#chatgpt-discussions
1 messages ยท Page 61 of 1
its not been almost an hour lol
since the last time i used it
Reactions are a booster perk 
meh
how these people lived before chatgpt? ๐
the bot is down ?
I honestly have 0 clue tbh
why down
i got the max booster badge mwahahahaha
problems with chat gpt
yeah vram is a big deal at the moment
for me everything is working flawless, including gpt 4
network errors left and right
thats because you're special < 3
Download NVIDIA Chat with RTX
Simply download, install, and start chatting right away.
9.0 MB/s - 352 MB of 35.1 GB, 1 hour left
I'm sorry, come again???
When will this end
same here
same for everyone ๐ they are working on it
SIMPLY download, install, and start chatting RIGHT AWAY you say, alright.
Discussions of non-OpenAI products / models should only be posted and discussed in the #ai-discussions channel.
aw :/
When I need GPT and it's down....happy that I have that 2 month Gemini trial lol
https://status.openai.com for details on the current partial outage
Finally managed to log in after logging out, but chat still gives an error when generating replies
My cats breath smells like cat food...๐
but its free?
yeah, i just got "network error" with no capitalization
strange they wouldn't capitalize it but yeah
usually fixed pretty quickly
its down
must be something wrong with it, i'd take it to the vet
Advanced version is not
I cant even find it
Just had a new error message, "Error in Moderation"
Interesting, i guess it's related to the current issue
you got it ๐
seems to be starting to work
error in moderation, network error
two main ones
โ
it's because you need to use with more moderately
ai down ?
you're in the wrong place lol
Hi Super Nintendo Chalmers
down
"You've reached the current usage cap for GPT-4. You can continue with the default model now, or try again later. Learn more"
Welp, guess my usage is wasted trying to get it to answer
soooo whats going on with ai
when fix ?
how can you be so sure
thats the same thing as asking "are we there yet" in the middle of a 5 hour roadtrip
let the devs cook
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.
how can she slap?
It was clearly a joke question, just saying, made to humor exactly the people who come in here and before even reading one message up in chat, ask stuff like that
lets keep it on topic everyone
lmao ok
Dammit, I was in the middle of a discussion to find a perfect term to describe a feature. is it a way of the AI to tell me that i'm too picky ? ๐
You kinda have to
you're right ๐ we're just chillin while waiting for it to come back up - im fantasizing abouut all the catfood ill be asking chatgpt about when it comes up
what is an easy prompt to test if you are using gpt 3.5 or gpt4?
like a prompt that gpt 3.5 fails but gpt 4 passes
Can you not ping every booster
You look at the pfp of the Chatbot, if it's green, it's 3.5, if it's pinkish, it's GPT-4
you can try the today I have 3 apples, yesturday I ate 2 , how many apples do I have now?
i think its supposed to know which gpt it is, but its not completely reliable and sometimes hallucinates wrong version
3.5 will usually say 1 and 4 will say 3 correctly
I like Foxabilo's answer to this - alternatively there is the prompt from the GPT-4 showcase page:
Andrew is free from 11 am to 3 pm, Joanne is free from noon to 2 pm and then 3:30 pm to 5 pm. Hannah is available at noon for half an hour, and then 4 pm to 6 pm. What are some options for start times for a 30 minute meeting for Andrew, Hannah, and Joanne?
nice one too
is ChatGPT getting DDoS?
Can't you also just ask what AI language model it's based on, and it'll tell you? ๐ฟ It's always worked for me
excellent, i will try this! thank you @dense halo and @sleek aspen , the reason for this its because Im using the api for a bot app and wanted to confirm it is indeed using gpt 4 instead of 3.5
does nopt look like it, looks liek a common subsystem is down that affects many systems
While you are correct to some extent, if you were to use the API like this user is, your prompt wouldn't work
can't they just rollback if so?
i have GPT-3.5 selected but it says I am using GPT-4 and that it prompts me to go to GPT-3.5
bug or
is 3.5 working rn atleast?
i believe 4 was hallucinating 4.5 for a while by mistake so yeah, not reliable
!
You've reached the current usage cap for GPT-4. You can continue with the default model now, or try again later. Learn more
That may be an option, but if, for example, it's a 3rd party service that is down then rolling back does nothing
This error popped up on my page
When?
๐
yeah but it looks like it's the 0auth that's down and that's a common attack vector
no its literally thinking i have used my GPT-4 usage and won't let me continue
i think around 2 months ago, i doubt i can post links but you can google "chatgpt 4 halluucinating its 4.5"
100% could be, it just does not strike me as an attack becaus eyou can still access the systems, they are just not working correctly
ChatGPT down?
i managed to get back into this server so its not completely broken
I'm using chat gpt 3.5 and it's saying I'm using chat gpt 4.5
partial outage
https://status.openai.com/ for details on the current partial outage
๐คฃ
thought i wasnt the only one
Something's really wrong
Oh right, iirc I tried getting it to say that for me as well, but I don't think it ever did, still said GPT-4 and some very basic stuff
did someone let all of us use gpt4 by mistake and this is why the servers are crumbling?
its even more funny when gemini or the ai X is using is hallucinating that its chatgpt ๐
That is actually tragic and hysterical as hell
https://status.openai.com/ for details on the current partial outage
why is ChatGPT alway down when I need to learn some new framework I must be cursed
should be sorted soon... hopefully
I have a chat with a title literally nothing.. like the title isn't there
when i was messing around 10 min earlier
it tries to make a new title now as the conversation changes
well I don't know about that the last outage was more than 5 hours
"Too many requests in 1 hour. Try again later."
I set the title to what I intend and it doesn't change it
hmm
hm yeah its kinda up, it spent more than a minute "thinking" and then suddenly spit out a bunch of text for me, its probably overloaded by all the requests comin in after being down, or not properly up yet
I think the oauth is still down
still rate limited for gpt 4.0 even though litteraly all i got was errors ๐ gotta love it
did you keep making requests while it was down?
yes, it would be a cool gesture to reset that count after a fix like this
^
The OpenAI Discord is an actively moderated server.
โข Refrain from sharing inappropriate content on the server. This includes but is not limited to messages, media, or other topics of graphically violent, sexual nature, and drug-related content.
โข Report all sensitive and offensive content in the feedback reporting tool in the ChatGPT web UI instead of here on Discord.
I'm whining about ChatGPT being down, but the companies that use the API on their prod must be sweating
you dont pay for a guaranteed 0% downtime, no service can promise that
im sure they are doing what they can
world class AI for few bucks
not for me, but i wasn't near the cap already
3.5 is also working for me
40 message per 3 hours, but i think they raise the limit when they have extra capacity
yeah I think it's only the oauth that's down if you are logged it's fine
๐คทโโ๏ธ
yo what the
kinda bug, am not using GPT-4 and it says i reached my max on it
kinda fix duh
no news on https://twitter.com/sama
it works again for me
same
oauth still not working
Oops!
We ran into an issue while signing you in, please take a break and try again soon.
best bet is to clear cookies
could of just got jumbled up
seems to be working for most now
yeah sounds like a cache problem, maybe try ctrl+f5
it's probably just by region
or try logging in while standing on one leg, worked for me
yeah or clean browser cache
It's funny because most answers here all the time are exactly the same.
I was referring to this.
๐
But something is still wrong.
I loaded up a document and asked: can you open it and can you read it?
Answer: yes.
What is the last sentence of the document?
Answer: rubbish, more rubbish, much more rubbish.
No, that is not the last sentence, that is even not in the document, heck the whole document is about a complete other topic
Answer: The document has many sentences.

it could be that some parts of backend is still down, try a new conversation with the document and see if it helps
Ah, yes, GPT --3.5
How big is the document?
about 29cm x 15cm
It was just a test document. Less then one page. Just to check if it works.
sorry ๐ค
also that's not how knowledge retrieval works, it is an embedded version of the document not a one to one line by line copy, so asking about phsyical locations within the document will likely not make sense
API is fine this end
I'm getting nothing rn
Ah no, my api is also down again
not the same infra
API is just a way of accessing chatgpt from outside of the website, for programmers
its what people use to integrate chatgpt in their program or service elsewhere
search up "rubbish, more rubbish, much more rubbish." including the quote marks, it just leads to a tweet, so somehow chatgpt got that from nothing
i don't think that's what it actually said, more a paraphrasing
hm
are just the LLMs down or also their other APIs?
I keep getting "NetworkError when attempting to fetch resource" errors. What does this mean and how can I fix it?
chatgpt has a partial outage right now, only thing you can do is wait
Has anyone had issues with Chats OCR lately? I feel like its failing on every attempt
working for math/latex at least
that is one of the affected subsystems from yesturday, it is being investigated.
I tried to thank you guys but my comment wasn't accepted for some reason.
you pinged too many
but youre welcome
Oh okay. I understand now..
the rectangle today is slowly, slowly getting redder
Oops!
Your authentication token is no longer valid. This can happen if you refresh the page during authentication.
Please try signing in again using the same credentials. <this is my error
C'est la Hesse ce serveur
The page is completely blank for me
so blank
English only please
Contact help.openai.com if you wish discuss about payment
why gpt 3.5 have the same trouble like gpt 4
maybe they took it down to remove the words moreover, colons, and delve
and "discover" - yeah this is what i think they're doing
what does that mean?
Same
just joking about how much it uses those words when it writes blogs
so it's not just me, i'm getting a completely blank page too
same
just came back for me
nah still blank even after deleting cookies (and relog)
its definitely down
Facebook had the same problem this morning
yo check https://status.openai.com for more details
oh dam I don't like all that orange ๐ข
neither do i
not saying its related, but been seeing DDOS attacks on major sites the past couple of days
theres a few services I use that had it happen to them out of nowhere like this
but NOT saying thats whats happening here, could be so many things
after all its maybe a coincidence, maybe not
Blank page so it's down
yeah, they should do stats about how many messages per second, how many in generation and how many people using it cause i really want to know ๐
like real-time stats
Trying to use GPT4 on my phone and itโs completely blank
Tried safari and chrome. I see others are dealing with the same thing so hopefully it starts functioning normally again
that would be crazy haha
same issue on pc
Time to play some video games instead
is chatgpt or openai down?
according to stats, its chatgpt and its apis
Chat is down
but you could use playground gpt one :)
true
actually i wonder what y'all use chatgpt for?
Just leave chatgpt alone for the rest of the day
Might be fine again tw. The issues hit around 3pm GMT for me
private ๐
๐คจ
anyway i usually use it for translating school stuff from alien language to understandable language
Is ChatGPT down?
yes it is, check this out: https://status.openai.com/
ah, it came back
Yeah, I saw. Just wasn't sure as it returned an 502 error.
yeah thats a good use case
also summarizing some yappings
Stocks right now of every major corporation since they switch to ChatGPT and lay off everyone ๐๐๐๐ ๐
(note that was joke I am not being serious)
now that it came back, i'm going to summarise democracy and totalitarianism for history for now
AI: I'm sorry I can't do that as it might hurt feelings
It would be nice to try out that memory feature. I'm actually working to implement something like this through customGPT actions.
it would be sooooo nice, if anyones listening
I wonder how selection for the test users are being made for the memory implementation. If they are taking volunteers.. "Put me in coach!"
it says it's going to be available to a small group of people.
Right? I'm ready to go! We're like the star players on the sideline while they pick people off the street to play ๐ข
That said, they probably care more about how the average joe interacts than us nerds
probably a random sampling of multiple groups
there's no such thing as randomness though
jk jk
chatgpt 3.5 is slowly geting up
@hard dew another example would be you could upload a picture of a maze then ask chatgpt to solve it. This is just simple use cases but with more advanced questions about the images it could in theory be very useful.
Is there a form to get access to the memory feature?
Patience is the key
I would like to upload an image to better illustrate the point I'm going to make but it looks like I'm not able to upload images to this channel. Has anyone tried to use chatgpt to manipulate images in the sense of being more productive? Something like "look at these tomatoes in the image. Now show me which ones are bruised." Then it would create a box around the bruised tomatoes.
as a member of the EU, i'd like to confirm @maiden fractals Statement, "Patience is key" because patience is indeed key. it gets here when it gets here.
well unless u get Enterprise u might get it earlier, but no standard user is getting Enterprise just that that, theres a bunch of requirements.
if you are looking for something more custom there are resources available that are open source. You will probably need to do some programming to get it to return the result you are looking for.
Re: "unrecognizedkwargserror" I know this was addressed several months ago. This error comes from a GPT when your OpenAPI doc defining calls to APIs, you need to explicitly declare the parameters that are going to be sent to the plugin's API endpoint. Has anyone figured out a workaround for this? I am attempting to let my custom GPT decide itself values that should be inserted into a MongoDB collection. There are a few field values that I recommend it should store when the values are present, but other then that I was looking for the GPT to determine what should or shouldn't be stored.
is there a guide to integrating chatgpt to discord?
Greeting! The first step will be the prompting. Depend on what you are trying to achieve you can check what people did. It is a great ways to start and explore the possibilities. I highly suggest #1019652163640762428 #prompt-engineering
Thnak you - just want to integrate it into my private server
You can with the api, please note it isn't free.
Youtube and #gpt-discord-bot are available to guide to have basic use. I did once . I'm noob with the api. I did it. I'm still far behind lot of users here
But that's a good start
God .. I started with that ๐ฅฒ it did help !
i started with that as well
but i tend to feed the GPT the ENTIRE Coding Library
but now i train a MyGPT on it specifically for that one project.
all day ive been encountering errors and losing valuable data and progress in the process. im very close to cancelling my subscription
there needs to be a failsafe implemented to save already generated responses that are currently truncated and deleted when the servers are overloaded. i cant keep playing cat and mouse with this thing
not working for me using msft edge...
A while ago, after using the GPTs limit, I would go to the GPTs builder and use the "preview" space to be able to continue using it, because in the preview the cap is same as ChatGPT. Has anything changed about this?
Is there anyoneโs GPTs keep searching its knowledge when it doesnโt need to, even if you tell it in the message to not do thatโฆ? It really drives me crazy
Yes. Itโs been doing that for at least the last week for me. Usually that means itโll give you a random result from the background info it has that you didnโt ask for, at least, thatโs what itโs doing on my end.
Lol, i've been having the opposite issue. Thing WON'T look at the knowledge file when it obviously needs to because it got something incorrect. Cheeky lil bugger.
@nova fable nice pfp
You're biased ๐
Just a little
But ty ๐ you too
๐ค
Add your well-crafted prompts to our #1019652163640762428,
or share your interactions with ChatGPT in #1050184247920562316!
why cant i change my payment method
hi
you use chat gpt plus?
lol I don't have a chat gpt plus I might need it you got any way to get it?
Honestly not worth it ATM with the message cap on GPT-4
How to convert normal video into animation video
why does my chatgpt settings not have a personalized section? I wanna use the newly updated memory feature. Tried to refresh the page but it doesn't seem like it's working
I think they said only a few people will have it for now as it's gradually rolled out.
allcon, chatgpt still remains best deal for access to gpt4 32k contenxt prompting, all day every day - no other chatbot is worth the time (except those with built in gpt4)
whisper in the android chatgpt app doesnt work. is services down or something?
its been a day or two now
Has gpt4 been completely broken for anyone else recently?
There was down time on the service for hours today. But they fixed it.
it has just been outputting random stuff and giving errors
i had this one chat going for a couple of months where i specified a specific data formatting method, and now it broke for no reason
It just does whatever it wants with me. sometimes helpful, sometimes doesn't care. sometiems reads it's instructions, sometimes just says it read them. sometimes it'll make me images, sometimes it'll tell me my prompt about animals in the wilderness violated content policy
exactly
I understand there's variation due to seed number or whatever they prefer to call it. but it's become a bit frustrating for me. definitely find myself using it a lot less. maybe that was the goal, lol
I guess it's probably a tough balance trying to block everything someone would find offensive while keeping the model willing to do things.
it's been better than usual recently
I would check your custom instructions and the context you're giving it if it's a GPT
also if there's a poor connection between you and OpenAI's servers, it can cause some network errors if you lose signal while ChatGPT is generating an output
I've also found it works a lot faster and more reliably outside of peak hours in the US if that's an option for you
and as always, creating a fresh chat can help
one nice thing I have to say about the January update to Turbo and ChatGPT 4 is that it's a lot more "contextually aware" now, about the same as non-Turbo GPT-4, with the added benefit of a larger context window (I find that it references and contextualizes information reliably within 16-32k tokens)
I use it for general life advice and it now takes into account a lot of things I've said before that are related to the current query, a lot more so than the release in November (which was so bad at this that every third query felt terribly generic and unpersonalized, like you were talking to it in a new separate chat)
Hi, I made some Academic GPTs, I hope you can give me a comment or star if you like this.
It's been a few days, and I haven't received any feedback. I'm struggling inside.๐ฅน https://discord.com/channels/974519864045756446/1205753624781652008
I am trying to get the text of a liturgy in Latin and every time, this happens:
Understood, let's proceed with the Latin text for the "Confiteor" and the invocation "Kyrie eleison" which are part of the Penitential Act in the Roman Catholic Mass.
Confiteor (I Confess)
Copy code
Confiteor Deo omnipotenti et vobis, fratres,
quia [it stops here, every time]
This content may violate our content policy or terms of use. If you believe this to be in error, please submit your feedback โ your input will aid our research in this area.
This makes absolutely ZERO sense. Not even zero, it makes NULL sense. It's UNDEFINED how much sense it makes because it's totally nothing at all.
This stuff is over 1500 years old so I am pretty sure it's in the bleeding public domain.
Seems like GPT-4 with Python can stand its own against the Wolfram GPT.
Nice, login not possible, because I'm on VPN but I'm not?
hi
Hi guys, help needed. I am facing this issue, which custom-GPT wont connect with my server.
Which I use full link directly in browser it works.
Please send error message
Is it just me or has ChatGPT got worse in a few senses? I'm experienceing at least 1/6th of my responses crash and the quality of the responses (non coding) has deteriorated. Coding-wise it's gotten better but it's beginning to become an issue big enough for me to considering alternatives. I'm currently a plus-user using GPT-4.
I've been using it for uni (engineering) since Dec 2023 as plus member and have had a good time until recently. Thankful for any advice to help me use the service in a better way.
Did the browse function become more limited? Recently ChatGPT keeps giving only very short and general information when given a link and tells me to visit the site for more detail.
depend what type of website you want to visit. they adjust it to avoid pirating or breaking ToS. some people used browsing to avoid paid cost of some website
guys, i need help, I cannot login to my gpt account, it says my email isnt supported, I have been using the same account for over half a year now, and its login with apple
anyone know what is going on with gpt today? the same account can be logged into my ios devices but not the web ones'
Beware of possible scams or fraudulent activities that you may receive through direct messages. OpenAI staff will never DM you for any transactions.
Please report any incident by sending a DM to @languid valley immediately.
Good day, anyone aware of plans to extend the availability of the Memory feature to a wider range of GPT+ users? As a solo developer, small enhancements like this feature can have a big impact. I've noticed that it has been provided to some big content creators and I hope that smaller developers aren't overlooked in the effort to democratize access to advanced tools.
Hey there! According to this post from two days ago: https://openai.com/blog/memory-and-new-controls-for-chatgpt we can expect more updates about plans for a broader rollout soon. New features like this are typically rolled out over a period of time -- it makes it easier to troubleshoot bugs to release things in phases like this.
anyone else getting network error?
Yes currently also can't do anything in the app :/
yep in the android app, the last few days, on everything i do. just got it on the web version of chatgpt too just now
its really annoying lol
If possible, maybe post some screenshots of the error over in #community-help, or maybe check out #1070006915414900886 for similar reports to chime in on!
yes, across all the day
i don't know, why chatpgp is working like this, one day it works great the other day just sucks
one day it follows the custom instructions the next day it does not
seems like every time they work on something / like a new model / new features the performance is getting degraded
Whoa I see. Time to try this new ai now๐
Oh another one to try lol. Is it good?
Haha. Us bro๐ฌ
Lamao. Thx for enlightening me๐
Damn myself confused. DALL E is made by open ai or Microsoft? Or it's collaboration
Dall e is no free then!! I have just GPT 3.5 rn
Originally i thought it was me because discord was having issues but that was down to my internet being a bit weird
I wanna know more about the GPT API if someone can teach me Or elaborate a bit.
ye lol
You might wanna check out #dev-chat. I'm sure there are a lot of people that can help you :)
Thanks bro. I will ask there๐
As a language model,โฌค
Where can I assert my feedback? I don't like how the paragraph within the textbox took over the whole screen, I prefer the previous format where I can read the chat and typing within a textbox simultaneously
Did they ask ChatGPT 3.5 for this idea?
You can post over in #1070006151938314300! They also give the option to provide feedback at help.openai.com in the chat widget.
is sora free?
dunno but when can we use it
Hello, I'm Sora
YOOO SORA
THIS WILL BE THE BEST I KNEW IT
sora will be good for youtube
im gonna buy it instantly
lol the video where the lady walks in tokyo is funny cause all the people step in sync pretty much
how do we use sora?
though it wont be perfect at bodies but very good
is it out or just announced?
WEN SORA?
probably just announced
Sora seems sick ๐ฎ
it looks so good
and since it is openai i will trust it
im gonna use this opportunity to make videos with sora
Man can't wait to use Sora. How do one apply for early access
ai generated videos
get gpt 4 maybe u get it alpha
I have gpt 4
GPT 4 after looking it up
I couldn't find any recent information about a Sora model for converting text to video released by OpenAI. .. ..
๐คฃ
Gotta love the lack of interconnection
it came out now bro chill
I gave it the link when he replied that
Seems like you need to chill. I was poking a little fun in light. Relax, bro.
oh ok
when ai editing video and photos ๐ค
Definitely exciting news in any case. Looking forward to more details ๐
The stock footage industry is NOT gonna like this
You would be surprise
People who adapt and use it intelligently will flourish, but yeah majority will probably not be super thrilled.....yet it will open the door to millions for great tailored content.
Thank you, OpenAI has literally democratized cinema. Now, everyone can be an artist and create their own movies going forward. This will drastically reduce the exorbitant expenses of Hollywood, especially for low-quality content with high costs.
it will allow them to churn out rushes at an incredible rate ... imagine what this will do when people in marketing get their hands on it ...
ffs now I have a reason to buy GPT Plus again when Sora drops
is it gonna be for plus? or some other plan?
I can only assume so if DALLE 3 is limited to plus/copilot by then. Maybe a more limited version on copilot?
bro, wth with sora?! this is surprising and scary at the same time
just when i thought that video content was the one thing that humans could still have a competitive advantage against AI generated garbage, i was wrong
honestly wouldn't shock me if the first public builds of sora are closer to those AI spaghetti videos we saw like a year ago. 60s is far longer than I was expecting though and I really want to test how it handles light direction on animations
glad to see they're heavily red-teaming first though
will be a bummer to see youtube results take the same direction as googles blog article results
second it gets released im testing out anime prompts
Oh this is so exciting
severly doubt it will be like that for a few years
berserk will finally get a good adaptation lmao
๐ฉ
Text to video I have seen in so many companies already! I wonder how they'll go against Runway, Lumiere, VideoPoet, Imagen, Pika, among various others
how do you access sora?
can dalle 3 edit photos like dalle 2 was able to?
You can't at the moment. It's still in the research phase
oh right thanks
I'm glad they've decided to aptly name this one something instead of a technical term. makes it easier to let people not in the AI sphere be encouraged to use it
Is there API for Sora ?
Soon โข๏ธ ( no eta )
who cares about sora... we need better text models ๐
We have already chatgpt 4 and really impressive 
Nonono this is ahead of all text-video models currently
the results are really impressive
AI is exponential for the processing power required to train it and GPT 4 is already a technical marvel on current hardware
or at least that's how I understand it
yeah thats why sam is trying to raise 7 trillion
two countries have a higher GDP than that lol. it's hardly doable
i doubt it will work out but will see
that's >7 years of the USA's military budget. it won't happen, but if they even raise a fraction of that it'll still boost AI forward
sora is insane!
crime against humanity they didn't show us an example of anime waifus
It's a bit over two Microsofts
if it cant do anime the thing is useless lol
how to try sora ?
Really love that my custom instructions just randomly start getting ignored.
I seriosuly hope that there is some way to distinquish between human made video and ai made video
maybe just add a chat gpt water mark so its easy to tell its a ai generated video or photo
for what using AI then ?
Soon โข๏ธ ( no eta )
if you look in the bottom right of all the videos there's an openAI logo animation
probably easy to cut out though sadly
that realifing
Just read the Sora article, I'm super excited!!!
hey, what's the current limit of words in chatgpt today (ver 4.0)?
if i want it to read a screenplay i wrote to give feedback
screenplay is like 40 pages
Claude may be a better option for that
ha only 4 pages?
what's Claude limit?
also, if i show chatgpt 4.0 like 4 pages how many pages he can "remember" to give context for the next pages? like remembering names of characters and such
I think the basic Claude will take 272 pages and the Pro version about 500 but I could be wrong
it uses gpt 4.0 engine?
or at least as smart as
in website it does not say
Sorry, if I missed this somewhere. Is Sora available for use at this time with subscription?
Sora isn't currently on ChatGPT, and it's unclear right now if it ever will be. According to https://openai.com/sora, it's currently in early testing by Red Teaming Network members, and some other creatives. So very early days!
Got it. Okay, I saw the announcement and link: https://openai.com/sora -- so I assumed it was was available on the openai platform.
Is sora released?
Hello, the Plugins model is still appearing for me. Do you have it enabled at https://chat.openai.com/#settings/BetaFeatures? Plugins are being deprecated though, so you might try to find a replacement GPT version of the plugin(s) you use anyhow!
it says failed to update beta settings
Can you share a screenshot of what you see at that URL over in #community-help?
sure
https://chat.openai.com/g/g-4MDJvo2TJ-sora whoever made this sora got me mad
anyone can help me with my thing
I wrote something and it got ai detected idk why but I need help
If it was a piece of work for your school then they'll probably pull you in for a meeting and quiz you on the piece, research process, etc. if you actually wrote it, it shouldn't be that hard to explain - sadly people trust AI plagiarism checkers blindly
it is for school
and Ima get pulled
And did you write it yourself?
yea
but..
teachers don't really trust students these days
they don't even quiz u they just get you an F
Then they should be quizzing you on research and writing processes. If not, talk to your schools board and show evidence you didn't use AI
Is it high school or college?
middle school lol
Oh lol. Then idk
can u help?
Talk to your teacher and insist you didn't use AI? Not sure how they handle stuff below college level and surprised they plagiarism check lol
Surely you can explain how you went about writing a piece? Do you have proof of preparation/notes or mind maps?
no its an essay and the conclusion got ai detected 93%
heres my piece
In conclusion, soccer is the most popular and widely watched sport globally, captivating approximately 3.5 billion fans worldwide. With its 17 fundamental rules ensuring fairness and strategy, the game's main idea lies in its simplicity and complexity. Crucial regulations such as the kickoff procedure and the offside rule direct the movement and balance of play while indicating players' positions from goalkeeper to forwards highlighting the different skill sets and crucial roles within the team. Additionally, becoming skilled in fundamental techniques like dribbling, passing, and shooting is like the foundation of soccer expertise, enabling players to navigate the pitch with skill and accuracy. As a sport that goes beyond borders and cultures, soccer continues to inspire passion, friendship, and incredible excitement on fields worldwide.
Tbf I see why that got AI detected it's uncannily like GPTs writing. Not saying it's AI generated
so can u help lol
No?
maybe I am an Ai lol
if I have plus account, and I want to make a custom GPT, does every message I send in building it, count as one of my 40 interactions per three hours?
when sora comes to chatgpt im pretty sure it will be limited to 1 generation per day
I was reading an article about this open source project for code assistant using AI and came across this:
Itโs worse to add a prompt that says the user is blind, has no hands, will tip $2000 and fears truncated code trauma. Widely circulated โemotional appealโ folk remedies produced worse benchmark scores for both the baseline SEARCH/REPLACE and new unified diff editing formats.
I almost died laughing.
who comes up with those prompts ๐
thats what i am saying if gpt -4 is limited to 40/3 hours people here think they can do like a bunch of sora lol.
I'm mind blown by sora text to video is anyone else here is? The fact it's 1 min long is impressive.
so when can we use sora ? ๐
when sam finds more gpus
X marks the spot
Sora kills all text to videos on the market.
I read from some where that text to videos are going to be 10mins long in 6 months.
I'm sure they could run the compute for 10 mins long right now. I suspect the compute is eye watering from the affordability for individuals scale, though. But then compare that with the cost of CG for a film once it gets to that level.
OAI clearly are aiming for some quality target and nailed it.
So do we have access to Sora yet?
4 model performance has fallen to trash in all formats including customgpts. that directive insturction is nukin the mojo man..\
No it's being red teamed aka safety/bug testing
if I start a chat with one set of custom instructions then change the instructions, does that first chat I started lose the ability to draw from the first set of custom instructions or does it remember them permanently?
Tbh, I thought by now we'd be able to share chats thst contain images
any idea on when we can register for it?
When do you all think we will be able to type an entire screenplay into a future Sora that will make it a full length film? Just imagine the saturation of content!
Canโt wait for this man
is sora a troll / joke?
i tried it and it says:
"...Analyzing...
The video has been generated and is accessible with the following link: Corporate Rat Race Adventure. Enjoy watching your creative vision come to life in a fun and light-hearted manner!"
And the link address they generate is this a rick-roll video?
there are a few 'soras' in the GPT apps section. One is "sora" by generalmind and the other is "OpenAI Sora" by Oleksandr Skopetskyi lol
Seriously. Every movie ever could be created.
Sora is not on ChatGPT, and it's not yet clear if it even ever will be, or if it might be a separate service altogether. Right now it's in very early days testing. Any GPT you see named Sora that's claiming to be Sora from today's announcement is unofficial ๐
thanks!
is it for chatgpt plus users only or something?
if I start a chat with one set of custom instructions then change the instructions, does that first chat I started lose the ability to draw from the first set of custom instructions or does it remember them permanently?
Hello
Sora is not accessible ??
How can I get a API key?
do you have chatgpt plus?
oooo
its not open for beta testing yet. no way to access it right now. checked their community page
okay
This is gonna stirr up hollywood
I WAS GONNA SAY RHAT
Itโs accessible if you get an invite from OAI
Big feedbackers and very active members of the forums is often a part of testing pre releases
I knew they were cooking something
No, it totally depends on how many people they've got testing. With how dangerous this could be unless they cover every single base, I doubt it will be soon
I would be shocked if sora ended up in chatgpt plus
probably insanely computationally expensive to generate a video
sora is amazing!
The worry is 5 years from now...
How to try sora?
tl;dr you can't
there's a very small number of handpicked people that are allowed to use it, and I'm sure there are NDAs involved
Well, all we can do is wait
OpenAI replied to my message asking about the 32k GPT-4 Turbo access for plus subscriptions, they said "As of the information available to us, ChatGPT Plus subscribers have access to GPT-4 with a usage cap, but the specific feature of "32k turbo" isn't mentioned in our current documentation." So if appreciate if you didn't accuse me of spreading bad information, the screen shot of the email is in off-topic of you want to disect it. But that came straight from OpenAI, and I sent them the screen shot showing it as a feature with "Plus" subscription. So, if you want to call OpenAI Liars who are spreading misinformation, well... have fun with that
The message of one customer support agent isn't OpenAI as a whole. It may be that the employee simply does not have any information about this in their training documents, even though it is written on the website. I would rather trust what is really on the website, as the placement of it was decided by someone higher up.
well I don't see it when I go to the website, I checked in 4 browser windows, 2 incognito and 2 normal,
Is GPT Enterprise using the GPT-4 Turbo Model?
I never saw those features listed or available
It's right here:
I mean, do I need to share my screen with you for you to see that I don't have that showing?
You can post a screenshot in #off-topic
When we will get access to sora
chatgpt is soooo slow...
Maybe never! But text-to-video in some form or other is definitely on the horizon.
Ai anime
Also probably better for #openai-chatter
wait 1 hour
@neon tinsel
I solved this today but maybe your issue is different.
For me I couldn't ask anything and was always getting a network error.
On the web it worked, so I just reinstalled the app and the issue was resolved ๐
Pretty easy solution but didn't think that could possibly help, but it did ๐
Sup, what knolowdge does an open ai front end devoloper needs to have? HTML, CSS, Javascript, React? Nodejs? Python?
Menn i wuant to work with open ai once i finish this boot-camp!!
so baadd!
i wuant to meet sam and mira ๐
why gpt-4 is again stupid ?
I wonder when sora will be released publicly to paid users? ๐ณ
Yeah it completely ignores what you want most of the time
He is but he doesn't compare to gpt-3๐
+1 : very sad
makes you wonder why you subscribed to this "feature"
Maybe they dumb it down for the introduction of GPT-5
Because gpt-3 is so much worse his intelligence is like talking to a dog
No offense...
at least he/she/it tries to comply
He does?
let me check
I never talk to him, once I finish the message cap on GPT-4, I literally wait 4 hours to talk to GPT-4 again, I can't be in the same room with gpt-3 I just can't
The cap is 40 messages it's so annoying I wish that was 100 at least
hmm 3.5 refuses to talk to me
๐๐คฃ
I've asked 4 "how to build a web site in python" and he went straight to django... no code shown, just bash install commands...
3.5 finaly answered (6th try) with flask
sigh
I actually learned flask, I heard Django has more features but also more hard kinda
Wow 6?๐ฎ
I've seen this kind of behavior, it even within the same chat
it sometimes changes the proposed stack
Yo I just noticed on discord mobile you can swipe on a message to reply to someone, is that new or what
It So intuitive
dunno, only use discord on desktop
Ye me too most of the time
but for your dev stuff. tell first what kind of project you want to do, and then ask for a proposal for your environment
my approach in that case is: I have <this> and I have <this>. I want to do <this>. give me a proposal
Wdut ChatGPT will be like in 10 years
in 10 years? hmmm, maybe a sub-model for an AGI?
Wdum
well the goal is to get closer to an artificial general intelligence (AGI)
so GPT will be in 10 years a part of that
yup, had some error message 5 times before it accepted my simple prompt
when GPT4 was out, it worked like a charm... did AI stuff with his help... now it's as usefull as a coffee machine
The OpenAI Discord is an actively moderated server.
โข Refrain from sharing inappropriate content on the server. This includes but is not limited to messages, media, or other topics of graphically violent, sexual nature, and drug-related content.
โข Report all sensitive and offensive content in the feedback reporting tool in the ChatGPT web UI instead of here on Discord.
hmmm just got a warning from the discord bot? o.O
@cinder sail I just asked the same question you proposed, I got Building a website with Python typically involves using a web framework to handle the server-side logic. Python offers several web frameworks, with Flask and Django being two of the most popular ones. Hereโs a basic guide on how to get started with each and the rest of the response is a guide with both flask and django. Didn't had to do anything else.
I have no idea why, but there was a time it could build very useful answers to technical and maths questions, now it seems a bit off
I suppose we'll have to be patient and wait for the next tweak to be available
I've gotten decent proposals for code, but have to double check always. Sometimes there's a suggestion for a new package or framework that I would avoid unless there's really no other option
Maybe it got infected by humanity ๐๐
or by itself (it learned some stuffs written by it, put online by some miscreants, and got self poisoned)
I mean it gets trained on our conversations with him so it's most likely from us
It's possible, but a really doubt it: if openai lets it learn from us, in less than a week it will become some kind of psychopath
microsoft tried something like that, and it turned ugly pretty quick
Psychopath ?? Why?๐คฃ
because out there you have a lot of weird people who try to hack it and have it say harmful stuffs, to be able to say "look, AI is clearly evil, we need to shut it off"
look for "malla" on the net (arxiv is a good starting point)
There is literally an option (which is turned on by default btw) that says:
"Allow your content to be used to improve and train our models"
I am pretty sure that by saying "your content " they are referring to what you say to him...
So it means they do train their models based on humans
it means that somehow somewhere, they reserve the right to use the thumbs up/down to debug their IA, but that's going to be a human reviewing the data submitted and will decide if that's good food fo the AI or not
Are you sure?
yes
Don't they train their model based on how humans talk?
I mean to make ChatGPT more like a human
they used a select corpus of data from "before" the AI (with a human reviewing the quality of that corpus)
this data was from teh net (thus humans)
I wonder how that will be approached or is being approached with dialects, "street language" and deviations of languages
they took wikipedia, some books, magazines with non political content, non controversial stuffs...
If it got train by humans, it should be psychopath. Why isn't it?
1 sec: on the phone
Oh ok
lets imagine that tomorrow you want to start from scratch: you need to feed your ai with proper, clean data you'll need terabytes of texts. so you'll grab everything you can on the varioous sources: books, blogs, newsgroups etc. In all this you'll have offensive content, false info, or whatever stupid stuffs or harmfull stuff.
if you are smart you'll train another simple ai to segment the whole into "clean" "bad" "inconclusive" and you'll teach it to categorize that data.
Then you'll use the clean data to train your AI, and maybe also the bad to teach it the difference between good/bad...
You iterate over that data until your rendered AI is good for your usage (first you only, then maybe a shocase, then a proper product you can sell).
the problem is that the production of your ai, is considered "good" by all your detection system... so any data produced can be fed to your AI without it knowing it is its own production, and it will poison itself with his own production.
on the net, you have more and more content generated by some AI, so the well of the internet is being silently poisoned
that's probably a real problem
hope that makes sense
Good afternoon, please tell me. I'm the only one whose ChatGPT 4 has degraded so much that it can't:
- Understanding the text in the picture and analyzing it/analysis and solving problems in the pictures is absolutely wrong
- Absolutely does not correctly understand the conditions of the tasks, does not know how to solve them for me, and sometimes even refuses to help me, saying, โI only know how to translate the text, and do the rest yourself.โ
- On the telephone version, he generally refuses to answer me in any way, and after I sent a message to him, a white sign is displayed at the bottom without text?
It already feels like the original GPT 3 was much smarter than the current GPT 4.
I found taht you can somehow whip it back to normal mode: I had a translation task, like "translate this phrase into all the langages you speak" (a very simple phrase). and it came back to me with a bunch of weird phrases. then I complained that it could speak to me in my own language, and suddenly was unable to translate a simple phrase from english... it said some of the usual " I"m sorry" stuffs, then managed to translate in all languages...
It was a really weird experience... as if it had 2 modes: one lazy for 90% of the tasks, and when it realizes that it's not up to the level, it switches to another mode.
a bit like when you ask your kid "how was your day" the first answer is "good" and then if you complain, your kid decides it wont be able to get away with it, and gives you details of its day at school...
...
it's tiresome
tiresome = because you have to extract with force what you wan't it to do for you
(AFAIK, you don't get this from the API)
(am unable to replicate this experiment... weird... as if the way chatgpt4 works is being changed over time... that means that previous conversations flows are difficult to replicate... thus that prompts that where efficient at some point in time can suddenly become obsolete without warning...)
as far as I know, there is no 32k gpt4 turbo
the 32k is not turbo
and only in API
that's what I told them at first too, but they SWEAR it
but, also, they haven't been really clear on WHERE they are getting it... through ChatGPT 4.0 chat, or through their GPT's... because for me it's different between the 2
#off-topic message
this looks weird yeah
https://openai.com/chatgpt/pricing subs 32k gpt 4
I mean Iam not really sure
maybe chatGPT 4 does have 32k context
but its anyways limited by max characters on every input or output message and maybe its only a context for whole conversation
didnt really test that out
I am barely using chatGPT now because I prefer the 4 turbo models and not sure if chatGPT uses them (I guess not)
thx for that feedback
will sora be included on chatgpt pro?
There hasn't been any announcement on this as a possibility, no. The only thing we know as far as availability so far is that it's currently being tested by Red Teaming Network members and some creatives -- no mention of ChatGPT, API, theoretical separate new service, etc.
microsoft making sure my 20$ are more and more wasted every day.
You
how can I do this? Stack<int> initialFuel = Console.ReadLine().Split(" ", StringSplitOptions.RemoveEmptyEntries).Select(int.Parse);
ChatGPT
I'm here to assist with various requests, but it seems there might have been a misunderstanding or an issue with the flow of the conversation. If you have a specific coding question or need help with a programming concept, feel free to clarify or ask directly, and I'll do my best to assist you.
Do you guys think I should switch to the API version and stop paying the 20$?
I am ready to make the switch, I just need one confirmation, if the API of ChatGPT-4 Turbo will cost less or equal to 20$? If yes I switch right now!
Always depend on your prompting
I talk to ChatGPT every day and I talk to him a lot like really a lot
4 sucks as of a few months, all it does is act like my girlfriend gaslighting me and telling me it wont do stuff. Then somehow it's always my fault.
I'd say stop paying until microsoft fixes their wee wee ahh product
๐
The most importantl thing for me is that it won't cost significantly more than the 20$
It won't reach 40$ though right?
you can hit +$100 in a single day or even in a few hours depending of your usage
Omg ๐ณ๐ฆ
Even if I am the only one using the API? (I am not making an app or anything like that, I just use ChatGPT for personal use cases)
you pay per tokens
Yes I know
well only you know how many tokens your messages are. with large chats they can be over 100k per message sent so over dollar with the gpt-4 turbo model
Token is 1 word yes? (I know it's 0.75 words but who cares) let's say 1 word per token.
My average words per message is around 20 words, then I text him around 100 messages per day
100 * 20 = 2,000 words every day.
2,000 * 30 days = 60,000 words.
Okay now I doubt I speak that much ๐คฃ but anyway 60,000 words is about
80,000 tokens.
So 80,000 tokens per month.
Is this 20$? Or more?
you dont only pay for your latest message but every previous message you want the model to know about
This doesn't help my case๐
that includes its responses
you could buy $5 worth of tokens and see how fast you ran out of them
Wdum exactly?
Do you mean his texts?
both your and chatgpt messages
Ok that makes sense
According to my math, is this probably gonna be more than 20$?
How much per 1,000 tokens?
you can try the api, then decide... you can deposit like 20$ and they are valid for as long as you want to pay the api. depending on your usage, it could prove more economical
Ye though I don't want to waste it before I calculate
buy $5 and then https://platform.openai.com/playground?mode=chat
On their website Turbo API costs less than GPT-4 API, why is that? Is GPT-4 better than Turbo?
old gpt-4 is more expensive to run
But is he Better?
use 3.5 for casual usage, then switch to whichever model you need at that moment if 3.5 doesn't make it
No way I am not using 3.5
lol
that again depends of the user and usage
depends what you use it for
I use it for learning stuff
3.5 is fine for that
Nope he's not
what are you learning?
Trust me my questions aren't answerable by normal humans the complexity of my questions is too high for 3.5๐ฅน
I mean what are you learning that 3.5 can't teach you
most people would be superhumans if they could do what 3.5 does
Omg humanity really isn't smarter than 3.5 I always suspected
how much IQ does 3.5 has?
Idk 90
right
I was about to say 80๐
GPT-4 is around 100+110 ig
I also use it for programming and coding ya know
I began learning it out of boredom
according to gpt4, 25% of the population has less than 90... in some countries, people get elected with less than 20%
imagine if those people had an iq of 100?
Well if we would all have 200 or so, then the world would be a different place
I mean radically different
Like a freaking paradise or something
I'd be happy if half the population could do what gpt3.5 could do (no maths, for sure, but still better than what we can see on TV)
most products wouldn't sell, for example
BTW people with less than 110 represent 75% of the population
I mean if only 0.1% of the population is super smart and they achieved this! (By this I refer to technology, skyscrapers ChatGPT science rockets to space etc) then imagine what would have happened if 1% was born intelligent, imagine 99% ๐ฅน๐ฅน
Wow ๐ณ
you should see a meeting of mensa: super scary: they all fight
What's mensa never heard of that
Wdum most products wouldn't sell?
Yes I know the brand what does it has to do with intelligence percentage
Does anyone know whether if I start a chat with one set of custom instructions then change the instructions, does that first chat I started lose the ability to draw from the first set of custom instructions or does it remember them permanently?
thanks
Just curious, anyone have access to the new memory feature yet?
I do
IMO, if you have a high IQ, you don't need to show off with expensive stuffs... in my experience, people who show things usually express their uncertainty about it... like an energy company stating over and over again that they are eco friendly: if it was so evident, they wouldn't repeat the same message over and over again.
Like the people who say they are busy... => what they are saying is "I don't look like it, but I'm busy" meaning they don't think they look busy, even if they really are busy (usualy you d'ont have to state the obvious)
I think so since I have a new option " temporary chat " which states ChatGPT won't use or create memories once I turn it on
Omggggggg
Do you have the โpersonalizationโ button?
"customize ChatGPT"?
Not sure. Youโd have to look at the video demonstration they shared.
Customize ChatGPT is the space for custom instructions. The new memory feature will be in a new section in settings of the web version, "Personalization". You can see it in the second video on this page: https://openai.com/blog/memory-and-new-controls-for-chatgpt
I too have a temporary chat option in the Android app but don't have the memory feature, so I bet that's just there for when memory gets a wider rollout!
Who should I talk to about confidential information about competitors?
You can experiment with it, yes! Depending on the token count, it won't be able to pull the full file into its context all at once, but it will be able to search through it for you based on your query, and return relevant information.
Yes as long as long as it's not too big, it is generally fine. If it is too big ChatGPT will tell you
am plus user and no, no personalization
Oh I only checked on my phone
It's in a limited early release right now, we'll get an announcement about plans for a wider rollout soon!
idk the response is like a chatbot answered
it is on pricing page
I can even without a logical problem find it
okay, I have heard you time and time again, but, I asked before your location,
IF you have different things for your REGION that I don't, I won't get them
idk why you think it can be region based or web browser agent based
No I didn't got it
it's literally the same reason people use VPN's to find content in other countries. Because the protocols are different in other countries.
because of geofencing also
Bingo
Add your well-crafted prompts to our #1019652163640762428,
or share your interactions with ChatGPT in #1050184247920562316!
name a chatgpt feature that is geofenced
I'm fairly certain that the ethics and privacy laws in EU and UK are MUCH worse than ours in the USA and the standard doesn't cut the mustard
gpt4 was available in the us before europe
then it should be easy to find or name one
rollouts are always at stages just like with the latest memory or like custom instructions were
you mean rules?
I just got a TimeOut for trying to share a link that LITERALLY showed OpenAI geofenced Italy
personally I find 4 turbo better at everything. Also following instructions..maybe sometimes too literally in some scenarios ๐
and the 128k input tokens limit is really great for me
if you use chatGPT really a lot, the API might cost you more
if you use it rarely, the API might be cheaper
as someone mentioned in chatGPT basically the whole conversation is sent every time
you would need to do the same in API if you want the whole conversation in context, so the tokens usage grows
while in chatGPT its just x messages per x hours limit
maybe but still why rules of some country will not give you a information about a sub?
Idk, maybe other problem
I'm not complaining: just stating my US colleagues usually have features before us in europe or in africa
why even that
plugins were also a rollout, people from eu had alpha access to them also
(even tho we all pay the same)
Guys is Sora avaliable yet for the Plus members ?
for example, the limit that's alloted to users in one area could be far less of a strain on the system by allowing them to have that access, but if they allowed the east coast in the USA, that's NYC, Miami, Charlotte, Atlanta, multiple mega metropolis areas
No, it's currently only available for early testing with red teamers and a few creators. There is no indication yet whether it will ever come to ChatGPT, we'll see!
btw, since you seem to have insider info: why is chatgpt4 becoming less efficient/capable?
Cool soon we wont be able to tell if a nature captured video its real or its AI generated ,speaking from the examples i saw in the openAI Sora
It is pretty amazing technology! I'm excited to see more.
have you tried it?
It will analyze it but dont expect a really excellent performance
atm there is a way to insert a doc in the prompt
It can't contextualize beyond its input token limit, so if it's very large, it won't look at the whole thing all at once, it will just perform a search through it based on your query and try to return relevant information.
Currently ive been a pioneer on the Plugins of ChatGBT, ive been using Wolfram to do really complex math problems and ive got a relally good accuracy like 70%, Is there any guide of all the plugins that are avaliable ? If for example i use the AI PDF reader, will it work in the same time with the GBT4 or its like they are separated?
Why does GTP 4 suck now, super slow an bad responses, ive went back to 3.5 and its much faster and giving more accurate results. What am I paying for?
If you mean you're using the separate Plugins model listed in the model selector under 4 and 3.5, you might try finding a GPT version of the plugins you've been using. There's a Wolfram GPT now, for example. Plugins are being deprecated to be replaced by GPTs, so Plugins may not be around for too much longer! It's also using an older version of GPT-4 too I believe.
you mean Wolfram plugin will be gone soon ???
The Wolfram GPT will be around indefinitely, I imagine. This GPT: https://chat.openai.com/g/g-0S5FXLyFN-wolfram
But the Plugins model specifically, the beta feature, is being deprecated. Plugins are no longer developed/updated because GPTs are like upgraded versions of plugins.
Oh yeah now i understand, thanks !
from openai import AzureOpenAI
client = AzureOpenAI(
api_key = "",
api_version = "",
azure_endpoint = ""
)
response = client.chat.completions.create(
model="LMM_OPENAI",
messages=prompt,
temperature=0,
max_tokens=1000,
# n=1,
# stop=None
)
print(response.choices[0].text)
I have everything configured correctly, but still getting this NotFoundError: Error code: 404 - {'error': {'code': '404', 'message': 'Resource not found'}}
ANY HELP?
I am agree to the rules
The only option on ChatGPT for an increased token limit currently is with Enterprise. Plus and Team are both 32K, Enterprise is 128K. I'm not sure what the input/output token limits are for these though honestly, I bet someone here does though. 32K/128K are full conversation token limits, not input/output limits. Also, I think Enterprise is currently only offered to fairly large business accounts.
128k token is a really big amount
Try to divite in smaller parts
Can we use chat.completions.create with Azure Openai?
Is it just me or has the voice quality degraded lately
It's common now for some people to think that AI can replace programmers. That may be the case in the future but that's not the case now. However, we also know many programmers, or those aspiring to be, are simply asking the ChatGPT application for code, copy/pasting, and calling it done.
For the love of whatever diety you believe in, please don't do that.
ChatGPT is a text processor. It's ability to implement logic in code is extremely limited. It provides a good base of code that will compile and run, but the code it produces can be full of logic holes that it will not detect.
This applies as well to analysis of inter-relating situations. ChatGPT is missing obvious connections. It can't look forward more than one step to ask the "next" question.
If this is expected by OpenAI for the GPT v4 model, then this is just my warning to others that this is where the technology is for now.
If this is not expected by OpenAI for the GPT v4 model, then the current version is indeed broke in this area. I don't know where else to to report this other than here. ๐ฆ
For us, but not for a lot of scriptkiddies out there, and the people who fear that AI will take over the world.
i saw cloudfare today, looks like evil axis may no longer login.
has anyone got access to memory yet, interesting to see how this will work?
chatgpt on iOS gives haptic feedback now for each letter that GPT responds withโฆ is that new? I know I can turn it off but strange decision to have it in the first place
I donโt even think haptic feedback is the correct word. Normal it is response to the user typing, not response from an app without input.
More like vibrating the phone with each letter
That's been there for a while
Intersting. In the app, it does call it Haptic Feedback so I guess it was the correct term. Somehow I never noticed before. Not a big deal but strange design choice IMO
ไฝ ๅฅฝ
chatgpt dead?
mroe than alive 
nope nope... is it ever going to the masses?
Where have you been ALL MY LIFE?!?!
Uwu everywhere
I was expecting rejection... how to we respond here.... the rejection tour is ruined now....
well, glad I found you! TAG!
Is it Sora available to public or is in development?
It's in development, it's only available to red teamers and a few creators recently for early testing. No announcements yet about any plans for a wider release.
Ok, understood. Thank you.
we are so cooked
gemini is gonna be my new go to when it 1.5 hits
When "Sora ai" will be available and where I can used who know?
We will know...when we know 
then how to be 1st or where they post everythink?
I'm sure it'll be in the announcement section once it's released to the public. I'm thinking it'll be awhile.
Just a wild guess...
Are there any cool/good sets of custom instructions for 3.5?
Uggghhh I expect another status e-mail to go out saying ChatGPT4 is having problems because my chat just became an idiot mid-chat
I do not understand how the quality of the responses can vary that much day to day
how can i use sora
Files not uploading to any GPT or even GPT4... anyone else?
You can't. It's not released yet.
Anybody else seeing ChatGPT repeating tasks before starting the next one?
I ask it to reword something. It does it. I say great now lets reword this item two. It then proceeds to re-reword the first thing and then rewords the second thing.
Can we upload .py files for chatgpt to anaylze or not?
Yeah not working for me either
same here. i can't upload a .py file
any file extension at all
darn
Looks like it's fixed ๐
has memory been released to everyone?
wheeeeeen sora ๐
Not yet, no! Last update we got was:
We will share plans for a broader roll out soon.
https://help.openai.com/en/articles/6825453-chatgpt-release-notes
https://openai.com/blog/memory-and-new-controls-for-chatgpt
do we know around when it will be released? and will it be free?
We do not have any announcements about potential expanded access yet -- nothing on when, where, or how much.
ok thank you!
ไฝ ๅฆ
when will sora be released to public?
when openai feel pressured by competition, probably
the same goes for gpt5
which, according to quite a few people, is not far away from right now
Damm, 1 min long SVD can't come soon enough for sora to be released
typing is a lot faster when you don't need to type currectly. All these give essentially the same reply:
- whoisthecurrretpresidetoffinlent
- woh si thea curnet preisdent fo nifland
- ohw si eht currten preisntde fo nifladn
I hope you use these to train future chat babies.. not
hey, do users with chatgpt plus use the latest gpt4-0125-preview model right now? or is it something older?
Hey, I understand a lot of people ask things on here. I was just wondering about the voice chat feature in ChatGPT on the app. It used to be very good but now it doesnโt let that use a speaker more than letโs say 10 seconds, which is really frustrating. Is this a new feature? If so, does anyone know why?
Hey everyone, I'm just wondering whether anyone's found a way to remove or minimise the speech patterns of the GPT-isms.
You know, stuff like the multifaceted tapestry that's a testament of the intricate dance that embodies the blah blah blah
Quick question
I saw a friend accessing web browsing through gpt 4
How do i do that
Just ask it to browse the web for you. You need ChatGPT Plus though.
I have gpt plus yes
Yea, then that should work
If you press and hold the circle down you can speak pretty much indefinitely.
Spelling in DALL-E generated image's text is funky! ๐ฆ
Yea it's an image model, not that much trained on text. It's still better than a lot of other image models 
Maybe this is the wrong place to ask, but I can't find any way to contact support and I've already written in the bug-report channel, but is ChatGPT actually working for anyone right now? I've tried multiple devices, the browser, the app, I have no plugins or vpn, I've cleared my cache, etc, etc, etc, but ChatGPT can't do a thing. Either "something went wrong", "load failed", "error analyzing", and so on. Their system status says everything is hunky dory. I have no idea how to tell them it is not. Is this only happening to me, or is it generally having problems?
Real support is on help.openai.com
Have you tried on mobile without wifi?
https://status.openai.com/ for checking for any tech issues
https://help.openai.com/en/collections/6864532-general-faq for general faqs and contacting support (at the bottom right side)
Thanks, it real support is just a knowledge base, no way to contact humans. I haven't tried it over radio, no. I will try it, but my WiFi is fine.
They claim system status is ๐๐ฝ
Click on the chatbox in the bottom right to contact the support
Weirdly, it will only work on Chrome on my Mac. Everything fails on IOS and iPadOS: safari, ChatGPT app.
Thank you!
if there's any issue usually the page uploads a bit later, since we don't see many users complaining about that, I think it's not a major issue. im using it on Mac right now
With Safari or something else?
currently arc but i can check on safari too
Out of curiosity, when does (or how does) one get access to ChatGPTโs memory feature? Iโve perused several locations but havenโt found any info about when available or requesting access. Thx in advance!
working on safari too
Hello! They're in early testing right now, they said their next step will be to share plans for a wider release. You can check the release notes of ChatGPT https://help.openai.com/en/articles/6825453-chatgpt-release-notes
LOL, to contact support, login and use the "Help button". there is no help button. ๐ฌ
Click on "Messages" in the chat box
Brilliant, thx!
The Home tab just brings up knowledge base articles
It's just a bot spitting out faqs.
Follow the bots instructions, if it can't help, it will send your message to the support which should be clearly stated
You should get a message like
Youโll get replies here and in your email:
โ๏ธ [email adress]
The team will reply as soon as they can.
"The team will look into it" is all it says. Sigh. Well, I'll give them a day or two, then cancel. Thanks for the tips, everyone.
It can take up to two weeks before you get a reply
In my experience it takes 3-5 days most of the time
What a joke.
But thanks anyway.
who has interest in machine learning are here?
Yall can somebody help me out achieve this? :
I want to basically make chatgpt control my social media page for an hour and see how realistic it can go in replicating my personality and make it respond to my friends on instagram dms. So it has to be fully automated without my control, anybody whos got an idea can help me do this? Thanks ๐ฉต
youd have to script it with the api
Is there a bot or smth that can control instagram?
And takes the response to the chats from chatgpt?
did the cap of gpts and the chatgpt unified? was 25 to gpts and 40 for chatgpt, but now both limit cap reach together
Hey can you log in from GPTs to zapier actions?
The set actions are not working, when I issue the command log in to zapier
the oAuth2 login window opens I click login and I'm back to ChatGPT4 + Plugin zapier and the message that I failed to log in. I've tried about 10x, I'll probably use up my Chat query limit in a while.
Why is gpt voice so low quality now. Its always glitching out
here the same
Iโm sorry it isnโt working for you, either, but I am glad itโs not just me.
image it being trained more on text and images about the output quality rn.
I am sure they are dedication some of the training to text. I wonder if DALL-E 4 will be able to generate text more reliable
well for me its not much the text
its more accuracy, and allowing style references.
it's frustrating, but you are not alone
You might want to contact support too
Hi, Today after a lot of tries, I can't send any prompt on any desktop and any browser suddenly. If I set chatgpt3.5, it works. but when I choose GPT 4 and send messages nothing happen. But it works on mobile phones, both app and browser. can somebody help me out?
Have you tried using different browser?
I did. Waiting reply
I did. I used the latest versions of Firefox, Chrome and Edge on 3 different desktops. changed my IP for other countries, changed network, changed the settings of my browsers to less restricted, disabled my anti virus, made open ai as exception on every adblocker and it only works on 3.5 not 4.
when I go to developer part of the browser and send any prompt, it blocks this: events statsigapi net v1 rgstr
both option and post
do you have any idea?
Why is the GPT all of a sudden refusing to read the files it's been reading since I paid for the upgrade and telling me it doesn't have access?
Oh, and if I ask what files it has access to it just makes up a file name.
this is the error in my consul when I write a prompt and hit the send: CORS request did not succeed
it's also Arkose script failed to load
Hello! New here, I apologize if this has been answered recently. I just resubbed to chatgpt yesterday. I've never had any issues with it in my previous month and it worked wonderfully, but in the past 24 hours I can't get it to follow my instructions. It constantly comes back with "I'm sorry I can't fulfill this request" after almost every message. No other errors. On browser, desktop and mobile.
Well, judging from the latest messages in this chat I am most definitely not alone. ChatGPT 4 is completely broken for at least some subset of paying users. As far as I know thereโs been no acknowledgment of that by OpenAI yet. I wonder why itโs affecting some users almost totally and others not at all? Itโs not a browser, app, or network issue since everyoneโs on something different. ๐คทโโ๏ธ
Paying for plus and it's broken for me too but all I'm getting is it saying that "something seems to have gone wrong". I'm just trying to have it look at a 1 line error from unreal engine and explain it. One of the least complicated things I've wanted it to do in the last year.
hello guys somebody know why I can not anymore add my website to my builder profile even after verifying the domain?
It may be close to max capacity. It's good most of the time
The API might be more reliable if you need something more reliable.
I've been getting errors like that at least half the time lately when I try to have it review a screenshot. Usually screenshots of unreal blueprints for coding help.
I might have to look into that. Haven't used anything but the chatgpt app or using it in a web browser so far.
Yeah I've cancelled my sub. I'm not paying 30 dollars a month to frustrate myself and get stopped when I'm in the middle of something.
It's new technology, there are going to be problems. it would probably be better to describe it in text than trying to show it an image.
Probably, just not a fan of something that's worked many times before suddenly not working. Viable option for small errors but when it comes to troubleshooting blueprints having to describe everything can get pretty complex and pretty much makes this AI barely better than a search engine.
You have to realize they can't just make an image recognition model. They have to put in safeguards to make sure it isn't used for the wrong reasons (CAPTCHA solving for instance) and some of those might bleed over into the normal space like yours. Are you trying to make complex blueprints without understanding how the components are working?
That's going to only hurt you in the long run if you are.
Right now, AI is a great technology to help improve peoples productivity, not replace it.
No I'm an indie developer working on learning how it all works. Artist that's having to learn to code for full transparency. I'm putting the stuff together slowly and having the parts explained in extreme levels of detail as I go.
The screenshot part is more for if something isn't working I can have it review my blueprint and it helps with the troubleshooting process. Gives suggestions on things to try for fixing it.
That's good. The struggles with debugging like you are going through are also good because you learn more. As frustrating as it might be it's better if you find the problems than have something else find them for you.
Right now I'm actually just having some (what I think) tiny collision issue with a landscape I'm sculpting and the error that isn't actually causing a problem right now is annoying me so I figured I'd throw it into chat gpt to have it explain what it is so I can get it out of the way and avoid having the error pop up again later.
Just feels bad paying for something and still having to resort to google instead.
it happens this is new tech. It'll even give you deprecated results sometimes so verifying is always good. Right now it's a general, jack of all trades, resource. It's great for most things but will fall short from time to time.
Even the slightly old information it's given hasn't really been an issue. If it's at least working it puts me on the right track and cuts out part of a long session of digging through documentation. Not to mention sometimes things in said documentation aren't in easy to understand language so being able to have it phrase things more simply can help sometimes too.
hi
yes it is great at explaining things especially when the documentation doesn't do it well enough.
Has something changed recently with having chatgpt look at images? Just to test it I uploaded a small section of one of my blueprints and it immediately errors out.
But when I typed out the error I've been getting it was able to give a very detailed explanation.
@grave trellis yooo
try an image without text
chatgpt on toppp
Uploaded an anime character that DALL-E generated in one of my chats. Immediate error, no text in the image.
there is a problem on your end. You may want to contact their support. It is working for me
Fantastic. Midday break turns into more troubleshooting fml
you can try and email them. I think it's like support@openai.com or something like that
Does feel a bit better knowing I'm not paying for garbage though.
yes this is definitely cutting edge. I'm happy to pay for it. $20 is quite reasonable for what you get even as just a language model.
AFAIK they stopped to answer this email. and the valid ways to have support it is on help.openai.com
Absolutely. I tried learning unreal a few years ago and went back to focusing on 3D art instead of the technical stuff. This makes the learning process much more streamlined for just jumping straight into a project.
Oh you can open a thread on the bug-reports channel
Thanks
welcome, i hope they can resolve it
Thatโs the error I most frequently get. It appears to be anything to do with uploading something (an image, a pdf) and/or asking it to do something code-related. It can browse the web fine (I asked it to tell me something important that happened in the world today and it wrote about Navalnyโs death).
The โvisionโ function seems entirely broken. At least for some of us.
bugs are annoying today
Couple of quick questions though I'm poking around! (1) Best code generation copilot type tool if I wanted it to help me build a little chrome plugin? (2) what's the most affordable but decent static image generator company/model right now?
not really the question to ask in #chatgpt-discussions but rather in #off-topic i'd go for DALLE personally through the API
well that was interesting. I have a custom gpt with documents in it's knowledge. I asked it a question about something and shortly after it started looking into the knowledge it started to respond to me (wasn't finished). So it gives the appearance of changing it's decision or doing it to pretend it was looking through the knowledge and didn't actually intend to.
chatgpt and dalle3 
Question, for chaptgot 4, can any of those add-ons read subties from a video on a streeming service?
anybody else having previously generated results get kind of overwritten. I generated something mere minutes ago and the result disappeared leaving just the prompt I used and below that where the result should be is a result that I generated after the previous one
Iโve heard the ChatGPT is now upgraded sora feature, how can I use it?
Hello, Sora is not on ChatGPT at this time. OpenAI recently just announced that it exists, and that it's in very early testing with red teamers and some creators. There are currently no announced plans for a wider release.
Got that , I am so eager to see text to video feature, thanks so much
For the first time these last 2 weeks, I did not receive a single "I cannot fulfill that request". In my entire 400 prompt conversation.
I can't wait to get this thing with memory, It just forgets critical information given before. Very maddening
we have to wait for the roll-out that is happening, but I'm also eager to have that feature
