a glitch because it says i did it right

[onyx kraken]

For now, that's the case, Sora 2 and the app are only available right now in 7 countries, https://help.openai.com/en/articles/12461230-sora-app-and-sora-2-supported-countries

[livid mulch]

@onyx kraken thx

[icy surge]

Where do I see answers to AMA on Reddit ?

[next stump]

Question, so I have used sora for a while but one thing i need help with is unblocking users i have previously blocked, but there isnt even an option to manage your ban list, can someone assist?

[onyx kraken]

https://help.openai.com/en/articles/12461090-blocking-and-reporting-on-the-sora-app

Does this help?

[next stump]

this doesnt even exist

[onyx kraken]

Are you in Sora 1 or Sora 2?

[onyx kraken]

Exploring more, I may see what you're saying though. I've never blocked anyone, so I can't fully explore - but I am not seeing a 'privacy & safety' setting in any of Sora 1, Sora 2, or ChatGPT itself (no idea why it might be there, but I would expect it to be somewhere.

[onyx kraken]

I note the helpfile there does specifically say 'The Sora app', and I'm using website to access Sora, unsure if that is why I don't see it.

[next stump]

I checked both and it doesn't exist in either or any version, so i imagine they need to add it @onyx kraken

[onyx kraken]

Do you have the Sora app? That's something I can't check, and that's specifically the name of the helpfile that discusses unblocking through that settings - privacy and safety option

[next stump]

Yeah I have it and it isn't there, or do you mean a sora app on my desktop

[onyx kraken]

You can #1070006915414900886 the issue, the channels up here where we chatter are for community members like ourselves to discuss, it's not clear at all that OpenAI monitors these channels, but both #1070006151938314300 and that bug report channel sometimes even get answers from the gold named OpenAI employees, #server-staff .

By Sora app, I believe that is specifically the Sora iOS app, https://help.openai.com/en/articles/12456897-getting-started-with-the-sora-app

This shares more about it, including a link: https://openai.com/index/sora-2/

But it is only for iPhone, and I don't have one.

[next stump]

thank you thank you

[onyx kraken]

You're very welcome, and I don't have more ideas, except perhaps #1070006151938314300 that a way to un-block be added for those who don't have an iPhone.

[hasty olive]

i am sure this has been brought up before but this lag when having long conversations with chatgpt (web version is what i use) is sooo fustrating. Chatgpt people, please fix that before you do anything else. sometimes it almost becomes a deal k`ller for me. from what i have read is because the front end is react. i dont know, i dont care. I need the slowness in long chats to stop. i need it faster.

[river cloud]

Why don't I get the option to download...?

[river cloud]

It doesn't appear in the desktop too...

[green summit]

Click the file in the message and it'll download.

Or should*

[sudden compass]

maybe you can try it on the web or ask the question again on the web

[charred parcel]

soliciting is not allowed on this server, see #server-rules rule 7

[rustic marlin]

there was never a download, the AI is hallucinating

[stray pollen]

Is the Sora 2 website down?

[river cloud]

there isnt any file in the message its just a text like this:...
📄 ADDC2025048_Horizon_Report.docx
📊 ADDC2025048_Horizon_Presentation.pptx, and it pretends it a to be a download option 😐

[river cloud]

i did try on the web, but it didnt make any sense...

[river cloud]

download option do exist on ChatGPT (or did exist* 🤔), i actually remember of downloading .pdf files of the replies/answers that it provided me...

[rustic marlin]

yes, I need more cotnext to help tho, but what I mean is that the AI is hallucinating in the sense of not producing a correctly formatted clickable link, or more likely, it thinks it have a file, but in reality it does not, so the link does not work

what I mean is that there are several things that could go wrong in the process that cause the AI to output like that when it says a link is clicable, but it is not, and several of those steps can get go wrong because of AI hallucinations

for isntance, Im guessing you isntructed the AI to do something and give you a .zip file with the data.. however, soemwhere in the process, the AI probably didn't made the zip file, despite telling you it did, so.. on this message thread, there was never anything to download in first place

when the AI hallucinates like that, either start a new thread or, go back several messages to the point you thing it begun to hallucinate and start again fro mthat point

[rustic marlin]

the AI didn't run anything on the system, it just formatted it like a code block xD

tell it to use python to run a ls on the directory, then, look at the output of the python script, the output of the python script is never AI generated,

[river cloud]

it wasnt actually just a zip file that i requested; at first, i asked for a .docx followed by .pdf & then .ppt at last i just asked for a .zip but it created none

even after asking repeatedly in different manner

[river cloud]

it did generate something like that too

[rustic marlin]

if you ask the AI to do something and it hallucinates an output, don't send more messages, edit previous messages to prevent it from hallucinating

when the AI starts hallucinating it will already be primed to keep doing it wrong, followups will only make it worse

[river cloud]

i see

[river cloud]

start a new thread
but then if i start a new one, i should explain the entire content back to it rite?

or does it link to the previous thread and generate the results...?

[rustic marlin]

the cleaner solution is always to start over

a bit frustrating, but relying on the memory to trasnfer the context might prime it to hallucinate again

in fact, for this case, it might be a good idea to disable memories as to not bloat the AI's context with information unrelated to the task

[river cloud]

well, thx for the info ❤️

[rustic marlin]

happy to help =)

[mellow marten]

Hey, is anybody getting the "Stopped thinking" after chat GPT spends 10 mins - 12 mins reasoning?

It's like if the reasoning phase takes more than 10 to 12 mins, it is likely that it just stops and show "Stopped Thinking" and no answer.

[charred parcel]

yeah, I was getting that on the macOS app earlier

[mellow marten]

Well, at least it's not only me.

[mellow marten]

Thanks for responding

[hollow delta]

does api access to sora 2 pro require a chatgpt pro subscription?

[trail cairn]

nope, anyone can use it from almost anywhere as long as they verify themselves with a gov id

[hollow delta]

that's what i thought as well, and i even got to test it in the dev playground. but when i tried to use it via API, i kept getting a "billing hard limit has been reached" error

although my budget and limit are much higher than what i've spent so far

[trail cairn]

projects also have their own limits on top of account ones

[hollow delta]

my project's budget is the same as my account budget though

do i need to explicitly allow sora 2 pro model and set its RPM to the max?

[trail cairn]

billing errors are unrelated to rate limits

make sure you are looking the right place and not the dashboard with monthly usage

[hollow delta]

i was on the dashboard, looking at both the organization tab's billing and limits and then the project tab's limits

are those the wrong places to be looking?

[trail cairn]

only dashboard is. does the error show more details if you use curl or just raw call, no sdk

[hollow delta]

hmm no errors with curl, seemed to work fine

[trail cairn]

what did you use when it failed?

[hollow delta]

im so dumb, had an old api key in my system environment variables that was overriding everything

everything works now, thanks for your help!

[real wave]

since 5 days same situation tried everything reedit openai chat support every possible way working without login any help ?

[vernal wyvern]

Do i just report thos to trading standards along with the failed to cancel screenshot from 7 month ago

Don't worry I'm ADHD with real bad memory. I did try cancelling 6 month ago but then got same problem and forgot. Due to my disability I forgot all about it

[cerulean stirrup]

Hello. all my messages to support are being ignored on both the desktop help feature and via email for over 3 weeks now. Please can someone help me

[rustic marlin]

the only place to get in touch with OpenAI support is at https://help.openai.com via the intercom widget at the bottom right of the page

[supple steeple]

Deepseek?

[wet flower]

“How do we use our cameos in Sora2? Is it enough to just use their names in the prompt?”

[weary rose]

hey , can i get the chat gpt 5 free trial without putting my card details on the app?:

[torn ruin]

Is it just me or ChatGPT is broken atm? Been thinking for more than an hour now, no replies

[gloomy trout]

Someone please help me!
When I try to generate a video, Sora shows this error:

Oops, an error occurred.
"Failed to execute 'removeChild' on 'Node': The node to be removed is not a child of this node."

Does anyone know how to solve this?

[charred parcel]

are you using a VPN?

[gloomy trout]

Yes.

[tawny delta]

Need help figuring out where to start with an AI-based iridology/eye-analysis project (I’m not a coder, but serious about learning)

Hi everyone,

• I’m a med student, and I’m trying to build a small but meaningful AI tool as part of my research/clinical interest.
• I don’t come from a coding or ML background, so I'm hoping to get some guidance from people who’ve actually built computer-vision projects before.

Here’s the idea (simplified) - I want to create an AI tool that:

1. Takes an iris photo and segments the iris and pupil
2. Detects visible iridological features like lacunae, crypts, nerve rings, pigment spots
3. Divides the iris into “zones” (like a clock)
4. And gives a simple supportive interpretation

How can you Help me:

• I want to create a clear, realistic roadmap or mindmap so I don’t waste time or money.
• How should I properly plan this so I don’t get lost?
• What tools/models are actually beginner-friendly for these stuff?

If You were starting this project from zero, how would you structure it?
What would be your logical steps in order?

I’m 100% open to learning, collaborating, and taking feedback. I’m not looking for someone to “build it for me”; just honest direction from people who understand how AI projects evolve in the real world.

If you have even a small piece of advice about how to start, how to plan, or what to focus on first, I’d genuinely appreciate it..

Thanks for reading this long post — I know this is an unusual idea, but I’m serious about exploring it properly.
Please DM me if possible

[heavy patio]

If you are student, you really need to talk to your teachers that have higher education degrees, they have the knowledge you seek. The technology part is the easiest part of your desired project, in my opinion. The hardest is to get approval for what you want to do, as Human studies and research has a lot of controls and ethics involved, and usually needs to be reviewed & approved by an IRB (international review board).

[sudden compass]

The price in the Malaysia region is one cup of coffee more expensive than the original price of 30 USD🙃

[tawny delta]

That part is easy for me! As this will be my thesis to be submitted for my uni
I have 2-3 projects in my mind! Felt this is unique and worthy no one ever tried so looking for feasibility

[heavy patio]

Ah, ok you will probably go through IRB naturally then. Masters Thesis?

[tawny delta]

Ug

[heavy patio]

Image Machine learning is fairly straight-forward. Your teachers can help you find the datasets, but basically just grab a couple thousand photos of eyes and spend a lot of time spent on each photo to mark out to tell the computer what is where. Then you hit submit, wait a bit and the AI service can output a model with "understanding"

The more time (and photos) you spend marking things, the smarter and more accurate your resulting learning will be.

Its been a minute since I used visual machine learning but last I checked, Azure, Amazon Web Services and Google Cloud all offer an machine learning service for images.

[tawny delta]

What will might be the cost of it! Approx

[heavy patio]

I have no idea. But getting grants for compute funding is fairly straightforward, once you have everything in place.

The school should have either a system for compute grants or support resources in place to help you get grants from outside sources.

[tawny delta]

Potentially me marking down bounding boxes to show this is this, will get the job done! other than that like this will be the primary part

Right

[heavy patio]

There's plenty more because nothing is ever easy, but pretty much yeah. I think it's common to have to go back and modify the validation set again, because of something you learn when you evaluate the first output? I would personally plan to fail and have to do a couple rounds in the process before I expect see a level of success I am happy with.

[tawny delta]

Is there anything like books or any sources that I can learn to get a grasp of what is what!
Cuz I don't know what is what! Some how I can drive through stuff but without knowing what is the start or progress point or milestone kinda behaviours is missing for me a roadmap!

To connect them what should I do!

[heavy patio]

Check out youtubes with the step by steps. I think tensorflow has a couple. Stay away from using "AI" in searches, you want to use the terms "Image classification" or "Image machine learning"

[tawny delta]

Also failing is the part I like the most! Cuz I like to be challenged! I wanna crack something up! But without knowing anything will be a mehh

[heavy patio]

Good luck!

[tawny delta]

Thanks

I got something to row with

Means a lot

[cerulean stirrup]

Hi Lugui. I've been trying to contact them on there for weeks now and I don't get any response. Thats why i'm here. I don't know what else to do

[rustic marlin]

are you sure you oppened a ticket?

that intercom thing can be very frustrating

but if you get apst its article suggestions, it should start a ticket, which is replied by a real person

[mental thistle]

Hello! I had some questions for this hobby project I’ve been thinking about doing for years and now I feel like is a great time to utilize AI to help with what I was looking to accomplish

[quasi bluff]

Is Sora 2 already in Brazil?

[charred parcel]

no

[onyx kraken]

If you'd like, you can track https://help.openai.com/en/articles/12461230-sora-app-and-sora-2-supported-countries which shows as they add more countries.

[toxic helm]

Hello there! Does someone knows that why can't I use Sora ai even tho i have a business plan?
I can't seem to find any solution to this, I contacted the support but they said nothing useful

[mental thistle]

Hay I was curious if it’s possible to integrate a logical thinking AI that can have its own beliefs and opinions with multiple perspectives and discussions and actually struggles to choose the correct logical reasoning for a particular situations into a game..?

[magic flame]

Can someone help?

After i deleted my data in sora 2, it suddenly locked me out, and when i try to log in it wouldnt let me and kept on saying that i dont have access anymore.

i dont wanna stay locked out of sora 2 forever

[charred parcel]

what country are you in?

[magic flame]

the united states

[charred parcel]

then contact https://help.openai.com for assistance

[magic flame]

ok

welp, i guess i have to wait a whole month to have sora 2 back...

And it also thinks that its not available in my country

[charred parcel]

are you using a VPN from outside the USA?

[magic flame]

any vpn i find is extremly expensive and i have no clue how to set it up. (im broke and i have no money. im only 15)

Is there a free VPN chrome extention i can use?

nvm i found one

[charred parcel]

I think the issue is that OpenAI think you are outside the USA

[magic flame]

oh

also i already found a VPN i could use

[charred parcel]

OpenAI deactivate or ban accounts using a VPN to evade geographic restrictions

[magic flame]

oh

ill probably not take the risk then

[exotic ravine]

I couldn’t reliably edit the DOCX directly on my side just now

sigh - anyone else having trouble right now

[main bison]

Uh have I been compromised. My name disappeared

[sudden compass]

Why is the price in the Malaysian region so high? The Pro tier is $43 more expensive than the original price.🙃

🙃

[twin gulch]

can i opt in for pay per use/pay monthly instead of pay as you go for api ?

[cloud quail]

Help. Invalid Client when attempting log to ChatGPT on Brave

I already deleted cache and its still not working

[rustic marlin]

the API is not a subscription service, there is no such option

[rustic marlin]

try disabling extensions, as a sanity check

[cloud quail]

Sure

Didn't work

[rustic marlin]

is the problem only on brave?

[twin gulch]

i understand, but i mean where i am billed per use and i pay only for what i used at the end of the month

it only looks like i can prepay

[rustic marlin]

on the API, you are billed by the exact amount of tokens you used

on chatgpt, you pay a subscription, the value is the same weather you max its limit every day or if you barely use it

on the API, you pay exactly how much you used, if you don't use it, you pay less, if you use it a lot, you pay more, as simple as that

[rustic marlin]

you gotta add credits to your account

so, if you want to spend a maximum of 20 bucks in total, you add that amount of money to the account

and if you reach that amount of usage, the system will simply stop working, it will not charge you more unless you enable th option to automatically purchase more credits

[rustic marlin]

those 20 bucks will sit there until you use the equivalent of 20 bucks, does not matter if you take a day or a few months to use that

[cloud quail]

Nah, I solved it

I just had to go to OpenAI.com and hit the login button LMFAO

[rustic marlin]

lol

yea, that helps :V

[mighty garden]

Mine is very strange... I done last night the trend with the year summary of gpt... It said my ip location as per the vpn and a local time that was exactly the time zone of the country im in. Then it said "probably you're using vpn 😁

[drifting hinge]

Why is this happening

[fading osprey]

watz the issue

[grizzled mantle]

Guys I need help so what basically happened is that I'm trying to create an anime video completely harmless but they keep showing me this if though I stated that he's not a minor I even changed his pic multiple times here is how he looks and what they told me

[drifting hinge]

It tells me it's a gpt 4o model.
But it should be gpt 5.1 actually

[fading osprey]

do u have pro

[cerulean stirrup]

Hi Lugui. Yes i've opened many tickets, all have gone unresponded to for weeks, and i've emailed [e-mail] and [e-mail]. I haven't had any response from any of them and ChatGPT keeps adding charges to my account I haven't agreed to or authorized

[charred parcel]

what are you being charged for?

[cerulean stirrup]

An additional business seat I never took out and never agreed to keeps being added to my account. They have said they will refund me for it for months, but they never do and then the additional seat keeps being re-added and then I get recharged for it again. Its been going on for months now. And now i'm just being ignored.

[mossy prairie]

contact alex or smth

or @burnt galleon

[charred parcel]

how many seats should you have?

[cerulean stirrup]

Who are Alex or Smith please?

[mossy prairie]

alex is the owner

of this

server

@cerulean stirrup

[cerulean stirrup]

It won't let me add him as a friend or message him

[charred parcel]

server staff can't help with billing issues, you need to deal with support

[cerulean stirrup]

But they arent responding

[charred parcel]

let me try to help, you have a business/teams account yes?

[cerulean stirrup]

Yes thank you so much

[charred parcel]

how many seats are you supposed to have?

[cerulean stirrup]

1 seat. I signed up during a limited promotional offer that allowed some users to sign up with just one seat

[charred parcel]

ok, and you have two seats now?

[cerulean stirrup]

Yes. This is what happened:
• 9 August — I started the single seat promotional plan.
• 11 August — a second seat was added without my knowledge or consent and I was charged £28.56. Support removed the extra seat and confirmed a refund, but I never received it.
• September — billed correctly £30.
• 9 October — again billed £60 because another seat was added without my consent. After many exchanges I was told I was back on the promotional plan and would not be charged the following month.
• November — billed £60 again.
• 27 November — I submitted a support request. I received no reply.
• 28 November — my account was suddenly deactivated without my knowledge or consent and no deactivation email was sent.
• 28 November — I repurchased a business subscription only to regain access so I could follow up. And have received no reply to my many many support requests both via tickets and via email.

[charred parcel]

how did you contact support?

[cerulean stirrup]

Via ticket and via email to [e-mail] and [e-mail].

[charred parcel]

was the email [e-mail]?

[cerulean stirrup]

yes

[charred parcel]

when was the last email sent to [e-mail]?

[cerulean stirrup]

28th Nov- I rewrote everything out again, as I hadn't received a response to any of my other emails or tickets. And then I followed up on the 2nd Dec saying I hadn't received a response to the email on the 28th and still nothing

[charred parcel]

ok, that chronology is really good. If I were you I would contact my bank and have them issue a chargeback

[cerulean stirrup]

I did that but the issue is that i'm still set at 2 seats with an ongoing charge of over double what I agreed to and I can't keep doing this every month

[charred parcel]

is there a reason why you went for business over Plus?

[cerulean stirrup]

At £30 it was only £10 more expensive then plus and had more features. I could move my data/history from plus to business but it turns out you can't move your data back from business to plus. But with 2 seats its £60 per month. So i'm either stuck paying £60 per month for something I never agreed to or I loose all my history and data. And i have tried exporting my history and data so I could manually input it to a new account but becuase I signed up with an apple id there is also issues with that and the [e-mail] aren't getting back to me as to how to manually export it. So i'm stuck

[charred parcel]

this is unacceptable from OpenAI, however support is woeful

[cerulean stirrup]

I'm at a loss as to what to do

[charred parcel]

I think the only thing you can do is keep logging tickets through https://help.openai.com

[cerulean stirrup]

I've been doing that. the issue is they repsond to you once. You tell them what the problem is and then never hear from them again

[charred parcel]

it's not good enough, but I don't see what more you can do

[cerulean stirrup]

Its really appalling

[charred parcel]

I recently unsubscribed from ChatGPT and moved to Google, maybe it's worth reconsidering where you end up?

[cerulean stirrup]

I need to get my data off it. Which is also causing an issue. Everywhere you turn its feels like its not working as it should

[charred parcel]

I agree, OpenAI seem to be concentrating on stupid things like meme video generators rather than serving their core userbase well

[cerulean stirrup]

It's not great. Making sure you're not ignoring your customers seems like a bare minimum. Anyway thank you for your help. Happy Sunday

[lapis marten]

Hey, is there anyway I can make deep research cite the documents used as reference in the text?

[drifting hinge]

I have Business

[fading osprey]

my gpt model said im using 4 when im on 5.1nano

i think gpt

is brokne

[drifting hinge]

Kind of unfair, I'm a paying client and I get their old crap? I miss the logic.

[drifting hinge]

Or is it hallucinating

[fading osprey]

no it’s actually 5.1 or whatever it is but it just says 4

(I think at least)

but yea

[wintry parcel]

I tried, and results were inconsistent at best. I asked Deep Research to "cite sources in the APA format and provide a list of sources cited at the end along with the article links." The report did include article links throughout and there was a list of sources at the end. But only a couple of sources actually had APA in-text citation, and the list of sources didn't cover everything. I think it has a hard time distinguishing between the actual article vs. the website where it came from. Also, if you are doing academic research, you will often need to specify page #s, etc., and AI models can never manage that in my experience.

[lapis marten]

I see. Yeah, I have similar experience to yours. Right now I actually don't need the pages, just for the AI to refer to the source and/or display which document it took the info from (I attached like 30 documents for it to base the writing on lol)

[wintry parcel]

Oh, so, you actually gave AI documents to refer to? Then, it should be able to refer to the specific doc you uploaded. Have you tried giving specific instruction to "provide reference to the specific document" or something like that?

[lapis marten]

Great, I just gave something like "refer to this to justify the relevance of the topic" or "refer to this when talking about x phenomenon"

But for the most part I just fed it the sources straight up

[wintry parcel]

Deep Research would often take it upon itself to go out and look for sources on the web. In some reports I did with Deep Research, it completely ignored the docs I uploaded and went searching for my website to obtain the same information, lol. So, maybe, if you want it to refer to the attached sources in their reports, you might need to be specific about that, like "only use the attached documents as knowledge base. Provide references to the specific document you refered to throughout the report." Hope you get it to work!

[pearl lodge]

Hey, I'm trying to upload an image for editing, but chatgpt keeps telling me over and over that it can't see the image, saying it's an inline embed, or that it doesn't have an asset id. Weird thing is that it's an image generated in the same chat; its right there in the library.

[sudden delta]

How do you fix this? It's only happening on the mobile app. The desktop is working fine and web browser is still working fine.

[dusty brook]

Is does chat got plus give sora ai users unlimited generators or is that juts sora pro??

[lapis marten]

i wrote mostly because im open to suggestions

should i write "only"?

also, i don't want it to mess up timeline (jumble up historical events), i wrote a prompt to prevent it from doing that but i'm worried lol

[wintry parcel]

It depends on what you want/need.

[onyx kraken]

I would try restarting the device, and if that didn't fix it, I'd consider trying a different network if possible (like, wifi vs carrier), it could be related to a local issue and that could help you narrow it down.

If none of that changes anything, I'd reach out to OpenAI tech support, https://help.openai.com/en/articles/6614161-how-can-i-contact-support

[sudden delta]

In fact, my device actually restarted very recently because the battery died and required a full power cycle.

Also I'm using wifi

It's still happening on carrier.

[lapis marten]

wrote "overwhelmingly my sources" or something of sorts

whatever overwhelmingly means to it

[exotic ravine]

@onyx kraken do you agree with this gentlemans suggestion?

[onyx kraken]

Sounds reasonable to me, so yes!

When a chat 'goes strange' or otherwise isn't the way you want it to be, often odd patterns will continue inside that chat - even if you fix one thing, it's likely there's other weird things.

A new chat often is 'within normal range', only a few are strange.

I kinda think of starting a new chat, like throwing a net into the ocean.... somewhere. Guided by your prompt and the rest of the first input, and whatever personalization - to tell it roughly where in the ocean to cast the model's net of what to pull out of the training data -

There's some really weird stuff there.

And if it grabs weird training data as its core, and starts answering with that focus - often the easiest fix is just "Start a new chat"

[exotic ravine]

groan.

same problem yesterday and today

[exotic ravine]

okay so i tried the suggestion of creating a new chat and everythings working perfectly again go figure

[exotic ravine]

this begs the next question

how many queries do i get before its appropriate for me to open up a new gpt chat

[onyx kraken]

I think that may be affected by total tokens involved - it's not necessarily 'how many messages' - so much as 'how much total info has to be given to the model each new input' - because inside context window, absolutely everything in the chat is given to it.

As that gets longer, past 'some point' I notice slowdowns, and that's usually where I stop if I haven't already stopped.

I find that the model remembers (between chats!) better if each chat is fairly short, and it remembers the early part of the chat only if it's 'too long' (umm, that's measured by, 'okay, the new chat has no idea what we said later in the chat, that's too late if I want cross-chat memory working well).

So, that's totally up to you, how many queries (inputs, right?) you want to use! They're your chats, and if you want to have super long ones that go until they bug out - you can, or that go until you get told 'that's how long this chat can become' (I have seen people mention seeing that - I've never seen it myself).

[exotic ravine]

what about the option: "branch into new chat" - ?

[onyx kraken]

That's fairly new and not a lot of helpfiles about it yet!

What I'm aware of: If the original chat bugs out or is deleted - the branches have also been lost at the same time.

What I'm exploring: Prepping a core chat to branch from, and then making new branches to explore various directions with, going back to the otherwise-untouched original and branching again anew when I'm ready to. So far that's working in predictable and wanted ways.

[sudden delta]

So I really have to contact support for this. It's really annoying. I wish I could just fix it. Why does it say request is not allowed?

[vague coral]

Hello,
I'm currently working on a small application that generates images, but I'm getting this error message:

"OpenAI API error: Your organization must be verified to use the model gpt-image-1. Please go to: https://platform.openai.com/settings/organization/general and click on Verify Organization. If you just verified, it can take up to 15 minutes for access to propagate."

However, the verification went smoothly, and it's been over 2 hours.

If you could get some help, thank you in advance.

[charred parcel]

it can take up to a day for verification to take effect in practice

[vague coral]

Okay, thank you for your answer. It's not a problem if I use a Google account on OpenAI?

[charred parcel]

do you mean for SSO or just using a Gmail account ?

[vague coral]

I simply use my Google account to log in to OpenAI

[charred parcel]

that's fine

[vague coral]

Okay, thank you, I'll wait a little longer.

[onyx kraken]

I don't know, because I don't see that listed in the various OpenAI help files.

I'm a volunteer community member, like yourself! Only the very few folks who have gold names work for OpenAI, #server-staff . But we can help each other inside what is known, and reaching out to OpenAI tech support is what I'd do next in your situation.

[bitter pecan]

I've been running into an issue where I'll prompt for an image or sketch for concept art. In the past I never had problems, now it ask me a bunch of clarifying questions and ends with "Working on generation now..." but then doesn't actually generate anything. I even got it to say "Sorry I wasn't actually generating anything for you." last night.

Any idea what might be up with it or was there some changes to its image generation rules?

[onyx kraken]

Are you using only 1 chat to ask for each of these in? Any 1 chat can 'go weird', if that's your situation I recommend explore in a new chat, and maybe use new chats fairly frequently.

Otherwise, I'd check memories, custom instructions, and my prompt for potentially conflicting or impossible instructions, because that pattern of response is commonly displayed by the models when we ask them for some task that they interpret (or really is) too big for them to do in one step, and appear not to welcome iteration towards what can be done.

I might also input something like,

Hey, describe back to me what I've asked you to do. Does the request conflict anywhere with your programming or other requests I've made? Are there areas where you need to guess what I want?

That can help the model potentially explain what it thinks you want, and why that may not seem possible to do. The models usually will not volunteer 'Oh, that's not possible, we'd have to change stuff or break it into steps', probably because that kind of answer may not score high on human preference testing, so the models may avoid that even when it might be useful and acceptable - and instead may slip into roleplay that it's being done 'in pretend' because that may be a closer match for preference training.

[sudden delta]

Alright..

[sudden compass]

Why does closeai charge such a high price for the Malaysian region? The Pro subscription is 43 USD more expensive than the original price of 200 USD.🙃

[little hound]

Do the Powers That Be ever retroactively go into a Sora 2 account and delete already-generated videos?

[trail cairn]

USD prices never include vat or anything else

[onyx kraken]

Sort of, yes, depending on the contents of the video.

Per https://help.openai.com/en/articles/12435986-generating-content-with-characters:

So, based on that, if you made a video with someone else's character in it, and that person requests it be deleted, that would happen.

Otherwise, it's certainly possible that a reported video might be deleted, though I'd expect you to be informed or warned if there were any rule concerns.

[little hound]

And yet...here we are.

[exotic ravine]

it also occoured to me, after looking into this for a bit, that it could be the jump from version 5 to version 5.1 (some of my conversations i was having trouble with were still in version 5 which i think was causing my issues)

[shut solar]

Wait is 5.2 happening?

[raven ore]

@onyx kraken Hi! Some of my submissions to Sora Challenge didn't get the star from Server Utility. Should I resubmit them? Thanks

[shut solar]

Is it dropping tommoroe?

[onyx kraken]

Hi!

I don't know much at all about the Sora Challenge, however I note that https://discordapp.com/channels/974519864045756446/1314861950211063849/1445839255036362802 mentions that the Server Staff Team will be selecting the winner.

I expect as long as your entry(ries) are in the channel within the time period for the challenge, they can be judged, but I'm not a part of this and do not know.

[raven ore]

Ok. I just don't want them not being judged since I work hard trying to get that Pro :}. Thanks

[onyx kraken]

I care a lot, I just don't know every answer, and that particular question, I'm not sure who to ask either.

Good luck!

[exotic ravine]

soon

[raven ore]

Appreciate that you care 🤗. If you know let me know.

[stuck quarry]

Hey everyone,

I am developing a voice agent and am just coming across issues with usage tracking.

My usage dashboard shows that I have spent $3 yet doesnt go into any detail as to how I spent that money. I only have one organisation and one project.

ive been trying to fix this for hours so if anyone could help that would be very much appreciated

[void zenith]

Can't login into my sora it asks for the dob as says you are already logged in and stuck on that pade any help

[onyx kraken]

I don't use the API myself, but does this page help with links how to see the costs endpoint in more detail? https://platform.openai.com/docs/api-reference/usage

[stuck quarry]

I tried that, unfortunately the API still returns with 0 tokens

[simple bluff]

I am getting error while sending messages to gpt anybody can help?
Chatgpt is loading successfully but its not giving any output. Saying request not allowed please try again later. Every history is loading correct but gpt is not giving answer in plus mode I am using gpt plus

[plain tartan]

How can I upload video on chatgpt?

[plain tartan]

Sign out and sign in

[green summit]

Sign out, and back in. That will fix it.

I just realized someone else mentioned this as well lol

[rugged wing]

I have a new discord account and i would like to link the new discord account to my openai account.

[green summit]

I asked GPT, here's what it gave me:

How to link a new Discord account to an OpenAI account

They can’t just “switch Discord accounts” inside OpenAI. OpenAI links whatever Discord account was used during the original login or connection, so to change it they need to disconnect the old one and then reconnect using the new account.

Here’s the clean step by step:

Step 1: Log into your OpenAI account

Tell them to go to
https://chat.openai.com
and log in normally.

Step 2: Go to Settings

Bottom left, click their Profile icon, then Settings.

Step 3: Go to “Connected apps”

Inside Settings, there is a tab called Connected apps or Connections.

Step 4: Disconnect the old Discord account

They’ll see their old Discord account listed.
Press Disconnect.

This removes the link, but doesn’t remove the OpenAI account itself. Just the connection.

Step 5: Reconnect using the new Discord account

Now they must:

1. Log out of Discord in the browser

2. Log into Discord with the new account

3. Go back to OpenAI → Settings → Connected apps

4. Press Connect next to Discord

Discord will open the OAuth popup and ask permission.
Because they’re logged in as the new Discord account, that’s the one that gets linked.

---

Important note

If they originally signed up for OpenAI through Discord login, the process might require them to:

Set an email and password first (in case their OpenAI account was Discord only)

Then disconnect Discord

Then reconnect with the new Discord

Super easy, just one extra step.

[broken zinc]

How can i do this guys?

[livid kayak]

Hi, can anybody assist in getting warnings about unauthorized uses for a never used ChatGpt account created long ago?

[rustic marlin]

run /unlink, then log on the new account and do the linking process

[rustic marlin]

this is not correct

also, plz, don't send AI generated text walls, use #1050184247920562316 for it =)

[rustic marlin]

change the password of the account

[livid kayak]

I have changed the password and activated otp signin - also i signed out from every device, there are still some (?) api keys active, showing last usage today, BUT i can not see any usage in the logs

[rustic marlin]

remove those apikeys then

I assumed it was about ChatGPT, not about the API

[livid kayak]

i am currently in process of removing ALL

[sudden compass]

does gpt5.2 has published?

[drowsy tundra]

Anyone here know about 2 cameos prob?

[thorn summit]

Is it Sora2?

[drowsy tundra]

yea its sora 2

[jolly plover]

Can someone help me

[simple bluff]

It's not working at alll

[green summit]

Hm, it should work tho. I've had that same problem a few days ago, and when others had the same issue, I also told them to sign out/back in and it worked for them as well.

[onyx kraken]

Perhaps reach out to OpenAI tech support, https://help.openai.com/en/articles/6614161-how-can-i-contact-support and ask for more info from the professionals, if no community member here (We're nearly all community members here, only the gold-named folks work for OpenAI) spots your question and has a working answer they can share.

[simple bluff]

I tried using multiple devices everywhere same

[spare yew]

should a pro user get this prompt

if anyone knows ofcourse it would be much appriciated.

[sudden delta]

I have the same issue and that didn't work for me.

[fossil shoal]

Hello everyone,
I’m French, can you help me to access at Sora 2 ?
I have GPT pro

[onyx kraken]

Sora 2 has not yet been released in France, and you can watch here to see as expansion continues: https://help.openai.com/en/articles/12461230-sora-app-and-sora-2-supported-countries

I strongly recommend waiting until it is released in your region, because it would violate ToS otherwise and when OpenAI noticed, they likely would choose not to do business with you again.

[raven ore]

@rustic marlin Hi! When is the submission deadline for the Sora Weekly Challenge? If the new challenge goes live every Wednesday, that means submissions are closed on Tuesday? Oh also done if my entries didn't get the star from Server Utility #3447, they failed? Thanks.

[rustic marlin]

you can send your entries up to the moment the next following theme is published

if you don't get the start, it is because te bot is acting funky, good thing you pointed that, I'll get that checked

[vernal wyvern]

Well that's my complaints and screenshot sent to the ombudsman in uk. I did report 6 month ago chatgpt not let me cancel subscription. I also said I will try every month and document every attempt.

I did just that 😁

I got screenshots of the website saying failed to cancel and nothing in my android subscriptions.

Before anybody wondered why I'm writing it here. Thos is the official discord for openai. It is the only place to contact them live. It is also evidence I have tried multiple ways to reach out to them.

[charred parcel]

I was able to cancel fine the other day

[vernal wyvern]

I have all the screenshot and evidence I need

[stuck ridge]

guys I have this problem, I am using chatgpt plus

[vernal wyvern]

6 months worth. Even video clips of me going to the website logging in then trying.

Then reaching out to chatgpt and being ignored

I should have waited another year but 6 months is long enough for a response.

I also reported it as theft. Just need to get a crime number now

[reef lion]

Hello. I'm trying to find where I can download my Multi-factor authentication recovery file/backup and can't find it anywhere under security? Am I missing something or?

[undone skiff]

Hey folks 👋 is there any open-source framework for building OpenAI-powered apps that’s more of an all-in-one starter kit?

I'm aware of the SDKs, but wiring everything together gets messy quickly. I’m looking for something Laravel-like for OpenAI: Opinionated structure - Built-in UI support - Easy setup...

If anyone’s knows something like that please let me know 🙏

[lethal minnow]

As a photographer I have a photo of a model holding a candle lantern. I would like to create an AI atmosphere of her walking in the dark woods. Is there an AI program that doesn't have guardrails to allow for partial clothed images? Nothing sleazy. Thank you in advance.

[hollow lark]

My ChatGPT is slow typing

And I can’t do anything with the app

[orchid holly]

my chat keeps showing a grey box over it when I click out of it, anyone else have this issue?

[solid ivy]

the moment i login chatgpt is useless. javascript eat crap the webpage is unusable

the phone locks up too. i cant even scroll without the app chugging and lagging.

if i try to use a mac. the web page literally never loads.....

no one from support is worth a crap. they are wayyyy useless! 100 emails and nothing has been accomplished.

[solid ivy]

thats just the deal after you ask a few questions. it uses javascrpt and JS sucks. it just locks up. chatgpt web page and apps are both garbage

[hollow lark]

it sucks even if i try computer or mobile it just still slows

[solid ivy]

It doesnt matter what you do. Their website is fing garbage

These idiots busy using JavaScript

[solid ivy]

what a joke

[prisma sinew]

Unable to generate
[{'type': 'literal_error', 'loc': ('path', 'task_kind'), 'msg': "Input should be 'winter'", 'ctx': {'expected': "'winter'"}}]

can someoe help me please?

[deft flare]

i have been trying to retrieve my openai data for about a month now, always either downloads an incomplete file or nothing gets downloaded at all, the privacy portal gui always gets either stuck between eleven to seventeen per cent and either produces a partial file or nothing or it flips into "download" complete all of a sudden, but again, either no file at all, or a partial file, support is abysmal, weren't really able to help me anyhow, as a network engineer of 14 years of experience i have proven my connectivity above and beyond (i NEVER had any issues downloading anything anyhow, anyways, but did go the lengths to check the network anyways), i tried two different browsers (brave and firefox), has anyone had any similar issues by any chance?

[onyx kraken]

For me, I request a new download, then within the day once it arrives (email) I download it then. I don't think the link's good for an especially long time, and the help file says it's only good for 24 hours.

https://help.openai.com/en/articles/7260999-how-do-i-export-my-chatgpt-history-and-data

Are you having trouble inside the 24 hour window?

If everything seems to be happening correctly except the actual link is failing, I'd reach out to https://help.openai.com/en/articles/6614161-how-can-i-contact-support and inform them and ask for help.

[dusty locust]

Has anyone gotten this error message in Sora 2 storyboard? [{'type': 'literal_error', 'loc': ('path', 'task_kind'), 'msg': "Input should be 'winter'", 'ctx': {'expected': "'winter'"}}]

[hushed crane]

How do I get Chat GPT to help with any subject above a G rating? I write some pretty mature fiction with a lot of adult subject matter. But the app keeps assuming I'm asking for advice I don't want* and redirects me to a hotline.

*I can't say the type of advice, because the word is flagged and my posts get deleted JUST FOR MENTIONING THAT THIS THING EXISTS.

[uncut saddle]

exists a way to remove the rejected request of images on SORA? I tried to delete it's a lot of times, but after deleting those black frames with the message are re-appearing once and another time again, has been impossible to erase it's from my images

[autumn nexus]

Why cant i see announcements?

[valid charm]

I have over 4K unused credits cos it’s been charging me each month without me being able to talk to anyone about refunds. Finally canceling it but jeez how do I contact anyone? That’s why I’m on this server in the first place

[abstract minnow]

I have a question, my chatgpt plus account is cracked, I already requested help from support and gave them what they needed, It's been 15 days since I've been able to use the account because I can't send messages or delete the corrupted conversation. What can I do?

[charred parcel]

the only thing you can do is keep engaging with support, nobody here can help with this

[abstract minnow]

Yes, I do, but what I'm wondering is whether sending a lot of emails to support might be more counterproductive?

[charred parcel]

the only people who can help you are support, so if you're having issues then that is who you engage with

[spice merlin]

anyone else having weird top_p errors with response api?

I'm getting top_p errors with 5.2/5.2-Pro as well, even though those models don't even USE top_p

i feel like it has to be something stupid that I've overlooked

[unkempt pendant]

Trying to test out GPT 5.2 for my project. Testing in Playground and in API with my project, get a 400 error every time saying Top_P is not supported. But Playground doesnt have a Top P value and my project isn't calling citing or specifying Top_P anywhere. Can't seem to fix

If I make a new prompt it works, but the second I save the prompt in playground, boom, broken and talking about Top_P

[frank ferry]

Hi everyone,
I’m exploring a few roles right now— AI Engineer, Data Engineer, ML roles, Python Developer, Data Scientist, Data Analyst, Software Developer, and Business Analyst—and I’m hoping to learn from people who’ve actually worked in these fields.

To anyone with hands-on experience:

What unexpected realities did you run into when you started working in the real world?

How would you describe the drawbacks or limitations of your role that newcomers rarely see upfront?

What skills or habits turned out to be essential for excelling, beyond the usual technical checklists?

How did your day-to-day responsibilities shift compared to what you imagined before entering the field?

In what way would you recommend a beginner test whether they’re genuinely suited for your role?

I’m trying to build a realistic picture before choosing a direction, and firsthand insight would really help.
Thanks in advance for sharing your experience.

[mellow harbor]

I am a solopreneur and my company is currently on Plus. I’m thinking of upgrading to Business because it assures no training on my company data. However Business plan does not support memory with past chats (yet). Will there be any issues with my old chats if I upgrade?

[north lodge]

Anyone know why chatgpt isnt working on my iphone 11?

[vivid mortar]

The website or the app?

[vivid mortar]

Nope! Should be no loss of memory or chats

[vivid mortar]

If that is something you are concerned about, I would suggest saving some of your chats maybe just to large text files just incase :)

I personally havent gone through that upgrade path so I cannot say for CERTAIN but you should not lose any history

[broken hazel]

Hello ! I just subscribed to the Plus plan that allows to create videos. But when I try to create a video, I got a message telling me "Unable to generate video
Upgrade your plan to generate more videos" !

[charred parcel]

what country are you in?

[broken hazel]

France.

[charred parcel]

Sora 2 is not available in France, you will need to wait until it's rolled out there

[broken hazel]

I see. I'll ask a refund then.

Thank you for your answer 🙂

[wispy scaffold]

i can help

[charred parcel]

#server-rules rule 7 says you cannot ask this

[wide fossil]

actually i am not promoting anything it's about educational i am just seeking help for my research thing

[charred parcel]

you are soliciting, which is disallowed

[wide fossil]

Oh, but this is the help section chat, so I can’t ask for help?

[queen temple]

Can anyone point me in the direction to make a refund?

[charred parcel]

https://help.openai.com

[queen temple]

Thank you Robert!

[worthy heron]

The MacOS chat bar has been broken for months. The cursor has stopped automatically appearing in the typing box, and the "remember last position" feature is accurate within the same side of the screen. It's ridiculous

[charred parcel]

cursor appears for me

[sterile anchor]

What is the contexte windows of chatgpt 5.2

[unkempt pendant]

still no luck

[charred parcel]

yes

[burnt pumice]

Hi. I was blocked by the openai and chatgpt platforms. What can be done? I've written to support and I haven't been able to get a response from them for a month. And I have my agents and money there.

[charred parcel]

did you get an email from OpenAI telling you that you were banned?

[shut solar]

How do I use gpt 5.2

[onyx kraken]

It's rolling out to all members. I hope the rollout reaches you soon!

Once it does, refreshing your page should be enough to let you see you have 5.2

[shut solar]

Has it rolled out to anyone yet?

[onyx kraken]

Yes, several report having it already

[vivid mortar]

400,000 Tokens https://platform.openai.com/docs/models/gpt-5.2

[median lotus]

How come my voice version of chatgpt is talking dutch to me in english. Its very annoying cause it doesnt like it if i speak dutch or english to it as both languages confuse it to whag im saying.

Where do you change the voice language its speaking in

[foggy pike]

Bro why chatgpt browser using 15 gigs of ram?

[vestal bane]

Hello, I have a business account, but I can’t generate videos on Sora 1 (the old website). The interface seems to have locked the video options (such as duration, aspect ratio, etc.), and when I press Enter to create a video I get an error message saying “Upgrade your plan to generate more videos.” I’m not sure why it says “more,” since I can’t generate even a single video.

[indigo cape]

Can I delete my chatgpt account without login? I was lost access to my mail that i registed to ChatGPT

[sterile anchor]

With API no in the app

[turbid agate]

I can't even afford 1 month free trial 😭

[steep cobalt]

i need a code to get in help

[worthy heron]

It (sometimes) appears when I first open the chat bar, but after inputting one query, the cursor goes away, and I have to click in the typing area to get the cursor back. You don't have this issue?? 🫠

[calm sky]

Help sora has changed i cant download any sora ai vids without downloading the sora ai app but the app isn't available in my country.

[stone fulcrum]

My OpenAI api key dosen`t work

[trail cairn]

what does it say?

[trail cairn]

and whats in e as in the raw api error message? also did you buy credits

[stone fulcrum]

No

[trail cairn]

api is not free

[calm sky]

Help sora has changed i cant download any sora ai vids without downloading the sora ai app but the app isn't available in my country.

[static saddle]

Has anyone explored with making ai stories? I been playing around with making a anime with ai and my method has been making images with chatgpt and then animating the images into videos with aivideomaker.. music with suno and voices with eleven labs.

My stump is figuring out how to lip sync the dilague with the characters. Has anyone played around with lipsync in any projects?

[calm sky]

Heeeeeeelp

sora has changed i cant download any sora ai vids without downloading the sora ai app but the app isn't available in my country.

[little hound]

Weird. I can download from the site just fine.

[calm sky]

Hmm maybe it's a glitch

I thought sora ai changed its rules

[serene bramble]

My create character option disappeared. Is this a glitch or did they remove the feature?

[dreamy cedar]

anyone else having problem downloading ? only says "Download Failed"

[mellow harbor]

With the Business Plan, is it possible to just have 1 person subscribing?

[rugged wing]

where is the ban appeal server or how can i contact modmail while not being on the server

[magic delta]

Has anyone been having issues with voice chat? It started acting up last night saying it couldn’t connect and didn’t have a connection. But my WiFi has been working perfectly fine

[drowsy locust]

Why isn't voice working

[royal blade]

Any idea why ChatGPT ain't answering and just saying "Stopped thinking" or "Thought for x minutes"?

[vague drum]

Me too. I reported it on the help chat. Been out since yesterday over 24 hours now

[stone yew]

pls help im stuck this screen

[little hound]

Maybe because there's no 17th month?

[onyx kraken]

What happens when you click 'next'? And it appears to be telling you to log in, because your account is already created - are you able to try and log in? (Don't share your personal info - the people here are just people, keep your info private!)

[warm fjord]

I paid for ChatGPT Pro to get Sora 2 Pro and the videos only show up with a watermark LMAO, need help

[onyx kraken]

https://help.openai.com/en/articles/12460853-creating-videos-with-sora

[warm fjord]

ohh, i working with image input of building

xD

time to work with open source watermover

[real vessel]

Is it just me experiencing an issue where you ask a question, and the chat responds with a question you asked earlier in the session, as it occasionally happens during my longer chats?

[white ingot]

It's seems to be a Bug in 5.2

[maiden scarab]

Anyone have any web browser suggestions to not get bogged down with lag after long conversations? The mobile app seems impervious to lag, which is great, but browsers just get hit with ram overloads. 😫

If i recall, the windows app also gets slow. Its just mobile app that goes on foreva.

[kindred oasis]

Anyone else finding GPT 5.2 buggy as f?

Dang, bro, it just passes right at what I've just said, reply what it already've answered and just don't follow simple instructions of wanted behavior and ignores them

What the heck, OpenAI?

[kindred oasis]

Like, literally: it's just like the beginning of GPT 5: it's robotic, zero conversational and with that "overwhelmed secretary" energy again

That's boring

[white ingot]

that's why a Lot of PPL going Back to 5.1 until this is fixed.

[kindred oasis]

I'm unfortunately stuck with the free version
Even out of the daily limit it stills replies to me with 5.2 instead of 5.1
Dang...

[worn ledge]

Hello guys, I was wondering if anyone face the same issue as I do:
I have lots of projects in ChatGPT (arround 50) which includes many chats inside (from 1 to 15). When I'm using ChatGPT online on the live website, it takes so long to load my projects (from 5 to 60 secs to load 10 projects) which makes it kinda annoying to me because I need to wait 2/3 minutes to find very old chat sometimes. At the same time, the search function doesn't allow to search for a specific project at this time.
Does OpenAI is aware of this and planning to fix this? (fyi: I'm ChatGPT Plus Subscriber)

[buoyant oyster]

Oh. Had that a couple years back. Yes, OpenAI is aware of issues related to projects loading and handling large numbers of chats, because this is a known issue within their dev community.

[drowsy locust]

🤦‍♀️ American thinking their way is the only way

Day/Month/Year or Year/Month/Day is way more popularity accepted around the world the the weird USA centric Month/Day/Year

[zenith zinc]

NEED HELP:
I created an APP SDK which was working fine few weeks back but today when I tested it was not working. Chatgpt is connecting to my MCP server perfectly as my terminal logs show. My mcp server is also working fine on playground and on MCP inspector.
But when I choose my tool from connectors and then give a prompt related to the availble tools, it says looking for available tools but then doesn’t call the tool.

And when I explicitly mention the tool name to be used, it says no such tool found in the environment.

I also tried with openAi’s pizza server also, it was not calling that also

Anyone else facing this issue ? Chatgpt not calling local APP SSK

[cerulean stirrup]

HELP: Support and Privacy are both not resposnding to me. They have changed my plan multiple times, without consulting me, overcharged me, said they will refund me and haven't, then keep ignoring me when I contact support.