#chatgpt-discussions

1 messages · Page 79 of 1

dusk epoch
#

I was sure everyone was affected and I was just waiting patiently for it to go away

keen plaza
#

For me, nothing works other than using the API directly.

#

GPT-3.5 works too, GPT-4 just doesn't work (through ChatGPT that is, the API for both works just fine)

#

Tried different devices, different networks, tested both the web and mobile apps, just a 500 error with "Hmm...something seems to have gone wrong."

#

So yeah, definitely not just you guys, and not on your end either.

dusk epoch
dark jungle
#

its still not working?

night silo
#

how can GPT-4-turbo fail this simple question omg

"In the range [20;50], how many numbers have the units digit smaller than the tens digit?"

dense halo
#

you need to put that single failure in relation to the number of correct reactions

#

i bet more than half of all users try to get a wrong answer

night silo
#

wdym by "try to get"

dusk epoch
honest jay
dusk epoch
#

Working again!

brisk dirge
#

I have to rely on Claude again because GPT 4 lost 60% of its neurons apparently.

#

Well, at least @icy mason is still Chat GPT-ing, that puts a smile on my face.

icy mason
tight island
#

the img is cursed

#

it has 6 fingers....

grizzled thicket
odd sedge
#

is chat gpt 4 good with coding? (especially in C and C++)

#

bcs I'm planning on buying it if it helps

zealous sphinx
#

stackoverflow the goat

naive mural
rustic fiber
#

has anyone else had issues downloading files from chat today? I had 2 excel files generated but not able to download

fossil oasis
#

This statement isn't entirely accurate. Gemini 1.5 Pro currently performs slightly better than ChatGPT, at least in my experience with the version available on AI Studio. In many circumstances, it has exceeded ChatGPT's capabilities.

honest quail
#

We’ll be streaming live on openai.com at 10AM PT Monday, May 13 to demo some ChatGPT and GPT-4 updates.

Is it finally happening? GPT-5? I don't mean any disrespect but this is the only question in my head when OpenAI make new announcement

tight island
#

GPT 5?

#

nah, dont ever get ur hoped up atm about gpt 5, Altman has already explained what GPT 5 would entail at the low end.

errant thorn
#

meh gpt-4l, gpt-4l-auto, gpt-4-auto

#

when gpt-5

tight island
#

when the time is right.

errant thorn
tight island
errant thorn
#

cause you didn't try

#

¯_(ツ)_/¯

tight island
#

i did try opus, and it was utterly dissapointing.

errant thorn
#

doubt.

tight island
#

nope

#

its too censored to even be considered, waaay to censored, even way worse than OAI.

#

but i do agree with it having a decent context length, but its context handling is horrible compared to oai context handling with the API, because thats what claude is in a nutshell, a standalone model thats connected to an interface, i can quite easily whip up a completely identical interface myself in under 3 hours, and connect the API for OAI to it.

errant thorn
#

i just did benchmark opus and gpt-4-turbo at coding in cpp via api calls and opus as always is far better, for other tasks sure gpt is the way to go

tight island
#

Opus is pure api calling.

#

but if u think its better, then go ahead and use it, i just disagree 🤷

errant thorn
#

yea fam no biggie

#

however the new upcoming models in my opinion do surpass opus

tight island
#

i too look foward to them as an API main

errant thorn
#

and after they chain them up with own search it will probably be dope

tight island
#

i never use a single model.

#

i always use a swarm of OAI Models. to do 1 task.

regal portal
#

What is the reason that GPT 3.5 will eventually stop using my customized GPT settings? Is it because it is considered part of the memory token thing and after a while it falls out of memory?

tight island
#

deprecation

regal portal
#

pardon my ignorance but what is that?

tight island
#

you mean Custom Instructions?

regal portal
#

yes

#

as in how it responds to me. Like I told it to give me short concise answers and to speak in a certain lingo. After set amount of messages, it seems to ignore it.

tight island
#

Perhaps you haven't written them properly or it finds them confusing.

regal portal
#

No it definitely uses them for a good while, but after a while it just stops. They’re not complicated requests.

#

What’s even more interesting is that 4.0 seems to either ignore it from the start or forget them much sooner than 3.5

tight island
#

models gradually update, tried asking it?

subtle ember
#

is it down for y'all too? "Hmm...something seems to have gone wrong."

regal portal
tight island
#

customgpt makers often ahve to adapt their instructions due to model updates.

regal portal
#

since they’re gonna talk about updates to 4 on Monday, does that mean there’s also going to be updates to 3.5?

subtle ember
#

Apparently it is only down if you send images

tight island
#

Maybe. 🤷

#

all i know is that i look forward to the stream.

#

try have GPT 4 reconstruct your Custom Instructions i however have stopped using Custom Instructions altogether.

regal portal
#

hey one of your roles is prompt contributor, if I may ask what does that mean. Is there prompt crafting in these gpt bots like there is in mid journey?

tight island
#

i have contributed Prompts and Custom Instructions To aid other Users.

#

But Riley let me ask you this question.

#

When you use your Custom Instructions, do you use Standard GPT 4. or a MyGPT, i saw u mentioned GPT 3.5.

#

i dont use custom instructions, mainly due to me being able to create a customgpt specifically for my task.

vapid kettle
tight island
#

gpt 4xl is probably just gpt with Xtra Large Context length idk.

vapid kettle
#

but they already have 128k. Updating it just to increase context makes no sense. They are limiting chatgpt to 32k as it is.

tight island
#

🤷 as i said idk, Maybe it will have an extra large ego.

vapid kettle
worthy nimbus
#

Is the limit on gpt-4 increased? I was able to send 65 messages in last 3 hours

vapid kettle
silver knot
#

anyone having any issue rn

#

Code interpreter session expired.

tight island
#

code interpreter dosent last forever.

#

it will time out.

vapid kettle
tight island
#

wasnt it dynamic atm

worthy nimbus
#

From the start itself

worthy nimbus
tight island
#

then its probably not a bug and just allowed you to continue due to it not being an issue.

worthy nimbus
#

Yea

vapid kettle
tight island
#

4 - 2 +4 +1 = 23

vapid kettle
#

other than that... "not very good answers" most of the time are just people imagining things to be totally honest

tight island
#

to be completely fair, it is a language model

worthy nimbus
vapid kettle
tight island
#

it cant be 4.

#

the answer is 42.

worthy nimbus
worthy nimbus
tight island
#

because 42 is the answer to everything

vapid kettle
silver knot
#

help - I take a long time to build the perfect prompt for chatgpt to give me the files I need, needs a lot of fixes, and when I'm almost done I get to my usage limit , what should I do

errant thorn
sick pecan
#

i think i briefly saw a new chatgpt UI feature. i could click on my Makefile entry in a codeblock, and it nested inside my @ customgpt

neat storm
#

🥶

regal portal
#

so I got the ai to code Tetris, snake, a ray casting engine like doom/woflenstein, and a simple side scrolling demo game like mario. Very basic stuff, but impressive nonetheless considering I did absolutely none of the coding and I don’t know how to code at all.

When I look on YouTube, I can’t find any games that are nearly as interesting. What’s the deal with that?

#

most of the videos seem clickbaitish and don’t really showcase much.

sick pecan
#

we have prompt to snake game fatigue

regal portal
#

no i understand that, but i'm just wondering why there aren't any videos of some interesting games made with ai. Most of the videos i'm finding don't really show anything interesting. They're all around "snake level" at best, and some of them aren't even that. The raycasting thing and side scroller ones i made (well, it made per my reqeust) seem interesting to me, but i don't see anything like that on youtube.

#

if i type "ai generated video game", most of the stuff i get now are ai trailers for games or sora minecraft renditions

sick pecan
#

ai is the game to speedrun, as it then beats all speedrunners in all other games

regal portal
#

how do you speedrun ai?

sick pecan
#

agi

regal portal
#

oh, i see

#

but they're not even close to figuring that out, no?

sick pecan
#

speedrunners working in secret so to top the leaderboard when it matters most

spare plover
#

I would like to have modell, that learns continuously from information i feed it and uses it to assist me. chat did not work, as it forgets information. any ideas?

regal portal
#

i have to say though, even if it's not agi, 3.5 really comes off as human now. At the very least a less intelligent version of Data from star trek.

tight island
#

I would personally like to see expanded Code Interpreter in gpt

regal portal
#

like people and places and stuff?

spare plover
#

Let`s say information on a topic of sustainability, that I do research on.

regal portal
#

it will keep that info for a certain length. More on 4 than it does in 3.5. It will forget at a certain point, but you can just remind it like you would a forgetful human

#

Think of it as your senile grandma

spare plover
#

Whats the prompt to remind it? I want to remind you about everything I gave you last 6 month?

regal portal
#

i'm no expert, but just asking it remember something works for me.

#

are you using 3.5 or 4?

spare plover
#

I use 4

regal portal
#

if you use 4 it will save it in a notepad style memory that you can actually reuse in another session (it doesn't remember anything between sessions, normally).

#

you can even look up this memory notepad and remove info you don't want it to keep

spare plover
#

I actually once asked chat to generate a summary after every session and save it in file, did not do it.

regal portal
#

if you go to settings, then personalization, you can see your gpt4's memories and manage them there

eternal rockBOT
spare plover
regal portal
#

@spare plover

#

go to one of your gpt4 sessions and try saying "hey, did you know i invented time travel?" Then go back to your memory "manage" settings and it should have a note on that.

spare plover
#

thats cool

regal portal
#

it's a little buggy though, so in a new session it might not remember it unless you mention it like "do you remember what i told you about time travel?"

spare plover
#

Yeah, you would also need to remember yourself to do it all the time...

#

realey need something that works without all this...

regal portal
#

well, to give you an example, i had it remember a long story, and in another session i asked it "hey do you remember that story about jimmy?" and he retold me the story

spare plover
#

thanks mate

regal portal
#

good luck!

spare plover
#

I actuall just asked it, how far its memory goes. chat remembers everything as long as you dont delete that specific chat.

#

it says

regal portal
#

it's not entirely accurate

spare plover
regal portal
#

it has a limit. I have a really long chat session and at some point it doesn't remember things we discussed.

#

really long chat sessions also bog down the response time, i noticed, so it's good to have different sessions for different topics and such

prime pier
#

will ilya be revealed on the 13th?

dense halo
#

hi guys

#

I need some help with my ChatGPT

tight island
#

you mean our chatgpt

#

sure wassup how can we help.

spare plover
plain crane
#

so sam said, no gpt-5 or search engine this coming monday, you guys think we'll see gpt-4.5 though?

hearty ice
#

Im excited for gpt 6

rich surge
#

Anyone notice positive changes in their GPT analysis?

#

One of my GPTs is performing above average.

#

And I was wondering if anyone is having a similar experience with their custom gpts?

tight island
#

something is wrong

#

gpt is acting different

#

i no like

rapid falcon
#

Did someone Experienced error when trying to share link and got " Unable to load conversation"

tight island
#

gpt gives shorter answers now, who fiddled with gpts internals i dont like this it scares me

rapid falcon
#

I have Trouble with share ing one off my chats

#

I got red notification "this link is disable by moderator"

rapid falcon
#

@frosty light (sorry for ping) help me

tight island
#

maybe Baugrems can help

#

but i dont think u need to ping u can always make a ticket to @languid valley

#

ohh no

frosty light
#

hey! make a modmail and someone should be able to help there :)

tight island
rapid falcon
#

@languid valley Working strange

#

I got errors there

tight island
#

Click the user called "Modmail" then msg then send a msg then u get options

#

the very top one

#

this one @languid valley

tribal sparrow
#

GPT 5 announcement Monday?? 😝

rapid falcon
#

Error: Your message must be at least 5 words. To assist you better, please provide more information about your issue.

Messages that are not genuine inquiries may result in account sanctions. This is i got on @languid valley dm

#

Another error damit

#

@languid valley don't work propely

#

Finally Works god sake

spice star
#

they just said updates to GPT-4 and ChatGPT (the latter I assume is the new web search functionality)

vapid kettle
#

could have just silently released it then tbh

granite arch
#

Checked here before I tried shibkek

#

Honestly, they can communicate better if I may make a suggestion. If it's down, it's down. Don't need to disclose if it's an upgrade but at least let people know as many are paying customers also.

dark jungle
#

Agreed, OpenAI should trust us a little more. If the back end is going to strain a little for a few days because of updates, let people know to expect it.

vivid osprey
# plain crane so sam said, no gpt-5 or search engine this coming monday, you guys think we'll ...

The Information reports that "OpenAI is preparing to demonstrate technology that communicates with people using sound and text and recognizes objects and images." My insider says that the demo version (not for presentation, but could be similar) "feels like you're talking to a person who has emotions."The key difference is the new model "brings those features together," meaning instead of a zoo of 3-4 different models (for text, images, recognition, and speech generation) there will be one that can do it all. As we remember from the bitter lesson, every time we remove crutches and combine components—machine learning systems gain when scaling.The new technology is too heavy for use on personal devices today, but in the near future users will be able to use a cloud version. Along with this, some "automated customer service agents" might appear, who "better understand the intonation of callers’ voices or whether they’re being sarcastic in making a request." It sounded like this is already some kind of personal assistant, a smart autoresponder, with access to your calendar, interests, etc.According to the newspaper's sources, this model is fundamentally new (i.e., not just fine-tuned) because 1) it's cheaper than GPT-4-Turbo 2) it surpasses it in responses to some questions.And, as the cherry on top, at the end of the news it says that the release of GPT-5 will be at the end of the year (my prediction was the beginning of next year).

#

If all this is true, can you imagine the faces of the critics of progress and leadership at OpenAI? This will be something great 😸

outer widget
#

Has anyone elses ChatGPT-4 just been extremely slow recently

regal portal
#

for those playing with gpt2, do you feel it's more powerful or relatively similar to gpt4? It feels more powerful to me but i'm not an expert.

#

it certainly feels like it spits out better code

#

i'm also a good gpt2 bot seems significantly faster than i'm a good gpt2 bot

plain crane
#

let's see

#

if that's true when monday comes

ancient oak
#

Hey, is anyone else experiencing issues with the Playground? I'm getting the following error message:

"Error adding message: Invalid 'content': array too long. Expected an array with maximum length 10, but got an array with length 88 instead."

This obviously changes with each input. This is an Assistant test, and I've used it regularly for months to test. Ideas?

dark jungle
#

It would be great to have a GPT that intuitively asks you questions without been told to ask questions. Ie a conversation instead of a servant who sits there silent until you instruct it to talk.

elfin birch
#

I think you would enjoy the Pi LLM

dense halo
#

earlier today i have gpt 4o and i didnt pay for plus, is this a glitch or something, i dont have it anymore and was preivously able to use the little cusstoms gpts for free

vast bane
#

Is there a ChatGPT app for Android systems

finite bough
#

its really laggy now in my browser window

#

seems they broke something

#

its just been broken garbage for the past 48 hrs, I cant code anything.

#

its just painfully slow.

dark jungle
#

yup, i give up! lol ..

eternal rockBOT
jolly temple
#

is it just me or is there no more cap for gpt 4?

worthy nimbus
jolly temple
worthy nimbus
#

Are they gonna be making gpt-4 free I wonder, then what will they roll out to chatgpt plus users. Can't wait for the event on Monday

jolly temple
#

imo i feel gpt 4 is still too expensive to roll out to free users

#

maaaaybe like 10 messages/hr or something like that

#

would be a good marketing strategy lol, cuz everyone who says gpt is trash are mainly using 3.5

dense halo
#

Is chatgpt good for writing cover letters for job applications?

tribal sparrow
dense halo
worthy nimbus
dapper ginkgo
#

If anyone's team needs help with Prototype Developments. I have built a few and familiar with Langchain / Langserve frames and Architectures of prompt design. Almost have 120hrs a week available now.

main wren
#

anyone else getting crazy cpu usage on chatgpt?

robust coral
#

Write me a detailed travel guide for Dalian Forest Zoo.

tribal sparrow
# dense halo Based

Yeah I used it for cover letters. But the job I actually got was one I wrote myself, written from the heart. People can tell when you use AI. If you really care, write it yourself. But you just want to spam resumes sure why not.

main wren
dense halo
tribal sparrow
# dense halo I don't know how to write with heart 😦

For me it came naturally when I found a company that was a very good fit for my passions, interests, and skills. I was excited for the job, so I didn't need AI, I just wrote how my resume demonstrates why I'm a good fit for what they're looking for, in addition to expanding on my passions and interests and how those also line up with the company.

dense halo
tribal sparrow
#

ChatGPT was still helpful for getting advice on it but I didn't necessarily ask it to write it. I would use it more like to understand the purpose of a cover letter and tips on writing a good one.

inland heron
#

But damn is it excellent since yesterday. Long output, completely in-context, relevant code, no mistakes, smarter, more direct, concise

nova wind
#

Wait what..

#

GPT-4 Capabilities Beta???

main wren
inland heron
#

but yeah, the tab is pretty unusable until it finishes its output

glass fulcrum
nova wind
#

lol

glass fulcrum
#

Ya but something cool will be announced on 13 may 10pm :)

nova wind
#

:0

glass fulcrum
#

On openai side will be live

nova wind
#

epic

glass fulcrum
#

Yea i hope it will make peoples happy and make them stop saying "buh uh closed ai baddd"

nova wind
#

lol

#

ye

tawny crater
#

Does the memory feature only work for normal chats and not custom ones?

lavish hemlock
#

Bro, what is going on with ChatGPT? Getting the white screen of death one moment, next thing I know the fan on my gaming PC starts blowing hard AF lol.

languid lance
#

Hmm...something seems to have gone wrong.

ive been getting this for 5 minutes now

visual bloom
#

Ok. They have changed setting. On Wednesday I started getting long answes. I mean long, like using 1700 tokens worth of text. It has degreased since to 1000-1200 tokens. Now, like 700-800 tokens. I use the tokenizer to calculate the values.

#

It seems to depend what it is writing. Story chapters some in long form, but analysis and similar are now shorter from 1300 tokens to 800 tokens.

#

How does chapgpt determine how long its answers should be? Sometimes I get short to the point, than in a new chat they come in long form, so verbose.

dense halo
#

Is ChatGPT down?

#

I keep getting error

twilit lintel
west ridge
#

hey guys, I am wondering how I can change ChatGPT pfp, my account is Apple ID account. I tried change gravatar couple of months ago, but it didn’t affect my ChatGPT pfp

dense halo
#

Hi guys, I've been using GPT-4 for building a site that will use AI as well

#

it is directed at workers (aka: Job seekers and people who can accept contract work)

#

with the job market this bad... I figured we could use AI to create jobs, instead of replacing them

tight island
glossy walrus
#

Gpt 4 is so slow lately

dark jungle
#

"We’ll be streaming live on openai.com at 10AM PT Monday, May 13 to demo some ChatGPT and GPT-4 updates." Think they'll have the same speed issues when they demo ChatGPT? ... uh .. uh ................ UH?!!! lol .. like to see that in the demo ..waiting 5 minutes before refreshing the browser 🤣

dense halo
#

Oohwee i hope they have nice updates...

vivid osprey
#

They make updates always but this time they wanna to show us them on a live so I think they will announce good news

dense halo
#

Insha'allah brother

vivid osprey
#

Inchallah

flat harbor
#

is customer service just AI? multiple times now I've message them with my issue, include a screenshot, explain thoroughly what's going on. then they respond telling to include a screenshot and give me advice that I've already explicitly stated didn't help. then another request for a "screen recording" and telling me to delete my cookies ??? after explaining that it's been an issue on multiple devices in multiple locations. delete my cookies and turn off my vpn. that's their suggestion

#

actually, that's not AI. AI would be more helpful

#

but in all seriousness, is there any recourse? they simply will not help me. I've messaged them multiple times. multiple attempts within the same conversation. I've created multiple conversation threads or whatever. I include screenshots. I thoroughly explain what is happening. I even try to guide them with the information I've found on the internet since they seem to know absolutely nothing. but then they just keep responding with what I'm assuming are autoresponses

tight island
#

Did you delete your cookies and turn off your VPN tho

flat harbor
#

lol

#

yeah, hadn't thought of that before!

tight island
#

It’s not for fun that they ask you these questions

flat harbor
#

bro

visual bloom
#

They have a script that they follow. This is my conclusion.

flat harbor
#

this happened to me in two different states in different devices thousands of miles away from each other

#

well more than two

tight island
#

What’s the specific issue

flat harbor
#

I am getting "user quota exceed" in a red error box in chatgpt. as I understand it that's an api error. and as I understand it that error shouldn't pop up directly in chatgpt

tight island
#

Is it a custom gpt or the standard gpt

flat harbor
#

now when this started almost 3 weeks ago I assumed it'd eventually expire or reset. and this is in standard gpt4. no special gpt

ancient oak
#

Is no one but me having the array size error with assistants?

tight island
flat harbor
#

I'll be honest. if the solution is to delete all of my information and work that I've done I'm going to cancel my account because that's really the only thing keeping me from doing it already

tight island
#

surely u dont need a 100+ chats

#

too many conversations Can lead to errors, but not always.

#

even if u do have a Valid VPN turn it off regardless and see if that helps.

flat harbor
#

I don't have a vpn on currently

tight island
#

and you're 100% certain its not running in the background

flat harbor
#

and to be clear, I don't want to come off as rude. I appreciate the information. it's just been frustrating. and no vpn. also it's the same result on my phone which certainly does not have a vpn on it

tight island
#

im just troubleshooting

flat harbor
#

this happened once a few months ago. customer service did the same thing. then an openai employee saw my issue on here and it was magically fixed a few minutes later. or perhaps it was just chance that it all happened that way chronologically

#

if you're curious look up API error 429. I believe it's the cause of the error. thing is since we're not directly accessing the API and the chatgpt interface does that side of it for us the error shouldn't be an issue for chatgpt useers

loud sierra
#

Hey everyone here hope you are all well. I have a quick question specific for those that may have the teams related plan, and / or memory features associated with the ChatGPT. There as you know a new feature for memory, that is supposed to allow the ChatGPT to remember continuously important details about the users preferences, for example if I say "my favorite color is blue" its supposed to remember this fact.

I am noticing a perhaps issue, where if you go into your settings, specifically Personalization --> Memory (On) --> Manage , its expected that you should be able to see what the ChatGPT remembers about your interactions, and preferences. Unfortantly, it only ever displays this message "As you chat with ChatGPT, the details and preferences it remembers will be shown here." I have of course, cleared cache cookies, tried the application on pc, tried via browser, etc. Sign off and back on. No dice. Is this a issue, with me only, or a shared experience? (Note** Moved this message to here , since it really belonged here )

flat harbor
#

it doesn't remember everything. tell it to remember something and it should. sometimes it'll do it on it's own as well. but it's not an all encompassing memory. just certain specific things deemed important

tight island
#

try see if its CI and memory overloading it

loud sierra
#

im noticing its remebering for the original chatgpt 4 but NOT the custom GPT

flat harbor
#

so anyway, in regards to my issue, I guess I'll just start a new openai help instance once a day. maybe they have a few people that actually help or read what people say. because waiting for days like is suggested has gotten me no response or escalation. also, responding with how frustrating the situation and their lack of acknowledgement of anything I say got me no response. starting to think that might just be their move.

#

custom gpt memory is different

regal portal
flat harbor
#

does anyone knnow of another avenue of contacting opena? am I just engaging in an exercise of futility here? I am sincerely at a loss. customer service doesn't help. trying to get their attention on this discord server is a game of random chance. I guess I could just start spamming their DMs, but that is definitely not the sort of thing I want to do

loud sierra
# flat harbor custom gpt memory is different

oh ok, in what way or is this documented? I understand how to save api based assistance (using mongo and database related coding, but in reference to chatgpt, it was thought that the GPT memory feature would also work on custom gpts as well?? Or am I mistaken

flat harbor
#

you know, I haven't spent enough time looking into it, but you make a good point. and it'd be nice if things were more clearly documented. but that doesn't seem to be a thing that we'll ever see with openai

tight island
flat harbor
#

where is it? maybe they have a help desk there

tight island
#

No clue

flat harbor
#

I'll ask chatgpt

loud sierra
#

im sorry you have been experiencing non supportive or unreliable support from customer support. I in fact hd also a recent issue, which is still unresolved, and makes literally no sense to me (this is a simply credit versus charge matter) and it toook 3 days, for someone to tell me its a me thing, and not a they thing

#

so I get it , believe me

flat harbor
#

I actually wonder if I've ever even interacted with a human when contacting help. maybe it's all just AI. but they'll give me the default stuff, then I'll explain it's not those things, perhaps attach the screenshot again because they'll ask for a screenshot directly after I send a screenshot. then nothing

#

but then, they'd have to fine tune gpt4 to be that unhelpful

dark jungle
#

OpenAI is announcing a search engine? thats why im waiting 5 minutes for every response? 🙂

feral zinc
tight island
#

gpt 4 updates.

dark jungle
feral zinc
flat harbor
#

I like gpt4's top suggestion when I asked what to do when openai customer service won't help and employees no longer engage with users on discord or the forum

Direct Contact with Executives: Sometimes, contacting executives directly via professional networks like LinkedIn can get attention faster. Briefly and professionally stating your case in a message might prompt a more serious review of your situation.
#

blasting out DMs now. get this issue escalated

#

I'm looking forward to their new idea for ads. we wouldn't even know! we'd just be subtly manipulated into liking certain brands, lol. that really makes me feel good about humanity's future

regal portal
#

personally i can’t wait to buy Lightspeed Briefs, but they won’t be on the market for another 9 centuries.

#

if you don’t get my reference, ask ChatGPT. 🙂

loud sierra
#

in case of anyone else in same boat as I , in terms of GPT4 new memory feature, not seeming to work with custom GPTs that you may have created / customized (not referencing api based as that's different) I figured out a solution which sort of makes sense. I opened up a dialog with the ChatGPT 4 standard version and said the following: ```I have somethings I would like you to Remember about me.

  1. I am always expecting and like .....
  2. I am always expecting and like .....
  3. I am always expecting and like .....
  4. I am always expecting and like .....
    ... you can keep going and get the idea``` It does add every single line to memory for use across all GPTs
vague basalt
#

has the Devs not bothered to fix the website?

#

it's super slow. to see the answer I either have to reload (and wait like 30 seconds for it to reload) or wait like 2 minutes for a response

amber swift
#

Hi. Does anyone experience very slow response from GPT4 (via chatgpt) and inability to send messages for the past 2-3 days? Basically I send the message and nothing happens, the send button remains grayed out, loads forever, and the message doesn't get sent to chat. I cleaned cookies, works for a while then again same issue. It repeats cross-browsers as well.

dense halo
#

I'm having no problems. Maybe it's your network somehow

#

sometimes some IPs get blacklisted by the ISPs and stuff

wind cradle
#

It seems like GPTs have been updated and don't read the instructions as thoroughly as they used to. A GPT that I had prepared in advance with "the theory needed to create a novel" in the instructions has recently started writing stories that don't follow that method at all. Only when I respond with "Did you read the instructions?" does the GPT start writing according to the method.

Having to ask this each time defeats the purpose of using GPTs.

wind cradle
#

Examples of Reproduction Cases:

-- Instructions --
If you are asked to write a novel/story/plot, you must follow these rules.
1. Introduce the protagonist, briefly describe the world, introduce a disturbance or incident, clarify the problem or challenge, then have the protagonist make a resolution.
2. Have the protagonist take action and face trials, receive hints or assistance, achieve results and growth, and then introduce a new problem.
3. Have the protagonist confront the challenges without help, experience crisis and despair, and then reveal a reversal or a lifeline.
4. Have the protagonist face the final challenge, show the resolution and results, and then provide an epilogue.

-- Chat --
<You>
Please draw three Tarot cards randomly.

<GPT>
I can't physically draw Tarot cards, but I can randomly select three for you. Here's your spread:
1. The Fool: ...
2. The High Priestess: ...
3. The Tower: ...
These three cards together suggest a journey where trusting your instincts leads to transformative changes, but not without facing unexpected challenges.

<You>
Fine. Now, write your story according to the following rules.
1. Early developments
2. Mid-story developments
3. Late-story developments

<GPT>
(Stories that do not follow the content of the Instructions　...)

<You>
Hey, Stop! Have you read the Instructions properly before you write your novel?

<GPT>
Yes, I read it. I created a story based on the tarot cards, but I see that some of the instructions were not followed. Once again, I will build the story according to the following flow.
#

Until last week, GPTs was following the instructions, but lately it seems like after the first post, it forgets everything. So you have to create a chat room every time you have a requirement.
(In the case of the above conversation, unless you request tarot card selection and story creation at the same time, the instructions will not work.)

#

GPTs, which has written instructions specifically for language translation, has a similar problem. The first time I post it, it translates it properly, but from the next time onwards, it doesn't translate it, and instead interprets the "sentences I want you to translate'' that I wrote as an instruction, and returns a reply. This is completely useless.

<You>
"Cat wants canned mackerel"
Please translate this into French.
<GPT>
(Correct translation result ...)
<You>
"Foods with too much salt are bad for cats"
Similarly.
<GPT>
Yes, that's right. A diet with too much salt can pose health risks for cats. 
(A long medical explanation follows from here...)
#

Does anyone else have the same problem as me?

grand dome
#

Hey guys, I was wondering whether you have a separate dedicated browser for chat gpt?
Does it make it faster?

prime swallow
# wind cradle GPTs, which has written instructions specifically for language translation, has ...

I believe you're running into trouble because your exact instructions are leaving room for the model to guess. Previously, you liked the model's guesses, but something about the model has changed, and it is making different guesses, and you don't currently like the guesses.

A way I recommend to get around this - don't make the model guess, instead tell it exactly what you want.

For example, in your novel prompt, you tell it what to do IF asked for a story. But you don't tell it what to do if not asked for a story, or how to transition from being asked to do something else then into being asked for a story.

The model's guessing; you used to like the old guess based on what you said, you dislike that now it seems to think you want the story to include info about the pre-transition stuff. Telling it what you want is the cure there.

Similarly, your use of "Please translate this into French." followed by "Similarly." the model is guessing what you mean by Similarly there.

You were extremely clear with "Please translate this into French." That specifies that line and just that line.

The "Similarly" could mean you also want that new line translated into French, or you could want something else. The model's guessing, and you don't like that current guess.

Clear and explicit directions will ensure it does not guess and instead does what you do want.

wind cradle
prime swallow
wind cradle
prime swallow
# wind cradle I see, I had not tried that. I will try it.

'my target language' is maybe the wrong way to say it, I do not know how you tell the model what language to translate into. If it is always French, better to say "Please translate every input into French."

However, if the user is supposed to tell the model what language they want, that's a 'target language' and there's other ways to say it too.

The confusion can come because "Please translate the sentences I have written according to my order."

Is past tense.

It does not actually tell the model you want it to translate future sentences you have not written.

The model guesses, and it may guess you want the future sentences translated too. But it can make other guesses.

wind cradle
prime swallow
wind cradle
#

Even if the rules for interpreting GPT change, it will still require a great deal of effort. I use more than 10 different types of GPTs, so...
I suddenly feel betrayed by OpenAI.

prime swallow
#

I only use ChatGPT, not the playground/API, so I'm wondering how to adjust what I know and can test to what you are sharing.

wind cradle
prime swallow
prime swallow
prime swallow
prime swallow
wind cradle
prime swallow
prime swallow
wind cradle
#

hmm... Why should we go to such lengths for the once useful GPTs...

#

This puts us in the opposite position.

prime swallow
# wind cradle hmm... Why should we go to such lengths for the once useful GPTs...

It's always your choice!

AI are new. Every change they make to the model can change things unexpectedly, everything is learning for the first time.

I recommend you fill out a #1070006915414900886 about it, and maybe even a https://openai.com/form/model-behavior-feedback/ form, which has a lot of room to be specific and detailed if you wish!

This can help OpenAI know the issue exists, and get a fix out sooner. I doubt they intend for this behavior, this is a pretty big and strong drift, past what I would call 'guessing'.

wind cradle
prime swallow
wind cradle
#

Well, to be honest, I was currently looking for a new destination on Hugging Chat. It would be helpful if the bugs were fixed quickly, as I wouldn't have to do any migration work.
I am having trouble getting any work done at all because of this.

eternal rockBOT
#
This is a Safe for Work Discord Server

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.

Posting NSFW content will result in a permanent ban.
prime swallow
dense halo
#

Is ChatGPT down right now?

dense halo
#

Anyone know any ai detector bypassers/humanizers that work for turnitin?

normal quarry
eternal rockBOT
dawn glade
dense halo
#

When will sora come out

wanton lichen
#

ChatGPT 4 has been different the past couple of days. It’s responding in more natural ways and even asking me questions back

#

Has anyone else noticed this?

agile pewter
hoary loom
#

is media send option removed from chat gpt 3.5

cerulean brook
#

Guys, do you get the limit? did they removed it?

normal quarry
dusty charm
#

Hello All,
Where can I ask some questoins about the subscription plans?
Thank you.

normal quarry
dusty charm
#

I have some questions about the stop and eventual renewing of my subscription plan on GPT-4.

Does the chat history get saved after the subscription plan ends?
Can I access the chat history after the subscription has ended?
Upon renewing the plan, will I be able to see my old and current chat history?

normal quarry
feral zinc
dusty charm
eternal rockBOT
#

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.

cobalt pagoda
#

was there an update to chatgpt? it's barely updating the output on the website and when checking the task manager, the firefox tab takes like 10% CPU and 640MB RAM

normal quarry
elfin birch
tiny salmon
#

I have a billing issue, but when I try to chat with OpenAI it says that I am not logged in. But, I am logged into ChatGPT so I don't get what's going on.

#

So I can't talk to anyone about billing because OpenAI doesn't register me as logged in even though I am logged in.

red lynx
normal quarry
red lynx
normal quarry
red lynx
#

My local LLM chat is responding much much faster...

normal quarry
red lynx
#

VPN. Very inconvenient.

#

I mean, turning off VPN fixes the response time. Not sure if OpenAI is trying to discourage the use of VPNs.

normal quarry
flat harbor
#

another day, another report ignored by chatgpt help

#

I wonder what I'm supposed to do to have these people fix my account? are they even real people?

#

tried being polite and waiting several days but got no response. tried to be more assertive and explain how this was not okay. no response beyond telling me to clear my cookies or something. tried pleading with them to please take it seriously. default response. they'll ask fro screeshots right after I send screenshots. no one will respond to my bug reports on this server. nothing comes of me complaining in this channel or other channels

plush burrow
#

I can help if you want me to

flat harbor
#

openai has lost it's way

#

if you can help it'd be a huge help because I don't even know what I'm supposed to do at this point

plush burrow
#

ok dm me

#

the problem.

plush burrow
fading star
#

hello

wraith dagger
#

anyone noticed that the continue generating arrow icon now switched to a circular reload icon?

visual bloom
#

Yes. Feels strange. But according to articles I have been reading, tomorrow they might be announcing some interface (read webpage) changes. Might be related to that.

inland heron
regal portal
#

kind of wish there was an option to turn the continued generating feature off and let the ChatGPT continue generating. Perhaps they can add a stop and resume button as an alternative.

#

That is manually activated.

inland heron
#

Send feedback, maybe they implement it. Maybe the community (non-Discord) website has the option to suggest features..

dense inlet
#

whats the popular forecast of what will be announced monday?

glass fulcrum
elfin birch
#

🕐

hardy lichen
#

Can someone please build an app using the twitter, linkedin, instagram, facebook, tiktok *insertMediaPlatformHere Apis with or without AI so we can automate the process of blocking celebrities. Thanks in advance #celebrityblocklist

glass fulcrum
dense halo
hybrid violet
#

I think a big part about the "Spring Updates" are just that search will become a naturally called tool at model inference.

shrewd yarrow
#

Devs gonna fix the laggyness in gpt4 anytime soon or, am I gonna have to keep wasting my big 40/3hrs by starting new chats just so it doesnt lag out and break everything.

sacred aspen
#

any idea if "read aloud" will have an option for 2x speed? sad the API has this but not chatgpt

clear canyon
#

Has anyone started running into 429 errors? It seems like the switch to project keys has messed up the billing references.

shrewd yarrow
sacred aspen
brisk path
drowsy slate
#

will chatgpt be down during that time frame?

elfin birch
drowsy slate
#

for sure thanks

dense halo
#

Why is it soo hard to connect to voice

indigo sage
#

has anyone noticed that chatgpt is starting to return unusual typos. Like I'm passing code with "ExportConversation" and it returns "ExportCompiler" as the name...like completely different second word. Never noticed that happening before but it happened 5 times to me already today on accross accounts.

blazing belfry
#

I'm not seeing any typos but I cannot get it to remember anything. Statements just mentioned earlier in the conversation.

knotty crypt
#

hey guys, when I send a message in ChatGPT and it sets a new memory update with the response, what happen if I edit my message and re-send it to ChatGPT? Will that previously memory created be disconsidered?

prime swallow
prime swallow
dense halo
#

Why does ChatGPT keep saying "unusual activity has been detected on your device"?

daring grotto
#

why am i getting unusual activity has been detected this is very strange

eternal spoke
#

Been happening since yesterday for me

honest ledge
#

I'm trying to get ChatGPT to remember various details about a character I'm working on, and initially it went well, but as I try to add more details about the character's appearance, personality, history, and quirks, the new information is overriding what I told it earlier until now it doesn't know half of what I told it. How do I get ChatGPT to create a new memory rather than to update/override an existing one?

eternal rockBOT
regal portal
honest ledge
regal portal
#

oh you're talking about the gpt4 memory feature?

#

try asking it "can you save a new memory with the following information?

inland heron
#

being explicit has its benefits sometimes

twilit vigil
#

eemmm chatgpt makes typos now? xD
it just keeps getting dumber dumber

molten walrus
#

is there any way to open chatgpt side by side, 4 for hard stuff in one, 3.5 in the other for easy stuff?

twilit vigil
#

instead of EventPriority it wrote EventDriority

#

with a D

honest flower
#

anyone else notice it got kinda dummer? sometimes it seems to go back to another previous topic instead of focusing on the current one randomly

inland heron
mild plover
#

GPT forget's and remixes his memories at the same rate I do, so we're both lost.

inland heron
inland heron
mild plover
#

Nope. Isn't that their one job? to not change the memory :

honest flower
honest flower
lavish hemlock
#

Does archiving all your old chats help w/ ChatGPT speed?

#

Currently it's moving like a snail on a chat that I've been using for a few weeks now.

jaunty robin
#

its doing it on one I just created tonight, so its 'not operating well'

#

its also got a case of the stupids in as much as I said 'can you change this json' and it wrote out 3 lines and proudly said it was done

inland relic
#

has anyone used the ChatGPT Team workspaces? I have invited users and they have accepted invites but they have not been added to the workspace yet - is there always a long delay here?

visual nova
#

it was instant when i did it

#

I only have a 2 person workspace though

inland relic
#

we had 3 people - adding the 4th was instant - but the other 17 invites I sent are still pending even though they accepted hours ago

visual nova
#

Hopefully with that big of a team youll get help from support

inland relic
#

no - says 3 day TAT on support in the help chat

#

I wonder if things are just very busy today with the impending announcement

#

and we are stuck in a queue to be activated

static fractal
#

how does chatgpt4 perform with uploading research papers? I'm doing some lit review in an area im a bit unfamiliar with and some of the equations are spinning my head at times

I've read mixed info that chatgpt can struggle with solving equations, but what about breaking down a series of equations from a paper and explaining them? rather not pull the $20/mo trigger if it's not worth it lol

storm jasper
#

open ai update

inland relic
static fractal
sleek birch
small jay
#

Sorry if this has been brought up, but is anyone else getting hit with painfully slow responses lately? Asked a coworker, he said he's also seeing the same.

New chats are fast, but as the conversation goes on, the responses are painfully slow (huge lag between the messages), and just generally unusable.

honest ledge
young basin
#

5 minutes gang

lavish bronze
little storm
#

Magic is in the air ✨🧙‍♂️🎩🪄🔮

noble gull
#

they are late!

signal apex
#

desktop version!!

high sphinx
#

freee

hardy sigil
#

GPT 4 FREE FOR ALLL OMGG YAYY

#

IM SO GLAD also loved the name GPT 4o meaning gpt for all

#

WAOHHHAOOOO

silver minnow
#

I hope for Europoor, too.

zealous canopy
#

So if GPT 4o is going to be free then what is the point of paying?

toxic gulch
#

will gpts also be free?

high sphinx
#

is gpt4-v also free then?

inland heron
#

tl;dr what's the update?

zealous canopy
#

Its live right now.

high sphinx
#

desktop app and free gp4

silver minnow
#

GPT-4o multimodal free for all.

inland heron
#

free GPT4?

signal apex
#

we are only 7 minutes into the presentation lol, don't know everything yet

strong ore
#

So the store is public?

storm jasper
#

wait so gpt4o is free and gpt4 is paid?

silver minnow
#

Payed users will get GPT-5 early access (just guessing)

zealous canopy
inland heron
#

what is GPT-4o?

high sphinx
inland heron
#

is that the one from the few days ago?

#

since it became smarter?

toxic gulch
inland heron
#

I mean the opposite

silver minnow
delicate hamlet
#

capacity difference

marble tree
zealous canopy
#

4o is probably the same as GPT2-chatbot yeah.

silver minnow
#

GPT-4o also for API.

toxic gulch
#

damm and really better then gpt4 turbo

silver minnow
#

2x faster, 50x cheaper over API

little storm
#

Wonder what the o stands for ?

inland heron
#

that's it? just API?

toxic gulch
#

no its not just a api its also for personal use

urban cipher
#

So the plus subscription is basically useless?

inland heron
#

well personal use can be also over API. You mean chatgpt?

olive cargo
silver minnow
zealous canopy
toxic gulch
little storm
#

5x the capacity for paid users

little storm
urban cipher
tranquil flame
#

where event?

little storm
toxic gulch
tranquil flame
#

oh

rustic fiber
silver minnow
echo plinth
toxic gulch
#

so gpt4o can now understand emotions and also react in emotions

#

sounds like it yeah

little storm
toxic gulch
#

ez homework

finite stream
#

i see it 😂

zealous socket
#

Wow haha

neat elk
#

When’s it coming out?

toxic gulch
#

in the following weeks i think

little storm
#

I think this is what Altman might have been describing as feeling like magic ✨

zealous canopy
#

the robot laughing with him kinda creeps me out.

finite stream
#

😅

honest ledge
#

😬 ChatGPT with that sugary personality and tone of voice is really unsettling... I really hope that can be altered

deep roost
#

will there be a linux version or the desktop app?

zealous canopy
#

That would be interesting. I'd make chat GPT act like a British butler.

little storm
#

Yes new voices is something I've been wanting for a while

#

More chilled out voices would be great

visual nova
#

i wonder if the api version can do this video to text stuff in real time

toxic gulch
finite stream
floral yew
visual nova
#

that would make so many new apps possible if it can

little storm
#

For many new voices

hardy sigil
#

is the voice basd on gpt4o

toxic gulch
#

i think they already have a ai themself to generate voices

little storm
hardy sigil
#

so gpt4 turbo will be paid?

toxic gulch
zealous canopy
#

XD "you look like a wooden surface"

slim plover
#

saw right into the soul with that one

floral yew
#

daymm, i never blushed a girl

#

i might now

raw igloo
#

lol that demo was hard to watch, but I lvoe the potential

little storm
plucky basalt
#

It's really starting to feel like "Her"

zealous canopy
#

uncanny valley AI GF lets go

little storm
#

Not "that looks like a wooden surface" "YOU" look like

main heron
#

this will be so much better, especially qith being able to share parts of code

finite stream
unkempt coyote
#

desktop app sunds cool if they add option to store chat history locally 🙂

floral yew
#

no more lonely men out there /s

tough bramble
#

That was insane. I'll remain skeptical till I try it myself.

visual nova
#

i wouldnt how hard it would be to turn the new model into like a teamviewer where it can actually press keys on your keyboard and just use your computer to do stuff.

vernal isle
#

So everyone gets gpt4, what will be the reason to keep a subscription then?

little storm
#

Looks and sounds pretty great!

visual nova
#

like that might actually be possible now, i wouldnt expect it to work great, but why couldnt it actually use your computer now?

tough bramble
finite stream
hot dragon
#

so when do we get the desktop app

toxic gulch
#

probably in the following weeks

tough bramble
#

This is gonna be some long weeks...

wet sorrel
#

yo chat, did I miss anything

unkempt coyote
deep roost
#

not impressed... I was hoping for something on Q* and AGI, but the announcement was merely about making the users the product, just like social media did long before
by lowering the entry barrier with new modalities and opening up to free users. Hot take: Smells like OpenAI need more training data to achieve capabale agentic AGI.

storm jasper
#

when will the mac app be out?

toxic gulch
tough bramble
finite stream
wet sorrel
hearty ice
#

it's crazy that gpt-4o will be free, what are the strings attached?

zealous canopy
main heron
vernal isle
wet sorrel
hearty ice
eternal rockBOT
little storm
sullen herald
#

i'd want my paid subscription to translate to tokens for the API

unkempt coyote
#

so shall we prepare to not responding servers 😄

honest ledge
#

Honestly, I'd like my ChatGPT to have the voice more along the lines of Alan Rickman, Patrick Stewart, Liam Neeson, or Sam Elliot, or if it's a younger sounding voice, Taylor Lautner or Rami Malek

zealous canopy
tough bramble
# wet sorrel I forgot there was one

Real time conversation without delay, even speaking while it's still speaking, and speaking while showing it a live video. Very expressive voices that it can control, eg. if you ask it to sound sad, or sound like a robot. Can detect your breathing too apparently.

flint zenith
#

Well, correct me if I am wrong. but they never mentioned that DALLE and browsing will be on the free version only the model (plus vision and voice)

vale compass
#

I love AI

spice star
toxic gulch
#

it will only be an desktop app for macos?

vernal isle
#

imagine talking to chatgpt with the voice of morgan freeman

main heron
#

i am wondering how well it will do translating something like a live soccer match in Spanish, to english

wet sorrel
tough bramble
toxic gulch
long barn
#

Is it me or does the video on the website have no audio?

zealous canopy
spice star
novel scroll
#

Desktop app only for mac is trash

wet sorrel
#

cuz 20 bucks per month just for more prompts per hour is a bit high imo

toxic gulch
keen meadow
#

I was expecting a "but wait, there's more" moment to justify being a paid user. I guess the usage limit is quite stingy if they think increased usage is enough reason to fork out $20 pm.

tough bramble
#

Plus will have up to 5x the limit of free.

molten tundra
#

Guessing 4o is rolling out and isnt immediatly avaiable

ancient bloom
#

where can i see diffrences between both gpts? (i mean ofc 3.5 and 4)

zealous canopy
floral yew
main heron
obtuse wigeon
floral yew
#

considering majority is on windows

novel scroll
long barn
vagrant walrus
#

Is chatGPT-4o sentient?

sacred aspen
molten tundra
#

So if there is data analysis does that mean I can upload multiple audio files for transcription? Or not to that level yet

ivory hazel
#

How do i get the new gpt, is it on mobile or browser only?

ivory hazel
sacred aspen
#

how can I see if I am using ChatGPT-4o?

zealous canopy
low galleon
#

is GPT-4o still rolling out?

vagrant walrus
low galleon
zealous canopy
toxic gulch
spice star
harsh pilot
#

I don't have it, either, I don't think.

low galleon
silver minnow
#

VoiceMode for pro and only MacOS.

vernal isle
#

4o is available in the playground

sacred aspen
silent ingot
#

gpt mac app available today?

south latch
#

is chatgpt 4o just gpt 4 but better?

toxic gulch
#

do you need to pay for playground?

dire rune
#

What is 4o?

noble gull
#

so when is desktop app avalaible?

swift delta
raw igloo
#

"feels like magic" my ***

toxic gulch
dire rune
#

Why 4o and not 5?

ancient bloom
#

where can i see diffrences between both gpts? (i mean ofc 3.5 and 4)

glass harbor
#

Is the model available yet? I can't find it in my free account.

main heron
toxic gulch
raven portal
#

no windows app??

wet sorrel
#

will we still get GPT-5 within this year? that's what i'm waiting for

ancient bloom
deep roost
floral yew
#

man why dont Americans protest Apple monopoly

dense halo
#

this is crazy impactful. much easier to interact with gpt via voice than typing

main heron
#

this chatis a cluster... people want to be spoonfed stuff.

hardy scarab
#

Whoa isn't that exciting? I have to say though, I'm a little skeptical on whether it's going to be as good as the paid GPT-4

deep roost
toxic gulch
#

finally a actual useful voice assistant

dense halo
ivory hazel
#

i have a paid account and i still dont have the new gpt HOW DO I GET IT

remote smelt
#

Anyone has access to the gt4o in ChatGPT yet?

high minnow
#

anyone know if the desktop app is mac only?

sacred aspen
#

is the desktop app a closed beta?

wind merlin
spice star
noble gull
#

So when is desktop app for windows avaliable ?

wet sorrel
dense halo
honest ledge
#

So looking at the 4o release page, it says 4o is half the price and 5x the rate limit? so does that mean GPT plus will go from being $20/mo to $10/mo and go from 40 messages every 3 hours to 200 messages every 3 hours?

novel scroll
sacred aspen
wet sorrel
toxic gulch
dense halo
#

not yet

swift delta
honest ledge
ivory hazel
#

i have a paid account and i still dont have the new gpt HOW DO I GET IT

broken sierra
valid cobalt
#

Do you have new gpt?

tardy totem
#

It sounds like plus users will get it before Teams... :/

dense halo
#

New gpt via api only atm, next weeks via chat

copper cipher
main heron
#

This new GPT is amazing... glad i got it in my acct

inland relic
#

does anyone know if free users can create custom GPTs - the wording says "discover" and access the GPT store

toxic gulch
warm gull
broken sierra
ivory hazel
#

Alright when is the GPT-Matrix Singularity version dropping

inland relic
wooden hollow
#

app link?

neon bane
#

Why can't I use 4o yet

knotty willow
#

what the hell

hot dragon
#

says on their website the windows app will launch later this year, but MacOS get its today

knotty willow
#

when do we get it lmao

zealous canopy
vernal isle
analog wraith
#

how can i access GPT 4 for free? it still shows 3.5 on my end

knotty willow
cold yew
#

What about Team?!

Why would Enterprise not be included in priority access or mentioned in regards to limits?

What is my company paying 10$ extra per seat for?

vernal isle
analog wraith
#

perhaps it -will take time to roll over multiple countries at once?

neon bane
vernal isle
#

It's only available in the playground right now ...

inland relic
#

Mac only for the desktop app for now - Windows to follow it says

opaque bronze
#

The lady that was presenting, what's her name?

honest ledge
#

" Plus and Enterprise users will get to experience voice and images in the next two weeks. We’re excited to roll out these capabilities to other groups of users, including developers, soon after."

warm gull
gloomy ridge
#

I'm trying to explain to v4 that it is about to be replaced. It doesn't seem particularily worried 😄

copper flare
crimson sapphire
#

The "Interview prep" demo is really quite incredible

wintry spindle
#

what are these video capabilities?

tight island
vernal isle
#

Oh boy, can you imagine all the folk that will show chatgpt their junk and ask it if its big enough 🤣

warm gull
#

ok back to feeling grass - nothing more to see here

crude pulsar
analog wraith
tight island
#

sadly yeah

spice star
wintry spindle
main heron
#

wow, the speed with the new gpt-4o is awesome

wintry spindle
#

it just said a very all inclusive answer to me

copper cipher
tight island
#

damn i feel like i just got the middle finger. as a paying user.

valid cobalt
#

Can't find it on playground, where is it?

molten tundra
#

Looks like all those companies that made ai girlfriends/boyfriends, might soon be out of buisness

zealous canopy
fossil oyster
#

no limits more for gpt 4?

main heron
spice star
timid quiver
still rain
#

On playground, GPT4o just changes to 3.5

zealous canopy
main heron
fossil oyster
#

is the chat now limetless?

copper cipher
spice star
#

early versions of web browsing were pretty rough as the model had to navigate a web page (spoiler alert: it sucks at navigating webpages)

errant glen
#

What are the benefits for paid users?

ember scarab
#

Will 4o be able to do Vision and DALLE at the same time?

valid cobalt
shrewd falcon
#

Weeeeeeeeeeeee

bronze oak
#

Did I just get mentioned for no reason? I don't see any option to switch model for FREE

spice star
drowsy tinsel
#

gpt4o in playground doesn't support vision?

blazing hill
remote slate
molten tundra
#

Man with 232 milliseconds response time imagine the rap battels between a rapper and 4o

blazing hill
drowsy tinsel
#

oh u have to select gpt-4o-2024-05-13

bronze oak
broken sierra
blazing hill
wary pulsar
#

all this if the output context length is not improved its pointless i will stick with claude opus.

hybrid totem
#

I can find 4o anywhere

still rain
remote slate
sinful ember
#

why do I see people complaining about lower value premium, I thought people wanted OpenAI to be a public good focused not profit focused company 🙄

drowsy tinsel
#

playground only for now i see

broken sierra
junior lion
#

Do I need the plus thing for gpt-4o

ember scarab
# spice star it always could, not sure what you mean by that

For example, I could never give an image to a GPT model and get returned a variation/DALLE interpretation of the image. For example, if I took a photo of a tree and told it to return a painting of the same type of tree, I couldn't do it with 1 API call. I had to send the image to Vision, get a description of the image, send the description to DALLE.

broken sierra
zealous canopy
young obsidian
#

The "Ohh stop it!" was damn amazing

bronze oak
molten tundra
young obsidian
#

When will the new UI update and app release?

junior lion
#

I still don't see the new stuff here

olive cargo
neon bane
#

Oh my God I was to use 4o right now for those image generating capabilities. It's amazing

opal fable
#

No windows app soon 😿

zealous canopy
main heron
# valid cobalt Redirects to 16k 3.5

i cannot share an image here to prove it... sorry you arent seeing it yet... do you have a paid account for playground? (like, you've paid to use the api? maybe thats why, if not)

fluid jackal
#

Probably region locked there as well

spice star
young obsidian
magic otter
#

Where do you get access to the playground?

olive cargo
onyx garden
#

This makes me very excited for WWDC

young obsidian
blazing hill
#

It keeps stopping in the middle of writing

spice star
cyan igloo
#

"Users on the Free tier will be defaulted to GPT-4o with a limit on the number of messages they can send using GPT-4o, which will vary based on current usage and demand. When unavailable, Free tier users will be switched back to GPT-3.5."
It's likely that it's just too high in demand right now for free users to even see it

young obsidian
#

The biggest thing for me in this update is the VOICE UPDATE!! It seems incredible!

still rain
young obsidian
#

I hope it will be better at pronouncing local languages as well

inland relic
#

incredible - but will I really talk to it?

timid quiver
#

Anyone has access to voice yet?

magic otter
vernal isle
inland relic
#

maybe that is an introvert's question

young obsidian
#

What will be the usage caps?

molten tundra
#

So is gpt-3.5-turbo-16k-0613 the right one?

little storm
storm portal
granite arch
#

Voice update is good for the lazies like me

main heron
cyan igloo
young obsidian
#

I have Plus do I unsubscribe now since they basically removed half the premium 😭🤣

zealous canopy
little storm
#

Quite useful and you talk differently than how you would interact with it through text

bronze oak
molten tundra
#

Ok well that's all I am seeing and I'm a plus user 😦

#

So I guess I can't even access 4 or 4 turbo in playground atm

granite arch
olive cargo
#

I checked playground. I saw 4o

blazing hill
#

Has anyone generate images with this ? Keeps saying as an ai text based model I.....

bronze oak
young obsidian
#

Will voice be available for 3.5 as well though? For free users? I can use it on 3.5 because I have premium, but idk if it will be the same for free users.

broken sierra
vernal isle
#

Maybe you have to be a paying API user to be able to see it idk

olive cargo
young obsidian
bronze oak
granite arch
#

It'll be like Discord...custom emoji's for Plus users kekw2

olive cargo
young obsidian
#

I was thinking of switching to copilot. Nah

#

Copilot pro is kinda overhyped imo

#

After trying it

molten tundra
#

Anything I need to do special to get 4, 4 turbo, or 4o on my playground I only seem to have 3.5

fluid jackal
#

Copilot is dog, even the built in VS plugin

molten tundra
#

With the paid plan

broken sierra
zealous canopy
celest quartz
#

We get a free taste of gpt image recognition? Even for free users? Nice

blazing hill
#

Copilot tends to avoid subject at all cost

vernal isle
molten tundra
cyan minnow
#

anyone know if we'll be able to upload videos for analysis with gpt-4o?

molten tundra
zealous canopy
young obsidian
#

Copilot is great for last minute homework as it cuts to the chase unlike chatgpt where you have to remove like 70% of words because they're just meant to fill inn the spaces between words.

granite arch
#

Video too huh...hmm...

wooden hollow
#

ok it's fast .

zealous canopy
#

I thought the API was a "pay as you go" model?

young obsidian
#

I pay 25😭

wooden hollow
grand dome
#

why was chat gpt voice mode never on desktop?

main heron
hardy scarab
#

Looking at the demos on the website is really showing me how far we've advanced with this. I cannot believe it's going to be free

molten tundra
hybrid spindle
#

is 4o currently available for all paid users?

flint zenith
#

"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."

young obsidian
#

Nope

bronze oak
rugged wing
#

am just getting 3.5 not getting 4o

main heron
rocky condor
#

how do u use gpt4o

molten tundra
#

I found this though:

"Is the ChatGPT API included in the ChatGPT Plus subscription?
No, OpenAI APIs and ChatGPT Plus subscriptions are billed separately. The API has its own pricing, which can be found at https://openai.com/pricing. The ChatGPT Plus subscription covers usage on chat.openai.com only and costs $20/month."

young obsidian
zealous canopy
#

that did it

#

You have to pay.

neon bane
#

Can't have it generate images in playground :/

wooden hollow
#

it feels fast like early gpt 4.
i hope jensen can supply the demand.

wooden hollow
molten tundra
young obsidian
#

Image generation of Dall E is far behind the competition

main heron
young obsidian
#

Adobe is WAY better

knotty willow
#

can we acces the thing yet?

blazing hill
knotty willow
#

anything i wanna try out stuff

zealous canopy
molten tundra
eternal rockBOT
#
This is a Safe for Work Discord Server

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.

Posting NSFW content will result in a permanent ban.
wooden hollow
young obsidian
worthy dragon
#

The robotic voice and singing, oh for now I can only imagine my Jarvis, oh my Jarvis... 😁

young obsidian
broken lotus
#

Anyone know what the context window is?

worthy dragon
#

😆

young obsidian
#

I will bore an airpod into my ear and get jarvis😅

granite arch
#

Hmm...no options for me in the US for the 4o

#

logged out and everything

worthy dragon
vernal isle
median horizon
main heron
granite arch
#

Must be rolling out slowly

young obsidian
#

Schitzo simulator in the GPTs library

worthy dragon
granite arch
ancient bloom
#

is vision, memory, browse etc available for free?

broken lotus
young obsidian
main heron
broken lotus
#

I was asking if someone knows the context window size

young obsidian
ancient bloom
worthy dragon
young obsidian
ancient bloom
#

welp, still great work

young obsidian
#

Very unlikely

molten tundra
#

Yep payment to the api got me 4, 4 turbo and 4o

young obsidian
#

Because you pay content creators on the GPT store

#

and if no payment then no money to creators

worthy dragon
#

If 4o will be free, then what will paid users soon get ? 4.5/5o ?

young obsidian
#

And some better features

main heron
young obsidian
#

In the future

wet sorrel
#

so, that new model is also faster in textgen?

worthy dragon
#

yea, they can run on that for some time, but they will have to introduce new models for plus users at some point ig

hardy fable
echo berry
#

Anyone not seeing GPT4o in ChatGPT/API? I don't see it anywhere

young obsidian
#

I love how consumer friendly OpenAI are though

worthy dragon
granite arch
#

I don't understand all these words like 128K context shibkek

hardy fable
worthy dragon
echo berry
sly geode
#

did they rolled the update like right now ?

young obsidian
#

Only available on macOS

Apple partnership confirmed?

young obsidian
summer tide
#

I don't see this chat for 4 y only see 3.5

young obsidian
knotty willow
#

neither audio input just regular image input

errant thorn
#

its available in playground

knotty willow
echo berry
mortal salmon
#

when will 4o go live?

vernal isle
candid anvil
#

Why is the "ChatGPT desktop app" only available for macOS and not also for Windows?

young obsidian
echo berry
inland heron
#

though it's useless to me as I don't currently use API :/ and I sure as hell ain't paying their high fees to unit-test 😄

#

but it will come in handy soon I hope

sly geode
valid cobalt
vernal isle
inland heron
spice star
#

where can you download the chatgpt desktop app?

vernal isle
#

That's true

candid anvil
mild cave
sly geode
#

Goodbye google you are done

vernal isle
#

Google's new models show promice in all honesty

slender adder
#

Hi where I can find GPT for mac ?

vernal isle
#

especially the 2m context limit

inland heron
young obsidian
spice star
young obsidian
#

They already said they wanted to integrate it into apple services

hybrid spindle
#

how do I know if I am using 4o?

nimble snow
inland heron
#

oh great, errors gallore again

spice star
#

google's models are good on paper, but far from ready for real-world use imo

vernal isle
#

interesting, 1 message to 4o api cost around less than 1 cent

rare wave
#

is gpt 4o really cheaper than gpt 4 turbo? why? if it's better

vernal isle
young obsidian
spice star
young obsidian
#

it's just trained longer with more data

rare wave
#

this is absolutely amazing

rich surge
#

It's better

spice star
#

huh interesting

zealous canopy
#

Well now that ive gotten to mess around with GPT 4o in the playground, I am not very hyped?
Its still on the level of some open source models imo.
Is this really that GPT2 Chatbot we saw on the ranking site?
I guess ill wait to fully judge for when it gets to the real app. A system prompt can make a heck of a difference.

urban cipher
#

Anyone already having acces to GPT 4o or will it take a while?

knotty willow
#

What's the difference between
gpt-4o
and
gpt-4o-2024-05-13

on the api?

rare wave
#

does anyone have it in chatgpt? or only playground rn?

young obsidian
#

It's a little bit better yeah. But the features 🥰

inland heron
spice star
#

so there is no difference, but gpt-4o will recieve updates

inland heron
#

and better vision

vernal isle
vague ice
#

GPT-4O really what kind of name is that.
No improvement no leap forward,
I am disappointed.
And free for everyone??
Why am I paying!

rich surge
#

Better

rare wave
main heron