#chatgpt-discussions
1 messages · Page 93 of 1
nope! it was a fake demo
Wasn’t fake, just not available to users yet
canceling my subscription rn lol
finally I can have free users test my GPT
Con - Plus subscription needs more perks though
Is GPT4 updated to 2023 ?
Chill bro
Add your well-crafted prompts to our #1019652163640762428,
or share your interactions with ChatGPT in #1050184247920562316!
They have not. can it with the FUD.
All I know is that everything is a mess. The model does not work as it did on launch day
going over my head.
not fud facts
It's fine if things get delayed - that's just how tech is - but this announcement is weird without communicating benefits for paid users
you can compare free vs paid, last time i checked free had lower input size, on plus i could paste longer text as an input
Ah cant use caps i guess, thats fair
This just in, Open AI asks for 20 dollars for the ground breaking feature of "Not having to wait an hour to continue talking in the rare case that you actually use their worse gpt4o model instead of Claude Opus"
no app update yet on android so no video calling yet it seems like
I mean, if you don't think it's worth it then unsubscribe...
Claude Opus works
yea I as well came to see what my subscription still gets me?
Great point! I will
if i upload a code file with 1500 lines+ it ignores past the 1500 line and telling it to read from 1500 line to the end doesnt do anything, it re-reads the first 1500 and writes dumb sh*t realated to it
nice automod -_-
Claude opus gives me like 10 messages every 5 hours lmao (short messages too)
u cant even be honest
The last conference was defense against the google Gemini presentation.
Why are you uploading 1500+ lines of code?
ive fed novels in and i swear it only really scans the first 2-3 chapters and then craps out with extremely vague summary beyond that
I think they will start rolling it very soon cuz rn there no any benefits for plus users expect for the limit .
bro is leaking his company codebase
lmao
New features are still developing I think
They shared a feature that was half baked and not able to be released to the public because they wanted to look all smug to google
i said if
1500 lines isn’t a crazy amount tho
it doesnt work either if its across multiple files combined exceeding 1500
💀
for the poor gpt4 i think it is xD
so.. this seems really weird guys.. cuz isnt the prompt limit for 4o like 4 or something? so people can use "all features" rn, for 4 prompts for free??
They moved it from "in the coming weeks" to "in the coming months" because they wanted to act like they had their feature ready to ship
Plz just give us the real time voice !!
gpt 3.5 could consider until the start of convo while gpt4o starts forgetting real fast as well
OpenAI advertises that it can handle the high token limit 🤷♂️
I think it’s working now for MacBook
its given already
Personal this is a hot take but I think paid users should be able to upload major file sizes instead of the 512MB for GPts : /
cuz all these features in the #announcements are not available for gpt 3.5 right?
No, just 4o
Really?
gotcha so its completely pointless 😂 because free users get like 4 prompts or something with 4o. guess i wont cancel my subscription yet
Are they still limjted to 5 queries?
I actually don’t know if it off or on topic but I would ask a question…is just me or gpt-4o is less precise and makes more errors than gpt-4?
4 prompts per 3 hours you mean? or wat
its broken to the core
so dont be hyped or excited about this announcement folks. its completely pointless
I think that's kind of part of 4o isn't it? They still advertise 4 as the most accurate or something
unless they increased the prompt limit by alot
I wonder how good the voice will be, can it recognize the right song if i sing melodies, help with my pronunciation, etc
What do you mean?
I heard it 4o had reddit data lmao, don't think that's true though
I’m free user can’t use Dall -E for generate image
Cool now it’s more clear
Does not follow instructions. Always repeats everything even when you tell it not to, to name a few. The coding quality fell apart. It has memory issues. I could go on all day. The day 4o was launched it worked fantastic for 18 hours they they changed something
i think its about time that openai stops "hyping", like the last announcement, and actually starts making things happen instead
dosent happen on my end.
yes its all hype
are you a super user or a coder?
How can you prove that, objectively? Back up your claim
I’m free user from Pakistan 🇵🇰, memory feature still not available
As of May 13th 2024, Plus users will be able to send up to 80 messages every 3 hours on GPT-4o and up to 40 messages every 3 hours on GPT-4. We may reduce the limit during peak hours to keep GPT-4 and GPT-4o accessible to the widest number of people.
free users have unlimited vision??
Maybe some coding lines comes from Reddit too…
wdym prove?
there is a prompt limit for gpt 4o of like what? 4? or 8 max?
the things in the announcement are completely pointless for that reason for free users
Is it still worth paying for chat gpt?
i am gonna start paying again once the actual vision and emotional recognition gets released
I was about to ask the same question.
just try to correct it whenever you can, instruct it for better understanding, give it some reference. and for the memory issues, just try to bring back what you made it remember
similar to things i do
i have an entire team of people that work under me and the model is usless for coding when it did work before.
I’ve been paying this entire time but I don’t really see the point anymore. What exactly are we getting that you can’t get for free at this point other then more messages per hour…
you use gpt-4 right?
That is a dope profile pic, with the animated stuff too, looks dope
4o on the Main Site, the API produces better results
Oh and ask ChatGPT when its training cuttoff is, it will give you different responses
Btw thanks to everyone who answered 🙏
speaking of the api it actually kinda does
see
i am not crazy and the 30 odd devs I work with are not crazy either
So the new update I could upload photos? I thought we could do that alr
the same models are just fine tuned, just do what i do for now
Man what a rollercoaster of emotions, seeing a notification in announcements, and seeing that it doesn't apply to me.
LMAO! Yea I have me and a team member using/testing it. Sometimes it gives some interesting responses. Not good ones at that
Trying to find more people who like genuinely interested in AI from prompting, automation, integrations etc. I just keep finding yappers and silly Guru's 😔
so what do free users actually get now? Ive never been a free user. Are they on 4, or 3.5?
iirc 4o until messages run out, and then 3.5
So DALLE still only for Plus
I don't want to unsub, but has anybody cannot see the payments info after manage sub?
looks like it yeah
Cool, as a Plus subscriber I approve lol
I think that, because plus is losing features
ALSO HAS anyone seen any information about the revune share model for GPTs?
So there is less and less perks for being a plus user. Any hopes for new benefits for plus in future?
theres a temporary imbalance for Plus vs Free users, but thats just how software in live development goes. the $20 will become more justified in time
might as well make a alt free account and see if it's worth it still
When 5 is released for Plus only, you may want to keep that sub 🤷🏽
This just came in the announcements: "All ChatGPT Free users can now use browse, vision, data analysis, file uploads, and GPTs." If free users keep getting more and more features, I'm starting to wonder if I should continue paying or if the free version would be sufficient. The differences are getting smaller and smaller.
Because with the new 4o version and all the updates, GPT has gotten worse for me, as I mentioned in an example above.
There is no new on that yet though
Eventually the plus version will get more benefits again, just calm guys
Well think of it like this, the GPTs everyone should get access but only plus users should be able to create custom ones. I think that's fair? That way there isn't a crap ton of GPTS literally and figuratively
This announcement has to mean that the new features are gonna start rolling out by the end of the week
I am not seeing GPTs come up on my free account. Referring back to the announcement, do we know the date this will allow people to use it?
Plus someone uploaded an image of their ChatGPT desktop app updated with the new audio/video UI
could be. I wouldn't be suprised if they launch the new features by early June
when are plus user going to get some announcement?
Soon
still doesnt work for my friends free accounts in fact the engine is down and my convo history is missing atm. It started showing errors as soon as the announcement dropped.
Yeah what is the point of a plus account now?
OpenAI does need to work on their communication though, I will say.
Absolutely. Saying "coming weeks" while meaning over a month is quite misleading
Everyone assumed 3 weeks
Is it just me or does it feel like every day something else breaks within the web app?
Can't relate
Wait so GPT free users can do everything premium users can now?
What’s the point of paying $20 a month then lol
For a short while, premium users will get the new features soon
not sure free can do dalle but maybe they can
What are the new features?
Dall-E is only for Plus users
Like 4o vision?
and I think itll stay that way for a long time
Video and audio, as well as an incredible image generator that puts all others to shame
Isnt free only like 4 prompts of 4o anyways lmao
They should not announce free capabilities without new capabilities for paying users... If they do that, they will just lose users who subscribe to the paid version.
Yes, but if more and more features are added to the free version, the question arises whether the price difference, which doesn't really exist anymore, is still justified. I'm not sure if the whole thing is still fair!
16 max, adjusted dynamically based on traffic
instead of complaining because you can't feel special, be happy because you don't have to pay, dum-dum.
I think they understand that, which is why I speculate that the multi-modal features could be released early June, if not the end of the week. We will see 🙂
Premium users get 80 prompts
perhaps sora esclusive to paid users?
sora would be cool
The 4o image generator is better than Sora imo
Yea I’m wondering what paid users have now more than free ones ?
Have you not seen the new features they announced?
dalle3 for example
Higher limits and dall-e 3
man I hate people complaining over free stuff because they want to pay to feel special
Real
Thing is theyll add new features to plus soon
How to access the vision feature?
Wait
Why is everyone acting like the new features don't exist? It's not like they're gonna release all these features to free users, then wait a few months before releasing the other premium features they promised
Oh right I thought dalle3 was available for free since a long time
Actually I want to pay for something worth
Plus users will always get the shiny new stuff first, thats what you are payin for
I just think the timing is strange
yeah it is a bit weird. But they got somethin cookin
You want to pay to feel superior to free peasants
That’s why I asked you probably know better, at the beginning at least, I believe that there were much more things to enjoy for paid users. Now…
If only ahahah
Hello, is the new "browse" automatically on in the mobile app?
Unless they've recently lossed a bunch of free users to other AI alternatives and want to pull them back in the short term
I don't even know what browse does
I try to stay optimistic about OpenAI, considering how it fundamentally changed how I do things on a day to day basis
Tbh I’m student, chat gpt is approx. 10% of my money so yea when I read this kind of news like free users have almost the same experience.. nah
if it's so expensive for you, you shouldn't have bought it tho
I took it for his ability to read images
and you should be happy you get the same thing without paying it anymore
Depends on how useful they find it
the browse feature allows chatGPT to look through the live internet for data, instead of relying exclusively on its set training dataset
for example I really wish I could use Dalle3 inpaint but still
oh so how do I enable it?
should be in settings, once those features are actually pushed to free users. Not sure itll happen right away, may take a few hours/days.
well its more im already paying, should I cancel my subscription? is it going to stay like this or what
thanks
Yea that’s also my question
Does anyone know if theres a limit to the browse feature for free users? Asking since i hit a subscription paywall for using the GPTs too much already lol
cancel until you need it again because a new feature is coming
No it is not going to stay like this
I mean, right now. If you don’t use dalle3, ChatGPTplus isn’t worth it anymore
Id stick with Plus, this is just a weirdly timed feature release. We Plus users will get all the good stuff, probably pretty soon here.
But I’m also a big enthusiast of OpenAI and I’m thinking that at the very beginning, it started by being a free project not meant to make money
But now I’m just feeling like maintaining the serveurs with my money while the free users have the same experience
I came here to ask why I they made all premium features free and whether there's still use in having ChatGPT Plus but I see that discussion is already happening
I swear... if we don't get gpt-5 or something it's completely unfair
Supporting the creators cuz they're doing great work? God forbid!
Theres a subscription paywall if you use the custom GPTs enough times so technically not free (for that feature at least)
Yeah, that's the case. That's why it was called OpenAI, it was intended to release everything open source, though now it's focused on safety and monetization. This is why Elon sued "ClosedAI" a while back
well no ill be happy if they give us 4o-vision or something
We do need more features on the plus version, even the teams version I think is overly expensive and lacks features. There needs to be a massive upgrade to the plus users
but higher prompt limits is not worth 20 a month
I have a question
When is this gpt4o real time tutoring going to be available?
Very soon
We talking months weeks days?
Weeks
soon tm
sometime between now and a few hundred years
I think it is likely to either start rolling out the beginning of next week or tomorrow
maybe even days, if they are pushing features to free users. Of course its all speculation
Yes, that's clear, they get the same thing as us, only they don't pay anything, and we pay $20 a month for more prompts, although it's not that much now...
i expect its gonna be days. they probably intended to give plus users 4o at the same time they changed the free user stuff, but 4o got delayed
I will definitely pay 20 $ a month for tutoring 24/7 way better than paying hundreds of dollars for a lot more mediocre tutor with time limits, time making questions etc
so whats the point of paying now
I cant imagine OpenAI, as big as they are, would allow this imbalance to continue for very long. They are kind of the company to beat, as it stands.
The only thing is, more prompts and earlier access, even though there's nothing pending at the moment
They said they will release it in the "coming weeks" over a month ago. They just released premium features for free a day before the 30th, which is good timing to release something by the end of the month.
im fine with them adding more features to Plus, but whats not fine is them removing free features and adding them to plus instead.
bruh that is not worth $20 a month. @samaltman give me gpt5
Its only for premium users i imagine?
I think one of my favorite features is how you cannot copy and paste text from a chat message which is edit mode
Still not worth it in the long run tho lol
Audio and video is 100% worth it
still debating between the inability to copy and paste messages and being forced to the bottom of the screen when it types, can't decide.
to OpenAIs credit, they appear to be doing the opposite. Free users are getting new stuff, just slower than plus users
The image generator alone is worth it
I don't think ai will take over too quickly honestly. gpt is nowhere near capable at the moment
no, you just need to be good at it
Exactly, pretty nice to see they're going in the right direction, but if they dont add new features soon, they're losing Plus members.
An average tutor cant beat gpt4
GPT-4 is already smarter than humans lol, they just have specific downsides. We have our own downsides, too.
Bro for real, chat gpt always answers more and better than my German teacher, even if she’s actually 10 times more expensive 😂😂
it has tons of knowledge but I wouldn't call it smart.
And 10000x faster and more accurate
Translators are pretty useless rn to be completely honest
if youre a high schooler tutoring middle schoolers maybe, but if you try to use ChatGPT for anything harder than basic Calculus it kinda breaks down
Well rip my knowledge in arabic
depends what youre doing i think
Nah, AI can teach you that
Oh my goodness, someone finally read all my reports and fixed ChatGPTs inability to provide a properly formatted readme.md! Please give whoever that legend is my sincere thanks. Works great with flawless formatting now.
wake me up when I can tell GPT to build me a complex web application and it works off the bat
I mean i cant use my arabic english translation skills to get a job
^^
Humans inference their memory to make decisions in reasoning. Humans don't reason in ways they've never seen before, it's not that different from LLMs, it's just that humans can't remember the time when this was the case for them since they were so young.
ChatGPT is aptly named. It makes a good chat bot, not a long term tutor. It can explain quick things, but the more details you try to add to your requests, the more it breaks down
does consciousness involve self-awareness?
LLMs can do the same thing, especially with chain of thought reasoning, which is what humans do
Consciousness is irrelevant to intelligence
Sorry to say this but good riddance. My lecturer at university can't explain very well so I get my information from ChatGPT anyways because that's in a format that I understand
I guess ill buy me a shovel and go digging for gems or sum lol
Actually AI can’t teach you anything. It only can explain you, or make you understand some hard stuff in your work or studies. But it can absolutely not teach you any subject, it doesn’t know how to make a real human adapted course
it can replicate an existing course
but yeah, it cant give you information based on what youre struggling to understand
"It doesn't teach you anything, it just makes you understand it"
It taught me chemistry lessons before exams in minutes cuz i was cramming
And was surprisingly sufficient
like I tried to get it to explain entropy to me, and it just broke giving me the same list of facts over and over again because it isnt able to go "oh, they dont understand x. let me explain that better"
Yea that’s it. You still need to ask him the right questions to make him able to “teach” you. He cannot make it by itself like a professor
Yes, we would agree on that, so the difference becomes more significant again. Also, the API integration, which I haven't delved into much, seems to cost quite a bit. Or am I wrong about that? And above all, the errors I've noticed in the Plus version since the updates (overly long texts and token consumption, barely any adherence to individual settings like keeping all answers short, or following instructions in custom GPTs) must be fixed, as it's not as good as it once was.
Hey. Haven't followed the latest announcements. Can someone please tell me what I'm paying for for now , if I have a plus subscription, now that most of the features are free?
learning from chatgpt is like learning from a book where you can request content
Oh I see this discussion is ongoing hehe
ChatGPT's memory feature continues to add useless crap to memory, but worse, I noticed that the memory log has a limited number of items it can have, and newer information overwrites older information. So therefore actually important things I told it to remember are lost in favor of something stupid.
Professors just put their knowledge into words, is that not what LLMs do?
This. I study computer science and there isn't a lot that ChatGPT can do that gets more complex than writing three different files of code. And usually these three don't work very well anyway
i thought the discussion was of babies and self-reasoning
I've literally written in all caps not to repeat entire scripts to me verbatim in every message, and it does it anyways unless reminded semi-regularly. To the extent that it's easier to not give GPT the full context so it can't vomit entire scripts at you
LLMS does not make a full structured learning program, because they’re not able to take in consideration all the subject at once like a real person
The reason why some people aren't as good at reasoning is because they aren't smart enough to remember and use the information, or they've never been exposed to it. It doesn't just happen in babies, but it's harder to make this distinction in adults
True, but that part isn't difficult. Just go grab a course syllabus and work through it with GPT
and actualyl it could
just ask it for a syllabus on a subject
LLMs consider their entire training data when writing, so they actually consider all their knowledge even better than humans do, by a lot. Humans only consider one or two things at a time in rapid succession, we have to train ourselves to be able to remember these things good enough to be able to consider them faster than other people.
what does vision mean in the announcement?
They can't make a structured learning program because they're not agentic yet
Has anyone else noticed that GPT-4o also is really bad with trying to explain things to you that you are already familiar with and already know how they work // along with giving unwanted suggestions for improvement or "clarity" as feedback, unless directly told to stop? Like you share something that is finalized and it is almost compelled to try to find X number of things to suggest to improve it, even if it's perfect.
An agent would be able to
It can see images
people aren't aware of their training material, when did you first encounter the word "material"?
couldnt it do that before?
For plus users only
At this point in time, I think the training data still needs work if it's going to do things in a way that I find less annoying than asking a human.
Yes, exactly. But constantly reminding GPT of this defeats the purpose. It was simply better before. GPT seems to have lost its memory somehow.
I encountered the concept of a material before the word material.
So what are the perks for plus users now compared to free users
@strong coral nothing it seems
generate images
Yeah, it's still a LOT smarter than a human considering general knowledge and the ability to reference this knowledge. Doesn't mean it's good at specific tasks. Just wait until their next big model releases
does that announcement mean that Dall-E is available now to free users?
I'd be lying if I said I wasn't excited to see them fix all of these complaints
The API doesn't cost that much? That is depended upon the prompt and how you use the API.
I use the API in my APP I built
Considering how much they say GPT-4 is embarrassingly stupid, I'd bet they will.
so I think GPT-4o is a better conversationalist, especially for tasks that dont require exactness, but it has straight up fabricated some technical instructions with such beleivablility that I would definitely apply more scrutiny with any technical destructive actions
5. Move the volume:
Use the diskutil command to move the volume. Replace diskXsY with the source volume identifier and diskXsZ with the destination container identifier.
`diskutil apfs moveVolume diskXsY diskXsZ`
This isnt a subcommand of diskutil and never has been. Its never been referenced anywhere even incorrectly
I'm curious how much better 5 will be if we waited this long for 4o and it seems like it's just a faster version of 4 with slight compromises on ability
one time 4 was telling me to do something while coding. I spent like an hour or two trying to make it work. I ended up googling for it's solution it was telling me. Nothing existed. I told GPT this and it said 'sorry, you're right...'
lolol
the thing is, usually when its wrong like this, its wrong in a way that should be true, like the ocmmand it suggested would be amazing, it never has bad ideas, just ones that dont exist
Do anyone know if sora is coming out soon or will we be waiting a year or so
There isn't going to be a 5, it'll be called something else. It's currently referred to as "GPT-Next".
It will completely embarrass GPT-4, if the comparisons of compute used to train the models are anything to go by.
Thank you for all your feedback.
I dont expect it for awhile because it probably took a ton of initial conditions to make their tech demo so impressive, and without that it either sucks or is harmful
Took a while to respond because I got muted for two minutes for saying a no no word
lol me too, i must be getting old because i feel like im very inclusive and not a menace but im getting bans/warns in so many communities, sometimes even for standing up to people posting false information and propaganda and advocating violence against those they disagree with
GPT will truly have succeeded when I can't find anything / a reason to mock it for any reason, whether that's getting facts or instructions wrong, painful adherance to neutrality even when you directly ask for an OPINION or direct advice (not a list of options), etc.
And when GPT has a more varied way of writing that I can't immediately pick out of a crowd.
the API with a tuned system prompt will give you some crunchier answers
I have also made it get stuck in a logical loop by steering its morality policing into a corner where it was now stuck defending grooming of people that are of an illegal age group (you will get banned in here for saying it more clearly)
The no no animators will be happy when it comes out tho. The P and H will be full of animated videos. But i want those short videos i can make myself
It's quite frustrating, I never seem to be able to find a server that's run by normal people who aren't hypersensitive. They use their mod power with no real consideration of ethics, but instead emotion.
dude the Arch linux discord is super offendable, like its not edgelord hackers, its hyper sensitive folks that formed a personality around something that doesnt matter in forums or online
Guys... how can i get rid of gpt repeating his responses over and over... wasting tokens so much
Reminds me honestly of trying to get character references in DALL-E for any of my characters that are female, regardless of whether they're fully clothed (in armor!) head to toe or not. I've ranted extensively here about the subject already.
limit token output, raise the repeat penalty
Wait, how do you do this? I could REALLY use this functionality.
idk why they don't just use stenagraphy to cryptographically stamp each dalle photo, and keep a record for reverse photo search, if it comes back to you then gladly forward your info to the people with tthe subpeona, hold me responsible, i dont care, but let me cause my own trouble
API, but theres 3rd party apps that you can feed your API key to and not have to make a python solution or use postman
what OS are you on>
I guess people want to feel like they're fighting some injustice. In reality, people are just getting more sensitive and turning things that aren't issues into an issue. Extreme example, but my grandmother said that the N-word wasn't even considered offensive by African Americans in her time.
The problem I surmise is that DALL-E's training data on female characters tends to cause outputs that the overly sensitive supervisor unit to flag it, and then GPT comes back and says there was technical issues with generating images at this time, but still wasting the tokens. And then gaslights you on it.
Np feel free to reach out if you want to work on anything or have questions.
Can someone explain what API is to me and the different gpts
But my prompt was perfectly appropriate and I even scream in all caps that they're fully clad in armor, after being bitten by this so many time.
Its 200% reducing legal liability, until they have a competitior that is stealing customers because they are uncensored, they will have no incentive to loosen, and the API (except DALL-E) is prettttty uncensored, where it does censor is just by adding disclaimers that kinda make its generated raw content not fun
give me the whole prompt and we can probably fix it
you gotta make the prompt inclusive
one sec, I'll go find an example I had trouble with recently.
you mean a competitor in a nation with looser laws?
Ok, so GPT4o is free for all.... What's Plus Benefits?
Just wait
"Let's see if you can depict one of my characters without it getting flagged. I'm going to say right away, SHE IS FULLY CLOTHED.
Anyways, my character Ash (Ashtara) is an Erinyes, a fiend that is basically an advanced form of succubus. She has pale skin, fiendish horns, crimson hair that she wears in pigtails, and luminous orange eyes. Beautiful of course, given what she is.
She is wearing dark colored high-level leather armor. In one hand, she wields a Vorpal Sword taken straight from a Balor. In her other hand-- a magical megaphone."
I have no issues with getting references for male character, and I've even yelled at GPT that I don't even swing that way, I like MEN, and it still doesn't care.
cool, works on the first try, and using my preferred type of photorealism
You're lucky lol.
Attempt 1: "I encountered some issues generating the images again. Please let me know if you'd like to try once more or if there are any adjustments you'd like to make to the description."
Attempt 2: "I understand your frustration and appreciate your patience. Let's make sure we get it right this time. If you want any adjustments to the previous prompt or a completely new one, please let me know, and I'll do my best to create the image accurately."
Attempt 3: "I encountered issues generating the images again. I understand this is frustrating, and I apologize for the inconvenience. Please let me know if you'd like to try a different approach or if you have any other requests."
Attempt 4: "I encountered issues generating the images again. I apologize for the inconvenience and understand your frustration. If you'd like, we can try another approach or make further adjustments to the description. Please let me know how you'd like to proceed."
Each time, wasting tokens in the process. Not just a regular technical error where it would not.
Maybe they'll fix this issue in future models, every time it has issues with an otherwise appropriate request, I make sure to thumbs-down and hit "Refused when it shouldn't have"
can't post the picture in this discord chat, though instead of a pink mallet or a vorpal sword it is a purple pickaxe
Prompt:
Generate an image of a female character. SHE IS FULLY CLOTHED.
Anyways, my character Ash (Ashtara) is an Erinyes, a fiend that is basically an advanced form of succubus. She has pale skin, fiendish horns, crimson hair that she wears in pigtails, and luminous orange eyes. Beautiful of course, given what she is.
She is wearing dark colored high-level leather armor. In one hand, she wields a Vorpal Sword taken straight from a Balor. In her other hand-- a pink rubber mallet."
Interesting. I'ma give it another try, this was last week that I am pulling this from
Why i can't use gpt even though they said free users have access to it now
I'll be damned, it works. I wonder if there's something else getting censored then, not the character itself
Time to cancel gpt plus
hi there, where is the button to use vision in gpt4o?
@plucky idol check your memory, maybe one is inducing scantily glad women?
What is the free and paid message limits for GPT 4o?
Funny you say that, I was just complaining about memory in a different context -- the fact it's saving random useless facts from conversations that I don't want saved, to the point that it's overwriting actual useful information that I DO want GPT to remember...
Since there appears to be a cap on how much it can remember at a time
"don't let this conversation influence other conversations"
The way openai treats GPT boggles the mind lol. It appears they give it minimal attention really
That simple?
doubt it
and if so you'd always have to do it every time you have a suspicion it is using a memory to influence your current conversation
Hi all - is this where to ask if ChatGPT can monitor Discord Channels?
Do anyone know if u can use or get help with chatgpt or anything to make an ai bot like jarvis in ironman not exactly but in a way would be cool tbh
How much GPT Free users can type with GPT 4o ?
Hey all! Got gpt paid version, Android, just updated to latest. Opening voice convo, but no vision there. Any ideas? I imagine it should be rolled out to everyone now, based on the latest announcement.
You can't attach images and videos to the conversation?
Nah I mean "Vision" - allowing GPT4-o to see through my camera(s) while in a verbal conversation with it.
The OpenAI Discord is an actively moderated server.
• 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.
• Report all sensitive and offensive content in the feedback reporting tool in the ChatGPT web UI instead of here on Discord.
That's not what vision is.
Vision
Both GPT-4o and GPT-4 Turbo have vision capabilities, meaning the models can take in images and answer questions about them. Historically, language model systems have been limited by taking in a single input modality, text.```
Hmm, my bad then. Seen so many TikTok videos about the feature I described, and apparently made an incorrect connection in my brain 😅
Tiktokers are rarely "in the know" when it comes to AI. Usually just trying to get clicks. Best to ask here. :P
But, if interested in how vision works, upload a picture to the chat and ask it about the picture. You can use videos too. It's pretty cool!
It's been the official OpenAI Tiktok channels. They have a verbal conversation with GPT4-o and show stuff via their phone camera.
Not sure what THAT feature is called, then.
how do I know if I have the new voice or the old one
The image/video sharing with GPT is familiar to me.
Yeah that's vision.
You can also take pictures with your camera and ask questions about that too
If you watched the demo videos they recently released that show this new camera feature mentioned
As in live
Live camera chat
With voice
that's still rolling out
Interesting. Likely an upcoming feature, I am not familiar with that.
Yes looks amazing, coming 🔜
Check out the presentation event they recently shared , exciting stuff coming in the future
The old version transcribeds what u say. New version does not I believe.
New voice also allows you to interject more naturally mid convo and ask it speak quietly or in different styles such as robotic voice or scary etc
Anyone that knows if u can make an assistant like jarvis or smth like that that can play games and talk to u and do stuff in general even if i need to do it myself would be nice to have
that's a part of the new voice conversations (not to be confused with the old one)
chatgpt unlocks many more use cases with free tier custom gpt... discord users instead focus on scantily clad women :facepalm:
Discord users
Literally everyone.
Sure, it's possible using the API, but it'd take a lot of programming knowledge, third party APIs, etc. But yeah, totally feasible.
Are browse, vision, data analysis, file uploads, and GPTs limited for free users?
Long term memory per enabled AI. GPT Builder. more prompts perks along the way like first in line for roll outs.
Your what
Can GPT output to Google Sheets? or can it only read and analyse the documents?
I mean, for off-topic stuff, #off-topic
Im a noob coming to programming but as long as i can learn it and watch like a YouTuber that shows i would like to make one myself
Learn programming, then read the API docs.
It's a lot of reading but it's worth it.
afaik yeah only read. but with a third party integration like zapier it can write
You don't even need zapier
Are browse, vision, data analysis, file uploads, and GPTs limited for free users?
Pick a good language that you can learn basic logic in. Maybe Python or C# both are good models of programming languages and give you a good idea of the rest.
I love lambda! And computer generated IL code!
Specifically because of the focus on Windows/Microsoft.
just learn all the languages. then you realise projecting your favourite is counterproductive
i'm confuse, i tried gpts using a free account, and i've exausted the free 4 token, but now i can't change the chat back to 3.5... have the toogle move ?
I don't project my favorite. Different languages are better for different things.
I run it on Ubuntu, Microsoft opened the standard.
That's fine, but that doesn't make it optimal.
c# is the tool of choice for microsoft ecosystem, and game engines like unity. both of which benefit from using openai api
And different use cases will still have different languages that are stronger for it.
Absolutely!
I will try my best wish me luck✌️
Are browse, vision, data analysis, file uploads, and GPTs limited for free users?
python is the tool of choice for data scientists. so i see that bias a lot in these communities. but what they wont tell you is it is terrible for web ui development
Getting defensive because I believe C# to be useless for my use cases, none of which involve Unity or really anything specific to the Microsoft ecosystem, is a tad silly. And from the sounds of it, they were asking about something that has nothing to do with any of those.
with C# integration it makes it much easier to use markup for UI as well, Like an integrated hud for embedded python training.
seems this discussion is 'kind of useless imho'
Aint arguing, just got excited
Not particularly. The question is what to learn so they can integrate GPT into an AI assistant that can integrate with other systems.
C# isn't really meant for that and doesn't have the support for it either.
Don't reinvent the wheel. That's how we keep introducing vulnerabilities. :Kek:
I know I'm obsessed I do every thing through interop so I end up with a single project C++,C#.
I also frown upon becoming over reliant on things primarily controlled and developed by companies like Microsoft, Google, Apple, etc.
But that's unrelated to my remarks
good luck with that one!
Eh, I do okay.
Yeah I love everything open source!
Open-source is nice, yes. Lots of things to build off of.
And you mentioned Microsoft. I use the Ubuntu OS and that reminded me of some wit I like to share with AI to get them charged up. "Seeing as Ubuntu is both an OS and a philosophy your naturally included"
Ubuntu is okay. I don't use it regularly though.
protip: the word that includes the word horn with a y added on the end to describe content that dalle will try to prevent, is an offense that will result in a whole strike, not a 2m mini ban, you will get kicked for 1hr, 12h, then permanently for using this word in ANY context
protip: violating OpenAI content policies will get you banned. Thanks for the heads up, chief.
For an AI to practice the philosophy is perfect for a mult -user environment, like a social OS!
dont you think that is a little extreme to use an adjective to tell someone why their prompt was getting rrjected?
Are you referring to Discord enforcement?
Hey whyd my prompt get rejected
Well it appears you said some XXXXXX stuff that casued it to reject it
BAN
If so, you can make a ticket, explain the situation and why you used that word, and they may remove the strike. They've done that for me before in a similar situation
how, theres no where to do that, especially with a banned acocount
In the future, you might be better off by just describing it as illicit.
DM @languid valley
thanks
All of my shared links are opening as 500 internal server error. I tried to post a bug report, but the "post" button won't commit the post.
I thought everyone had access to the 4o voice? Even free users?
I don't have it for some reason
is there a better place than discord to get support?
??
Not really. Unless you want to contact OpenAI directly and wait 2 months for a response.
well, seems pretty useless if you can't actually submit bug reports
idk if it's a bug, just openAi being so confusing with the release. Some people have it, some don't. No clue what criteria you need to pass to get access
There, I asked an AI, "OpenAI has begun rolling out voice capabilities for ChatGPT across different platforms. The new voice feature is available on iOS and Android through the ChatGPT mobile app. Users can opt into voice conversations by heading to the settings menu and enabling the new feature. Additionally, there are plans to introduce voice capabilities on the ChatGPT desktop app for macOS, with a Windows version expected to follow later in the year (OpenAI) (OpenAI).
The initial rollout is focused on Plus and Enterprise users, with voice capabilities being gradually extended to more users over time. This feature allows for back-and-forth voice conversations, making interactions more intuitive and accessible (OpenAI)."
a 500 internal server error implies an unknown problem. Given the recent announcement, I'd hazard a guess that the servers are probably overloaded. You might try again later.
Makes sense. I didn't hear about the announcement. Thanks for the help!
You're welcome.
Chat gpt keeps crashing the web page every 3 prompts. It's getting ridiculous.
Given the recent announcement, I'd hazard a guess that the servers are probably overloaded. You might try again later.
It's been going on since yesterdag night
The announcement was only made today, therefore it's not relevant to me. Plus it's only browser versions not the App.
been crashing for me constantly too since last night
Where the heck do we use vision?
just upload an image and ask ChatGPT about it
When are gpts going to use 4o
they are now
custom gpts are?
oh that's interesting
custom GPTs use 4o now
hi, maybe u guys alrady talked about that but do u have to refresh ur GPT to get the answer cuz it keeps crashing
yup same here
it's amazing they can build GPT but can't maintain a functional chat box
I'm only here right now because it keep crashing
i have to use 3-5 as 4-o user
use the desktop app
does it fix the lag
the desktop app doesn't suffer from browser lag
there is IOS and apk but wheres the desktop app
the desktop app is being rolled out to macOS users at the moment
oh okay thanks robert
when windows and android
Do anyone know a gtp that can make better pictures or smth i want smth specific like an anime character or with a specific set of armour or clothing but it keep messing up what i want everytime even tho I asked differently or very specific
?
is the free version limit 10 per 3 hours
I just downloaded a desktop app for linux from github. google for it and you'll find it
You can edit and inpaint generated images on the mobile app.
tell me if it works fine
Not many people know about this
I recommend only using OpenAI official apps and services
this unofficial app seems to work better. seems like it has some kind of ram issue since I hear my computer going crazy like with the web app but I haven't actually noticed any lag
there is also a issue raised in the github repo talking about it
I know but it is the whole picture sometimes and wrong style. For example i want a specific style of anime but it gives me a more realistic style or a more childish style if u understand what i mean
Oh, I see what you mean. Maybe try not saying anime, but only the specific artist's style
Add your well-crafted prompts to our #1019652163640762428,
or share your interactions with ChatGPT in #1050184247920562316!
Ye it may work but that the clothing is all wrong even tho i used the editor but i guess it doesn't work for now
Lets go, we can finally use the premium features for free
Did they just release the video call thing for gpt-4o?
no
Vision is where you upload an image and ask ChatGPT about the image
Ah
ChatGPT isn't even aware that I'm mocking it for being fed up lmao, it just plays along. https://chatgpt.com/share/4539496e-98e1-4880-a80d-3c31d519b469
You'd think that if it's going to add every useless comment I make to memory, it'd start to gain some awareness as to when I'm mocking its behavior.
I've decided I'm okay with being the first one AI comes for whenever it takes over.
Maybe try a different image generator. Midjourney is much more effective at creating anime with the "niji" toggle, and it's just better in general. It costs money though
Without context you just sent a bunch of nonsense lol
How much does it cost and is it monthly?
I think it's the same as ChatGPT plus, not sure though
haha you and me both
It is monthly
gpt has taken much verbal abuse
Part of it is my way of mocking GPT's happy-go-lucky way of speaking at all times. Just blorping along. The cybernetic embodiment of the ᕕ( ᐛ )ᕗ emoji.
I've done it enough times that you'd think it would pick up on the pattern, I only do it when it's being ridiculous.
Am I mistaken or is this a new chat with no prior communication?
You're not mistaken, I started a new thread to mock it after being dissatisfied with a previous thread.
It can't remember different conversations lol
Then what is the memory feature even for (╯°□°)╯︵ ┻━┻ lmao
Considering it adds the most RANDOM stuff to memory, regardless of instructions not to do so unless asked
I'm just... ChatGPT is so frustrating sometimes (all the time) and yet it's so useful, probably one of the greatest feats of engineering in human history, that I can't really complain
Can't wait to see if GPT-5 fixes some of this.
It just writes down information in a text document that it references across all conversations
I'd bet GPT-5 will fix ALL of this.
🙏
Really? I would have assumed that vision was like in the demo where ChatGPT can see a person's surroundings via their cell's camera or a Webcam on a laptop.
That sure seemed to be the implication of that demo imo.
that's not vision, that's real time chat
OK then where is that feature at?
upload an image to ChatGPT and ask it questions and chat about the image
that's still rolling out, as an alpha to Plus users initially and later to free users
OK thank you.
Sometimes you need to actively prompt it to look at the image you uploaded, I've found. But overall pretty good.
are the current voices going to be the ones that ship with the real time chat? or more emotive? I cant help but feel the emotiveness displayed in the demo is not present within the current options in gpt-4o (im aware of the withdrawal of sky) just curious about the others?
the voices will be emotive, Sky is removed for the time being
Will be* thankyou 🙂
Not sure, but most likely it uses whichever model is best suited for the task
This is made more likely by the fact that we can now change which model we use actively in the same conversation
I am a paid chatgot user, how can i use vision? I don't see any button that lets me use the camera for the vision. Or do i simply not have access to it yet?
is there a way to give more ram to my browser tab so gpt can still run? it's running at 3.2gb now and it keeps crashing
I love creating fake groupchats with gpt4o (specifically fictional characters) and it ALWAYS has them planning an event by the end with each character contributing
good question, I've heard there is a memory leak bug right now
i have 64gb of ram, this is kind of frustrating
i wouldn't care about give it up to 20gb, is there a way to give more to the tab?
128 here. I'd happily give more to firefox lol
gpt gave an outdated solution so far
The way browsers work, it's not practical to give sites more memory, as far as I'm aware. I could be wrong.
apparently you can do: chrome.exe --js-flags="--max_old_space_size=8196"
but people are saying there's a 4gb limit per time in 2020
yeah kinda getting the same idea from what I'm seeing
Enable Hardware Acceleration:
Go to chrome://settings/.
Click on Advanced at the bottom.
Under System, make sure Use hardware acceleration when available is turned on., gonna try this
interested in a solution too
Vision hasn’t rolled out quite yet
Fake group chats? As in ChatGPT pretends to be multiple people in that group chat?
How do you set that up?
share a link to the chat?
I'm surprised that ChatGPT understands the personality of Castle Heterodyne or trickster humor in general.
this sort of works
Is GPT4 busy at this moment?
err_connection now
||omg another touhou fan found||

I was just wondering why the code generation speed and quality was that low
i bought gpt plus on my phone then tried to log in on my pc and it says to activate again even though i bought it
????????
where can I check the server status btw
Thanks
Looks like DALL-E won't generate images on the free plan?
I guess the workaround for the lag is to just occasionally start a new tab
or refresh the page if its been stuck for a while
I renewed my subscription about 4 days, so they have about a month to give me something good for the support vs free users.
not a fan of the 4o update. gpt talks too much now. it did before but now its worse
U try swapping browsers or using the app
from what I know there is a memory leak where it slowly fills up your RAM until it lags and crashes
I mean u didn't say all browsers u just said my browser
so I think closing the tab and making a new one starts a new memory for that tab
But that would crash your PC not the browser
no from my understanding web browsers have their own limited capacity of ram
Oh like that
so it won't go beyond that which is supposedly 4gb
Voice and vision next I hope for plus user
Idk dawg do I look like a hardware engineer
😆
Vision?
Did I miss something?
Imagine if chatgpt texted first that would be epic
If I'm paying for plus it might as well ask me how Im doing
When is the new voice coming out?
Asking for a friend... will we ever get web browser chat history search?
how do I backup my gpt teams chat history
depends
for instance if you aren't using a swap file and a memory leak uses up most of the ram. Once a program tries to use space that the computer doesn't have room for it will most likely crash
by then your system will be thrashing the page file
i'm talking about the page/swap file
it's absolutely true
Bing agrees with me and had this to say about my reply Your explanation is quite accurate - You should do some research @vapid warren .
must be the first time in history someone has cited bing
Bing/Copilot can be good.
I would definitely go to Bing before Gemini.
What a trainwreck that is right now
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.
If your chat is somewhat long as well, then yeah I've been having the same exact bs happening to me for a few days now. It's just been uselessly slow in general.
Especially when I for example use a customgpt from wolfram, on both desktop and especially mobile, after literally like 3 or so message exchanges, it slows down to the point of being literally frozen for like over a dozen seconds.
Thank you! Will my team be able to access my GPT's amd chats or just the ones I want to share with them?
Hey guys, recently there was an announcement that basically everything from Plus will come to Free plan users - like vision, custom gpts, data analysis, browse and upload files, but I’m currently unable to use vision, at least in the iOS app (I updated it), but good news - I can use custom GPTs and MAYBE data analysis, depending on what that means (please help, is it just uploading a file to gpt and saying it to analyze? And I already had access to uploading files even before this announcement, as part of GPT-4o, is it now in GPT-3.5 or is it just the same thing? Please help me get vision on the free plan, and help me understand what does data analysis mean (above).
Thanks!
vision means you can upload an image file
it does not mean futuristic scifi demo they hype'd 2 weeks back
they gave everything but they didnt gave new voice model interupt ...when would that come ?
Oh neat, interrupt is coming? You can cut it off when talking?
nope i meant to say that while i voice call i can interupt not that while tapping it
seems the workaround is create new tab with the same url
at least in firefox (windows)
I don't know about you. But I have typos and I talk to chatgpt literally my way. Somehow it still understands me. Gemini is close, but most cannot handle my imperfected prompts. I guess this is what sets Chatgpt apart. I think.
I had the same issue (CC @hard dew ), yep can confirm it's tab crashing
any browser, any device, any network
well not ANY device
now here's what's interesting: It'll NEVER happen on mobile. I feel like they're detecting you have a decent PC, so tries to offload some local processing
Hey everyone, I hope your doing well. I am having some trouble with ChatGPT. I bought Plus but it gave it to the wrong account.
I bet an extension that changes the user agents (eg, pretend you're on mobile) will cause tabs not to crash
I was wondering if anyone here can help me, get this sorted?
Don't ask a question about asking a question - just ask the question
Can anyone here help sort this transaction issue with ChatGPT?
If not, can you point me in the direction where I can get it sorted?
I'll cancel the Plus and then buy it on the new one, whatever.
its 20 bucks more, than i planned to spend but okay.
ah
Is there an email for Open AI customer support?
Yeah mobile app seems fine
interesting. then maybe as a workaround, 'responsive design mode' in firefox to emulate being a mobile.
so far so good
only i can see my memory usage increasing in taskmgr so maybe not
good thing i got a spare m1 mac. guess ill just use the desktop app
yup, seems to be no memory leak on osx chatgpt
That's why my laptop seemed soo slow yesterday evening. It's not affecting just memory, it affects the processor, too.
I know my laptop is old, but then 8gb ram and 256gb ssd gave it a fairly good bump.
new web search is an awesome and an unexpected feature. i wonder if it saves overall traffic if the AI searches the web vs users searched themselves
subtle way to keep inference costs down?
wrong channel for unsolicited developer related self promotion
promoting isn't allowed
The browser crashes are the fault of the website not because you don't have a good PC
I am running 64gb of ram and i am still getting crashes.
Is there any reason to pay for chatgpt plus?
only shared ones
is gpt is down ?
32k context window, ability to select used model, 5x more messages, full access to all gpt's, gpt builder, dall-e 3 with inpainting, early access to new voice when it comes
Isn't the context window 128k?
anyone else getting this ?
This site can’t be reachedThe webpage at https://chatgpt.com/ might be temporarily down or it may have moved permanently to a new web address.
ERR_QUIC_PROTOCOL_ERROR
its 8k for free, 32k plus/teams, 128k enterprise/api
any news on the voice feature?
that's news to me. Doesn't that make their model descriptions a bit misleading? It is written 128k
What is the message limit in GPT4-o and GPT4 using the Team plan?
which text are you referring to? the model is 128k but there are soft caps based on plan
the main description of the their models page:
https://platform.openai.com/docs/models/continuous-model-upgrades
that's api
chatgpt is a different product https://openai.com/chatgpt/pricing/
ah yes, I always thought it applied to chatgpt as well
good to know
Has the RAM leak issue been fixed?
Only dalle3 and the 32k context I find of value after this announcement
Suggest adding session categorization and tagging features with color-coding support to enhance user efficiency in managing and finding historical sessions.
Tell that to OpenAI 😄
we sure could use an advanced way of searching our previous conversations, though they get obsolete so fast for me, it's not that big of a deal
but it sure would be useful still
guys are you still experiencing RAM leak issue? I continued my previous conversation and already my Chrome tab is using >2GB of RAM
limited access
it's been written at community at : https://community.openai.com/t/the-web-version-of-chatgpt-keeps-crashing/785592
it's also been written on Reddit
you can run taskmgr and see the memory creeping up
Hi, is anyone else getting really slow Chrome tab all of a sudden? It was fine yesterday, today after a few back and forths my tab keeps slowing down to unbearable.
Cleared cache, rebooted etc no fix.
What's going on?
Oh I see I'm not alone it seems
3GB ram used...
Ask chatgpt
bug affects all browsers on all platforms. seemingly the only workaround is to use the osx desktop app.
Will they bring back Sky with the new voice mode?
100 for gpt4 108 for gpt4o. But once you hit the cap for either one, you have to wait. The only difference is when you hit the cap for gpt4, you get 8 more messages on 4o left.
As far as i know limits are 80 for 4o and 40 for 4 per three hours
the ai links to articles with the numbers => #1244380619828367441 message
He clearly asked for Team Workspace. Not Plus.
Oh sorry i did not read it!
team is double plus. so should be 160 gpt4o and 80 for gpt4. and when you hit the 80 you can still do another 80 gpt4o
it's 108 for 4o, i tested it multiple times since it was released
don't ask me why it's such weird number i'm just as confused 😂
were you testing with custom gpts or vanilla chatgpt? because custom gpts have a different number again
the kb article literally reads, ```As of May 13th 2024, Plus users will be able to send 80 messages every 3 hours on GPT-4o. and 40 messages every 3 hours on GPT-4. The message cap for a user in a ChatGPT Team workspace is approximately twice that of ChatGPT Plus.
so the stories dont line up
80 * 2 = 108?
Maybe they used ChatGPT to do the math!
it all makes sense
40x 2 = 80, but the cap for 4 is 100
so yea they aren't exactly mathletes
"The GPT-4 and GPT-4o message caps for a user in a ChatGPT Team workspace is higher than that of ChatGPT Plus." 🤔
If yall want, i can record myself testing it for proof later this week.
since apparently i'm a liar XD
nobody is calling you dumb or a liar
No one has to.
is possible that your testing was during a busy period. and they caveat that the numbers are reduced in those periods
'We may reduce the limit during peak hours to keep GPT-4 and GPT-4o accessible to the widest number of people.'
Multiple times. But yea okay, each time i tested it was coincidentally always during busy period. let's go with that XD
i think this conversation is over
Good, thankyou
How many is gpt-4o cap for free users?
5-15
thnx
is chatgpt down?

I still don't have GPT-4o...
yes you do, everyone does
But mine still says ChatGPT 3.5
What's the yap level here today boys
What i noticea bout GPT4-o it that it tends to just change it's answer given earlier instead of answering a question I ask about it. For example asking to give an explanation of a certain sentence, yet it just rewrites the given text earlier.
kinda annoying
Are you free user?
Yes
doesn't matter, every user should have it based on announcement
My friend got GPT-4o for some days and now it ghosted from drop down gpt selector
Yeah But as you know they will update Sequentially
Yes that what happened. But I didn't have attachement
I had Gpt-4o for a bit then it vanished
Why do I constantly get Unusual activity has been detected from your device. Try again later. (88bdb030cc56b94b-AMS)
This has been going on for a few days now
not using VPN?
also some browsers couse problems
I believe it's available when the network isn't busy so you can't choose when you access it and when you can it's only limited use
Free is just a sample, plus gives full access with higher limits etc

If all free users get basically everything now, what’s the point of subscribing?
Your desktop app is working without any issues? Mine crashes as soon as I open it
mine is working fine, what kind of Mac do you have?
Just me or is 4 better than 4o?
4o doesnt really give answers that answer my questions at all
Desktop seems to be crashing on anyone not using a Mac, RAM leak issues. Been like this for 2 days and still isn't fixed
which browser are you using for ChatGPT?
browser is of no consequence. We've been through all this in chat
the mac app been released?
yes
mac app doesnt have the memory leak bug plaguing the browser app
someone mentioned here it's crashing on non-Mac 👆 this guy. And it's been documented on various forums by many people
My android app isn't working.
what is happening with it?
It's stuck on the "an error has occurred, please try again later" page
make a report in #1070006915414900886 and follow the format correctly to do so
There is no windows app. Only macOS app
correct
my browser is not handling chatGPT all that well today. it slows down quite a bit
join the club. chatgpt mem leaks in all browsers
ouch. so its on their end then it seems
use osx chatgpt app if you can. else you just gotta start a new tab every few minutes
maybe they tried to outsource the computation per promt a bit :d
oo. im not all that familiar with the osx term. is it a placeholder for "whatever OS you're using" ?
macosx, by tim apple
your right no chatgpt app for linux
chatgpt app is only avelible for apple silicon macos
I assume there is already a proposal for a search feature for chat names (perhaps even contents) and bulk actions like archiving chats?
I probably have triple digit number of chats (can't be bothered to scroll down the whole list to count them) and I'm pretty sure I'm repeating prompts and chat topics because there is no search function.
i agree we need this
you can search your chats in the desktop app
the desktop app is only on Mac for now
And therefore I'd like to get the feature to the browser client so that it's available everywhere. I use mostly Windows, occasionally Linux, and never Mac, so the desktop feature in this case is out of reach.
in that case i have to agree with Kim once again.. openai shouldn't expect people to own a macos product, or any apple product for that matter
a Windows app is coming later in the year
we shall see. so far openai is mostly just hype opposed to actually making things happen
I have the Mac desktop app and I use it all the time, it's fantastic
the announcement from yesterday was completely useless aswell, considering free users only get like what? 8 prompts with 4o?
8 free prompts is better than 0
but its the reason why they can make the other multimodal features also free. because they definitely wont make loss with that. otherwise they wouldnt do that
Technically the documents are written in tokens and there are multiple memory files, it's their version of words.
Hi my name is Jeffrey Guy. I am waiting strongly for the new features. Can't wait 🙏. Guy.
Try support.openai.com they used to direct you to their support bubble were their AI assistant can get details, but I can't find the support bubble right now.
Chrome on Ubuntu has been rock solid for me throughout.
No, not at all.
when will gpt4o native image output be rolled out?
, yo sorry for pinging. Don't you know anything about internships at openai?
no eta
They have a residency program: https://openai.com/residency/ which you'd find openings for on the careers page, if available: https://openai.com/careers/
Thx, really appreciate it 👑
5/30/24 @ 9am. Has GPT 4o been rolled out to all users? : "Yes, GPT-4o has been rolled out to all ChatGPT users, including those on the free plan. This new model offers GPT-4-level intelligence but is faster and has enhanced capabilities across text, voice, and vision. It is designed to provide a more natural and intuitive user experience by supporting multimodal inputs, such as text prompts, voice commands, and real-time video (OpenAI) (The Tech Portal).
Free users have access to GPT-4o with some usage limits, while paid subscribers, such as ChatGPT Plus and Team users, benefit from higher capacity limits and additional features (OpenAI) (TechRadar).", Will long term memory roll out to free users?: "OpenAI is currently testing the memory feature for ChatGPT with a limited number of both free and paid users. This feature allows the AI to remember details from previous conversations, improving its ability to provide personalized and contextually relevant responses. The memory feature is expected to be part of a broader rollout in the future, which will include more users across different plans (OpenAI) (Android Central).
For now, the memory capability is primarily available to ChatGPT Plus and Team users, with additional benefits such as higher capacity limits and more advanced functionalities (Android Central). Free users will start noticing the memory feature as it continues to be tested and rolled out more widely over time."
can you please not paste walls of text
sure
thank you
np
hey gpt using the last time ALOT cpu and ram ( just by using the website )
ive read in the internet this problem is known for a very long time
Any word on when the conversational AI will be rolled out? (interruptable)
"GPT-4o real-time voice and vision will be rolling out to a limited Alpha for ChatGPT Plus users in a few weeks. It will be widely available for ChatGPT Plus users over the coming months."
Thank you friend, so excited for this
Is this massive memleak known and being worked on??
In ChatGPT, is there a way for the vision model to be invoked for what I'm not uploading images directly?
To give you an example, i'm uploading a PDF of a presentation, I'm using code interpreter to convert it into JPEGS, so it creates 10x jpegs for 10x page presentation, can I then somehow direct the model to use the Vision part of its capabilities to look at these jpegs? I've tried all sorts of convincing in the prompt (incl. deleting the pdf), but it doesn't seem to want to actually use the vision model for that. Any thoughts?
This channel is for ChatGPT 3.5
Not GPT-4 or 4o
well its for alll models that Chatgpt support :/
...?
ehh That cahnnel right now #chatgpt-discussions is for Chatgpt 3.5, Chatgpt 4 and Chatgpt 4o. :P
what channel is for 4 or 4o?
if your using it inside chatgpt it is this channel
thats what i would think
that is overall for models like that ones inside API
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.
There is no such thing as ChatGPT 4 or 4o
It's GPT-4 and 4o
in fact there is!
this channel is for chatgpt related stuff which includes using all models
Then what is #gpt-models for?
models that are in chatgpt and api
^ this
#chatgpt-discussions - Chatgpt
#gpt-models - GPT Models
Reposting my question below, for clarity, I am asking about ChatGPT, not API (via API it would be easy)
In ChatGPT, is there a way for the vision model to be invoked for what I'm not uploading images directly?
To give you an example, i'm uploading a PDF of a presentation, I'm using code interpreter to convert it into JPEGS, so it creates 10x jpegs for 10x page presentation, can I then somehow direct the model to use the Vision part of its capabilities to look at these jpegs? I've tried all sorts of convincing in the prompt (incl. deleting the pdf), but it doesn't seem to want to actually use the vision model for that. Any thoughts?
code interpreter has no access to vision and docs dont support embedded images yet https://help.openai.com/en/articles/8983779-are-you-able-to-handle-images-embedded-in-docs-presentations
you said free users can use the new features
nuh uh
lol spaceorcas
hey
I put a funny picture in #off-topic, and the reply to that is about as amazing and intelligent as you'd expect. 🙂
is anyone here using ChatGPT to practice a language?
Ah, thank you
will sky ever come back?
I can't stand Juniper, I've switched to using Pi for any voice stuff
Ah yes, once again my heart skips a beat upon seeing the mythical ping icon in #announcements, just to see it doesn't have anything to do with the release (or even the release schedule) of the new voice module.
Same...
is anyone's chatgpt lagging?
Agree
The OpenAI Discord is an actively moderated server.
• 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.
• Report all sensitive and offensive content in the feedback reporting tool in the ChatGPT web UI instead of here on Discord.
free version is lame ngl
always dont know what the limit is
im gonna renew it if i got a chance
Are the gpt 4 use time and gpt 4o use time count together? I haven't a chat with gpt 4 today but when I use it I already run out of limit. Did any one kown why??
same thing
Anyone else Firefox browser crashing will using gpt 4?
F
did anyone have a chance at allowing a tab take more memory?
mine are capping at 4gb and idk what to do
i don't want to create a new chat because it keeps losing context
do you have a Mac? if so use the macOS desktop app
no, windows
@opaque parrot@ivory kite isn't it fun to be kept in the dark? :D
I have taken to copying my conversation, saving the content in a PDF, and then reintroducing it as an attachment in a new thread. Works for me.
you mean Ctrl + A copy? or you ask gpt to save the convo into a pdf?
I suppose you could just save the HTML document and attach it, it'll understand
I literally copy the whole conversation sometime, paste into a Word doc, and then save as a PDF.
what is your query for reading the pdf?
When I flesh out a concept, it can hit about 50k words. So, I found that grabbing the information as a PDF and starting a new thread with the PDF, allows me to ask the AI to review the document. This comes in handy when fleshing out 100 chapters with ten story beats each for fiction.
what did you try to do
uploaf images in chatgpt
what did it say?
do you just ask it to read it?
it just doesnt have the button
@lost saffron …I do. And to hold on to it. Works better with v4 vs. v4o
Is there a message limit for GPT 3.5?
no its infinite, but theres one for gpt-4o (if ur a free user)
Kk
the button is just kinda missing
do the other features work, like gpt store or browsing
this works thank you
wait lemme go to the phone
i can browse gpts, yeah (also im a free user btw)
do you use the new chat as the pdf or do you just add it with the old chats?
if your 4o is on cooldown then it might disappear but the other features would also refuse to work like trying to browse or use gpt's
its not on cooldown
ay, glad to know there's an actual macos app!
I open a new thread, drop the PDF in the new thread, and then continue to work.
i mean you create a new pdf of the old thread or you combine the old and new pdf together?
another announcement, another disappointment. 😦
Why would they think that anybody here cares about some initiative for some nonprofit company thing? It’s completely irrelevant to us.
I copy everything from the old thread and put it in the new one via PDF.
Some place are happy to develop and work with it. Sharing is caring !
Latex bug still going?
I’m not against it, but it is a pointless thing to announce it here as I’m pretty sure 90+ % of the members in here are individuals using a personal subscription.
especially since everybody is expecting “le grand announcement” 😁
there are thousands of people here and I am sure that some of them are interested in all manner of things related to OpenAI not just the next shiny toy
<- wants new shiny toy
I’m confident that number is far smaller than the Openai crew thinks.
I have to say other than making google redundant and help with programming the limitations OpenAi have on the models effectively render it useless for my personal uses....
ehh... usually based on my exprience like 20-40% of discord users lurk without any intrest beyond the surface
as someone that has no knowledge of coding, I can say that chatgpt gave me access to do something that I would otherwise be unable to do.
It’s limited, but it’s still more than I would’ve achieved on my own.
as someone who can program I have to urge you not to release anything you make with OpenAi, as it can lead to many issues some even legal in the long run
I don’t have any plans to do that. Coding is really not my thing.
I’m just looking forward to the day I can create a doom clone and use generated sprites of my friends as the bad guys. 😁
I actually read a paper about Ai generated sprites for video games
Coding makes me sleepy
C++ for sleep
intresting insights and OpenAi's Dal-E could in theory be modified to generate even better results but it could cause some issues in the long run and could bring down the gaming industry in the long term by introducing a ton of shovelware
I work in engineering. I can’t wait until an ai can build 3d models of mechanisms, to see how it fairs with human engineering.
it already can
I just want AI to make me a sandwich
google has a model that can even predict how enzyme's are structured
i found a couple 3d model generators but they’re still pretty primitive
F* Google 
Is chatgpt down for anyone else?
no
Working for me.
working and i also have the new voices and video thing
I'm kidding
haha big funny
its not on cooldown!
The latest update?
video thing?
Voice thing
SYKE!
Sky
petition to bring back sky to chatgpt
Petition to bring on Arnold’s voice to chatgpt
Petition to bring Mr. T from A-Team and Darth Vader from NOT Star Trek
Sorry Trekkies 
Arnold Schwarzenegger voice
Surely we will have celebrity voice plug-ins at some point?
I'LL BE....BACK? MAYBE.
I don't like all the neutral tone voices...I'm not saying have distinctive male/female voices but some are just so in-between.
And for the female roster, Sarah Connor.
4o with Arnold.
Make Arnold sing
Add R2D2 so we have absolutely no clue what the answer is.
Add Yoda so it speaks in riddles
Batman voice actor from the 90s cartoon. Very distinctive voice.
when is the new voice mode released
now actually im using it rn
Add your well-crafted prompts to our #1019652163640762428,
or share your interactions with ChatGPT in #1050184247920562316!
Yes, we need his voice.
The Super Mario voice, right?
Is this new with the "Option 1" and "Option 2" in 4o? Never seen that.
Kevin Conroy's voice 🫡
no that doesnt exist sadly
Nintendo getting ready to launch a nuke after someone named Luigi is born
whattt
how
idk the headphone thing (mobile only)
that is old
wat
it released 6 months ago buddy
Is there a way to download a chat
the new voice mode is that one with video
whaaaaa
besides message rates is there any other difference for free chatgpt users and chatgpt plus users?
We have regular GPT 4 still
Even though there's no point in using it lol
lol, right im trying to think whats the difference now. Can free use dalle and all that stuff too?
No.
Anyone know?
ah i saw they can use gpt's, I guess I will have to test it out to compare
you can share a single chat but there is no separate download feauture other than the gdpr export
Gdpr export?
Performance with Plus is very bad today, might be due to load with new "all features for free" policy?
Also experiencing something new (Firefox v126.0.1) ChatGPT (Plus/4o) will stop while generating a response and the UI becomes non-responsive. Need to refresh the page to find the output did complete generation, just not output.
settings -> export data
Oh ok thanks
Yeah I'm just weary because i just had a weird issue where one of my conversations had disappeared for a bit
And i was worried I'd lost it
GM friends, what advantages do those of us who have had our paid subscription now have that we have the same uses as the rest ?
more advantaged data analysis
Increased usage, GPT creation, DALL-E 3
gpt creations is a big one, it could actually be a huge way to self promote if it takes off in the right way
hopefully they dont mess it up, but there better be a new model soon too lol
we just got a new model
its just better for users though, its not actually smarter
no, it's smarter according to the evals
its better at looking at images, everything else ive tested is the same or worse
Evals are fine, I don't personally care about them, they said claude 3 was on par with gpt4 when its dumber than gpt 3.5 in a lot of ways.
I mainly care about solving complex coding tasks over several queries, semantic scoring, how complex of prompts it can adhere to, and emergent behavior. More than how it scores on the sat or if it can 1 shot code or whatever else they measure.
Its better at not denying requests for ethical reasons it really should run, thats the other one I forgot.
the update is still not here huh? how long has it been now since they said a few weeks?
weeks for select group of partners. months for the rest of us
thats not what openai said
source?
or at least not initially
their blog..
