chatgpt-discussions

[normal quarry]

they are saying they are bringing Deep Research to Plus and Teams soon but have provided no date

[cold sequoia]

Tnx

[rugged igloo]

so far these are all the papers i have done so far with deep search

[normal quarry]

you have been busy 😄

[rugged igloo]

yea trying to see what works and doesnt work for prompts

for gertting a good output

[gaunt magnet]

does ctrl+shift+i not work anymore to go into personlization?

nvm works now

[brave valve]

What are you first impressions?

[rugged igloo]

its good if you have a really detailed multi step prompt that you want it to exactly do

[brave valve]

It seems openAI is finally catching up with godmode v1 that was released 2 years ago...

[rugged igloo]

it feels alot like o1 pro you need to be very specific on what you want it to achive

if you want an mla paper you give it that an apa paper you give that ect ect

[brave valve]

Garbage in, garbage out, still

These agents are the future, 👍

[cloud pewter]

O1 pro is the only thing you’re missing out as a plus peasant

[normal quarry]

Deep Research
20 second Sora Videos
Operator
o1 Pro

Pro users don't have to deal with this -> #chatgpt-discussions message

[cloud pewter]

thank you captain obvious

that surely is very useful!

[dense halo]

he pays 300 aud for pro

every month

[normal quarry]

Pro is worth every cent

that's why so many people have it

[cloud pewter]

that’s $184 usd

[normal quarry]

yes, subscribing on the Apple iOS App Store works out cheaper than using credit card

[eternal rock]

Get notified for server events!

Visit id:customize to pick up the <@&1261377106890199132> role.

[elfin lava]

The only person here that always flex about pro would be Robert , and no I’m not complaining 😆

[leaden ingot]

still nothing on deep research availability in the EU …. crap

[whole roost]

I wish this list was true for Europe

[normal quarry]

yeah, EU is getting a rough time of it

[whole roost]

I'm not sure about operator, didn't follow that topic if I should have it but if it should be on pro, I don't think I have it

[normal quarry]

Operator is US only for now, I don't have it

[whole roost]

Yeah, that leaves us with o1 pro and unlimited messages
And 128k tokens, that's also valuable I think

[normal quarry]

I initially subscribed to Pro for unlimited o1, everything else since has been a bonus

[whole roost]

I subscribed for unlimited 4o

[sinful spear]

I'm thinking of getting a VPN, will that help? (not if it's against ToS)

[normal quarry]

Using a VPN to bypass regional restrictions is against the ToS and can lead to your account being banned

[sinful spear]

Then I definitely won't use a VPN, it's currently just a very bad customer experience watching continuous announcements of things you don't get when you're paying £200/month

[whole roost]

I didn't have any interest in stuff like sora, just think that Europe should have a lower subscription as I'm paying for half of the things that I don't have

But I'm still shooting around 400 messages per 3 hours at 4o so it's worth it and I grew tired of the o1 restrictions when creating my scripts

[cloud pewter]

they really really need a middle plan for only unlimited 4o and extended access to o series
Rn it’s either go broke or forced to ration

[dense halo]

Damn must be rich

[cloud pewter]

And Europe is another problem yes

[floral urchin]

We are a problem yes

[dense halo]

yes

[floral urchin]

Wait i'm uk nvm

😿

[whole roost]

I remember how long it took till we got memories

[dense halo]

Ex EU bruhhh

[floral urchin]

Stupidest thing to ever happen

[whole roost]

Begone deserter

[cloud pewter]

Yeet away legacy 4 and give plus unlimited would be dope

[dense halo]

Id not enjoy unlimited 4o

I barely use that Sht

[floral urchin]

use it more

[dense halo]

[cloud pewter]

use it more that you always hit the limit

[dense halo]

why

4o got so bad

[floral urchin]

did you walk 420km yet

[cloud pewter]

Also the price of plus isn’t being raised

[cloud pewter]

nope 😭

[floral urchin]

glhf

9am and i've had about 2 litres of coffee rip

[dense halo]

make a vlog u walking 420km

and talking to gpt

[eternal rock]

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.

[floral urchin]

@dense halo high key wanna get in my car and drive it to south of spain and twitch up petrol money

[cloud pewter]

ok

[dense halo]

dont forget to have a therapy session with avm

[floral urchin]

WHO

[cloud pewter]

I’m using this as my server profile banner

[floral urchin]

There's nothing wrong with that image

[whole roost]

Mine is amazing, I laugh a lot, but maybe depends what you do with it

[dense halo]

What are your tasks?

[floral urchin]

My 4o is gfuelled

rains emoji's and heckin's. I suppose that's more an inditement on me

[whole roost]

I don't know how the rules here are but trigger warning: RP death

[cloud pewter]

Well

[floral urchin]

Traineportation

[normal quarry]

these really belong in #sora-discussions as it's not ChatGPT related

[floral urchin]

ok

[cloud pewter]

ok

[floral urchin]

This coffee doesn't belong in my body but i'm drinking it robert

[dense halo]

Do you want to discuss the prompt you made with gpt for sora?

which gpt did you use

[floral urchin]

caffeine-4o

[cloud pewter]

Just the normal one

[dense halo]

so you let gpt write your prompts?

[floral urchin]

Come with me, to a world, of pure where you can use your own imagination

[cloud pewter]

Yeah

[floral urchin]

I can't afford oculus so I'm making one out of chocolate

Choculus Gift

[dense halo]

which gpt do you use for that task?

[floral urchin]

A heat gun and bars of chocolate

[dense halo]

Thats another word for o1

[cloud pewter]

i can’t look at this picture properly anymore

[floral urchin]

Brain Warp

[whole roost]

o1 has the big disadvantage that it cannot read files

[floral urchin]

wth

[whole roost]

And it's kinda slow

[floral urchin]

Souls character

Tap X to board the train

[cloud pewter]

I can paste whatever long text into o1 and it still accepts them

even the ones from dr

but files indeed would be better

[paper hornet]

is chatgpt super slow rn?

[whole roost]

Files because of projects like I have 25 files in my project

I hope that they will add a function that you can edit uploaded files in canvas so that you can edit them on the page directly

It is a huge problem for me because I upload them on my computer and I don't have them on my phone or mac so I only can edit them on my computer

It is really inconvenient

And canvas for the app and mobile web

I don't know why they always half-ass features

Why can't I access previous swipes on the app, I just don't understand why they don't add basic functionality

[twilit jetty]

GPT just generated me a method that obviously causes an infinite loop. 😂

[cloud pewter]

their priorities are wrong

[whole roost]

Character AI has added a feature where you can blacklist individual words and I would need that so much for GPT

[cloud pewter]

if openai ever added this feature there will be default banned words tbh

[whole roost]

I was inclined to ask which but that isn't a good idea

[cerulean pasture]

what is caas Containers-as-a-service? These are the chrome management policies applied in the Operator browser

[frozen birch]

openai are doing aprils fools earlier this year?

no sora for pro users in eu - ok, i can understand
no operator for pro users in eu - ok,, im bit mad
deep research - Nope for pro users in eu

..... ITS A FREAKING JOKE?

[cerulean pasture]

dude, you still get a better deal, I'd trade AI for a functioning govt

[frozen birch]

sorry mate, what you mean with functioning govt?

[cerulean pasture]

yall have a broken govt too?

[frozen birch]

i mean.... functioning government should be a standard when you pay taxes each time

[cerulean pasture]

How do the geographic lockouts work? You should never circumvent them no matter how easy it is

and are you paying $200 in whatever your local currency is for PRO?

[frozen birch]

yea, its sad since im paying in dollars, im from Poland, so okay, i can use vpn for avoid that, but srsly? i dont want to make exorcisms like that 😄

[cerulean pasture]

damn euro194.67 for none of the cool stuff sucks

[cerulean pasture]

I would be cautious about mentioning that, its super easy to get auto-banned from the discord.

[frozen birch]

seriously? 😮

sigh

[cerulean pasture]

completely unrelated: Split tunneling is a cool concept. You can make most of your traffic go over your regular internet, but then you can make certain destination IP addresses take a VPN tunnel

Maybe Advanced Voice Mode? IDK, I wouldnt pay $200 if I was in EU, but im also not rich

[frozen birch]

ive studied ToS and no words about vpns

so i cant understand why i can be autobanned?

[cerulean pasture]

well they must not care, maybe the EU's rules are written such that if you did something like VPN to put your end point in the USA then its no longer a liability to OAI?

[cerulean pasture]

I wasnt saying you would be, just that the bar to getting a temporary ban here is very low

[frozen birch]

idk, but i cant understand one you know?
ok, sora and operator can be strictly censored by UE law

but deep search? crazy

[whole roost]

I am pro User in EU and I pay because I like the functionality and features GPT has and I use it all the time like all the time

[frozen birch]

oh hi Vishanka, nice to see ya again 😄

[whole roost]

Hello yes I will mostly be here this time around because I'm at work and my work isn't very complex

[inland heron]

o3-mini a bit slow for anyone else? Just checking if it's just me or in general

[whole roost]

I actually have used different services before and just have the best result on GPT and the features compliment what I want to do well

[frozen birch]

ohhh noce to hear 🙂
@whole roost mate what you think about that situation with deep research? since i must say that im very mad

[eternal rock]

Join the Sora Weekly Theme!

Submitted your entry yet? Don’t miss out—there’s a new theme every week, and we’re excited to see what you create!

[whole roost]

I haven't really monitored any new developments with this other than that EU doesn't have it yet
Are we never going to get it?

[frozen birch]

dunno, sent email to support with question about that question

when ill get answer ill post it there

[whole roost]

Ah so no news about it? Then we just have to wait honestly I think it will arrive in the next months

[frozen birch]

i know one @whole roost - i feel really as a customer from Africa or any other country with limits, no offence for Africa of couse

[frozen birch]

i hope so

[whole roost]

Memories were said to never arrive to Europe and after a few months they did

I wouldn't do very important things with deep research anyway honestly

[frozen birch]

oh, for me it can be really helpful for cybersecurity findings, 0days etc

[whole roost]

To me these features are just a gimmick, I would need mainly improvements to 4o and canvas and bug fixes

Hire capabilities to follow instructions and retaining instructions and longer conversations, support for longer conversations so that the website won't slow down

Don't mind the odd words I'm using Google speech to text

[frozen birch]

ohhh i understsamd ya mate, we have so other needs from pro 🙂

[strange valley]

why gpt extremely slow today

[rare shore]

i have quesiton, i am a plus user and i dont have deep search available on my account
while other people who just opned a free account next to me at work has that option, i think it's a bit unfair.;.
is ther esomeone i can talk to on this ?

[elfin lava]

Deep research is for pro user tho for now

Unless you got confused with “reasoning” button

[rare shore]

maybe your right

i may hav egot confused

[elfin lava]

Free user get to try o3 mini hence the reasoning button appear 😃

[rare shore]

i can use o3 mini.. but i dont have that reasoning button

[elfin lava]

Are you on ios/ android app

Mine doesn’t show up too

[cloud pewter]

reasoning button ain’t the same as dr

that button is just used for free users

[rare shore]

i'm on pc

[elfin lava]

But if I do this it shows up

[cloud pewter]

plus users shouldn’t have the button since it’s only some free users toggle

[winter vine]

Couldn't agree more. Stability, memory, and responsiveness are way more important than flashy features

[cloud pewter]

real

[still hound]

is GPT lagging for anyone?

[topaz panther]

Anyone here from Australia have Deep Research yet?

[winter vine]

i have lag

[winter vine]

it's weird

[cloud pewter]

lags often occur when a chat is too long

[normal quarry]

yes I have it

[winter vine]

Yeah, I noticed that

[still hound]

even on new convo, it doesn't write smoothly, but with choppy framerate/lag

[dense halo]

hes currently in thailand and spends 300 aud per month for pro

[whole roost]

Yeah pretty slow

[cloud pewter]

🔥

[cloud pewter]

try clear browser cache??

[dense halo]

well lately 4o has been very slow for me, about a month already

they made it dumber + slower

[topaz panther]

Thanks Robert.

[topaz panther]

I’m not a pro user (in any sense 😂)

[dense halo]

No problem he has pro btw

pro is for people that can afford 300 aud, just letting you know

[topaz panther]

As a physics PhD I have conducted deep research including some with early instantiations of machine learning, just nothing with ChatGPT 🧐

[cloud pewter]

they love to do that

[magic lintel]

anyone here have deep research?

[lavish vale]

why o1-pro always have to refresh the page to see actual done work, most of time its just stuck on processing it but in reality at some point itwas done already.

[cloud pewter]

many people have it rn!

hmm. it’s becoming slow for me too

[magic lintel]

do you have pro?

[cloud pewter]

nope

but I’ve seen multiple pro users in this server have it

[magic lintel]

i wonder when it will come to plus

probably 3 months

[cloud pewter]

they said soon and “in a month” but who knows

that means another $20 gone to their pockets

[magic lintel]

i cancelled my sub

[cedar thicket]

Hi, I'm not able to see my Projects in the menu, Am I Blind ?

[dense halo]

glad to hear

[calm gulch]

ChatGPT is sssssssssssssssssoooooooooooo slow today

[cloud pewter]

trueeeee

[cloud pewter]

which platform are u using? also which plan you on?

[still hound]

ok i was worried it was only me. seems everyone's GPT is slow today

[normal quarry]

the ChatGPT Desktop App is quicker

[short moss]

I want to demo workflows to my team using the ChatGPT suite of models (and outside tools if needed). Here's what I have so far:

• Research to slide presentation. Flow:
  --- Use o1 Deep Research to generate a detailed research report
  --- Use o1 to convert that report into collections of bulletpoints.
  --- Use gamma to convert those into a presentation slide complete with images.

• Learn a new language. Flow:
  --- Use o1 Deep Research to create a vocabulary of the most common words in the target language and get their English translations
  --- Format them as Anki cards
  --- Learn these words over time using n8n to text them to you on WhatsApp (you get texted the English words, you reply in the translation)

• Use O1 Deep Research for various personal research: travel itinerary, researching purchases, etc.

• Sales trainer
  --- Transcribe a sales call
  --- Use o1 to extract key parts of the call: objections, objection handling, etc.
  --- Convert this extract into a system prompt. Feed this prompt to ElevenLabs to create a mock sales call agent.
  --- Have human call this agent and practice handling the objections

Any other cool ideas?

[still hound]

I didnt even know there was one. thanks, I'll install and see

[tawny whale]

Hi, Im a bit confused and would like some guidance. Are people still making customGPTs? What are the pros and cons compared to projects? From what i understand, projects can be used with atleast the o1 models, meaning we essentially have RAG for those models?

Is the only way to get o3-mini with RAG by creating an "Assistant" and use function calling to search knowledge database?

What im after is getting a model to be able to consistenly follow a specific format for generating Desmos graphs in JSON format, based on user instructions

[whole roost]

So on custom gpts you can add actions which can perform several tasks which some people need but if you don't use actions then there is no advantage of custom gpts over projects
One more advantage could be that you can disable some features like dalle and web search which clutter the system prompt and in projects you currently cannot deactivate them

Another advantage or disadvantage depending on how you see it would be that custom gpts cannot access memories

The disadvantage of custom GPT is that you cannot select a model, that if you edit a file you have to start a new chat while projects apply the changes to your current chat

And projects are a way easier to organize because of the folders you get

I personally only use projects now and don't use my custom gpts anymore

[honest jay]

Just one small point: chats in Projects currently use 4o: https://help.openai.com/en/articles/10169521-using-projects-in-chatgpt

Please note that chats in Projects use GPT-4o, and the model cannot be changed.

[cloud pewter]

custom gpts can set up avatars and names, it’s better on a specific task like roleplaying
projects is basically a fork over customgpts

Also both tools can’t use o3 mini

[fringe barn]

I purchased credits and yet im still getting errors that i dont have enouh

for api

[whole roost]

You can use o1 in projects

[tawny whale]

Thanks all of you for the answers! So for o3-m usage the only alternative is the assistant approach? That should be possible or am i misunderstanding something? I read this on the models page:

"o3-mini is our most recent small reasoning model, providing high intelligence at the same cost and latency targets of o1-mini. o3-mini also supports key developer features, like Structured Outputs, function calling, Batch API, and more. Like other models in the o-series, it is designed to excel at science, math, and coding tasks."

Which implies that o3-mini can use function calling and one of those tools could be knowledge search in external documents, correct?

[honest jay]

Interesting! What do you see when you tap the three dots in the top-right, and then "View details"? It should list the model at the bottom

[whole roost]

You only cannot use it if you have files uploaded because o1 can't yet work with files

[eternal rock]

**`` Important Notice ``**

Be cautious of potential scams or fraudulent messages you may receive in your direct messages. The OpenAI team will never contact you via DM for any transactions.

-# Report any incidents by sending a DM to @languid valley immediately.

[honest jay]

Ahh ok, that was what I just tested on my end, you read my mind 😁

I was like "whoa loophole!!"

[whole roost]

For a short time there was a bug where it was possible to generate messages even if you had files but it just was the result that o1 didn't process the files

[honest jay]

I see! Well maybe that help doc will get updated soon, o1 definitely works, you are right! Follows project instructions and everything. Thank you for this!!

[whole roost]

You are very welcome

[honest jay]

One thing to keep in mind is that o3-mini is currently rolling out on API tiers 3-5, so you'd need access, but I think it is compatible with the other Assistants API stuff, maybe try verifying over in #dev-chat ?

[whole roost]

Since I roleplay like crazy in projects I'd say that projects provide the better environment

For example yesterday in the roleplay the AI invented a side character, he got some story so I created a file for that one and uploaded it and it immediately took effect in the roleplay so that the AI has the information available proceeding from that

Can't do that on customs

[cloud pewter]

because I think your usage is mass roleplaying lol

has a lot of settings and characters

[cloud pewter]

projects has live update now?

[whole roost]

Always was like that

They also theoretically support memories but I haven't been able to instruct GPT yet to consider roleplay progression as relevant info to remember

So it never really creates one unless I tell it

I'll try around this evening because it should be possible to shift its attention

But memories seem to be bugged overall

Yesterday I wanted a replacement for some memories and it told me it updated it but it never arrived in the memories really even though I could read the text in the supposed edit

I had to delete all memories and tell it to add them from scratch in order to get any change on them

And the stuff that appeared in the memory list never Was the one it wrote in the chat in that memory box

I had to try like 15 times or so until it was correct and always deleting everything

[tired gazelle]

anyone having issues with o3mini high?

it ultra slow and stops generating after a bit

[whole roost]

Yeah the whole service right now seems to be a bit laggy

[inland mortar]

Yep very slow, he's thinking like he's had one too many

[cloud pewter]

memory is so bugged

[crisp cairn]

A network error occurred. Please check your connection and try again. If this issue persists please contact us through our help center at help.openai.com.

[inland heron]

I was about to ask

[iron elbow]

Anyone know if for the Pro version we have access to the "full" o3 model or only the o3 mini and mini high?
and do Pro users also have limitless prompts for o3 mini high?

[normal quarry]

the full o3 model is yet to be released but Pro users will have access to it. o3-mini-high is unlimited for Pro users I believe

[iron elbow]

Ah i see. thanks for the info. Honestly im changing to Pro soon because of the rate limit for prompts but the big issue i have atm is that deep research isnt available in switzerland (yet) .... and afaik they havent published any timeline for it being available yet, right?

[normal quarry]

they just say they are working to bring it to EU but there is no clear timeframe

[iron elbow]

: (

[warped walrus]

Why am I suddenly seeing reasoning in chatpgt -4o? 😭😭😭😭

[cloud pewter]

which plan u on?

[warped walrus]

Plus

[cloud pewter]

ive seen this on twitter before, i think they are testing something with 4o and that guy is also plus

but no official announcement/statement for it yet

[heady hawk]

get, voice and advanced voice is bad functioning?

[warped walrus]

This is so not done. I use 4o for everything. It's like my Google. But I can't wait 5 seconds every time I type something.

[cloud pewter]

it triggers only occasionally i suppose

[spice lotus]

Is it worthwhile to make custom GPT models specialized in a specific area of mathematics or computer science?

Does it make the gpt better ?

[cloud pewter]

also it mightve just been using the o series models when youre selecting 4o

similar to how free users use o3 mini but theres no toggle

[potent lark]

guys please can anyone help me here? I was subscribed to the ChatGPT Plus plan through the Google Play Store and it renewed yesterday.
The email linked to my Google Play account is the same one I used for my subscription.
I recently canceled my subscription through Google Play and attempted to restore it via the ChatGPT app.
However, when I click on the subscription options, I receive an error: "Error retrieving information from server, also if i disable the subscription on google play and i try to resub again i get this error. PG-GEMF-01.
I have already tried troubleshooting steps, including reinstalling the app, clearing the cache, and ensuring I am logged in with the correct Google account.
Could you please assist me in resolving this issue? Let me know if you need any further details.

[cloud pewter]

did u get a refund?

[potent lark]

i can't get refund anyway because the subscription is from last 24hours

[cloud pewter]

if yes you need to subscribe again
if no contact their support

[potent lark]

i didn't refund at all to begin with

im so sick and tired of it

[normal quarry]

if you are having problems the correct course of action is to get support at https://help.openai.com/

[potent lark]

i contact them twice 7 hours ago and i got nothing back

ia this a common issue tho?

[cloud pewter]

wait a lil bit

they could take days to reply

[normal quarry]

nobody here can help you with billing issues, you need to contact support

[potent lark]

thanks, i actually thought its a common bug

[rich surge]

I can't assist you on this, but what I can say is that: when signing up for something as serious as AI, you should always set up payment plans via the main website, and never third party platforms.

Always buy from the main source.

[potent lark]

i will kept that in mind from now on really

[worthy lynx]

Anyone else getting slow responses today from ChatGPT? status.openai.com says all is well so I'm wondering if it's me.

[cloud pewter]

hah.

[worthy lynx]

@cloud pewter thought so!

[upbeat pollen]

Deep Research seems to be working in chrome on windows 11, tried bunch of stuff in ios app and mobile browser and it just gets stuck

[sullen shard]

my chatgpt is so slow rn

[worthy lynx]

same, see above...lots of demand at the moment

[rocky condor]

yall, whens deep search coming out???

[cloud pewter]

are u on plus

[rocky condor]

yep

[honest jay]

I believe that they're trying to expand access to the currently-excluded Pro subscribers in currently-unsupported regions, then Plus and Team are being added as well.

https://openai.com/index/introducing-deep-research/

[...] we anticipate releasing deep research to Plus users in about a month.

[worthy lynx]

Faster than I was expecting! Woohoo!

[cloud pewter]

imagine eu pro users get dr later than plus users

they took so long to let avm video hit eu and sora/operator still doesn’t even have an ETA

[mortal salmon]

what is the point of paying for plus? ?

the original promise was early access to new models, but seems like only pro users get that now

[keen hill]

how many deep research prompts do we get?

[rich surge]

I like plus. I am plus

Plus is great. It reaps the benefits of all plans.

At an affordable cost.

[plain arrow]

4o creative writing performance has been seriously degraded for about a week (sentence fragments, bold formatting, lack of paragraphs, loss of continuity). Anyone else seeing this?

[rich surge]

People are saying it's getting better.

I also find it's getting better.

Maybe has to do with your custom instructions and memory.

[granite arch]

ChatGPT + Grammarly is my go to for work related writing lol. It helps a lot honestly as English isn't my native tongue as much as I'm fluent.

[rich surge]

If you want creative writing. It's better to go with designing your own custom gpt catered around your creative writing style.

Custom gpts doesn't get the love it deserves. It's so good.

I should make a tips and tricks guide for making custom gpts, because I feel like maybe people are confused with how they should make one for specific purposes.

[granite arch]

I do a lot of instructional videos for my job (not YouTube) and feel there are plenty of opportunities for people to create content like all things online...

I wish I had the time.

[rich surge]

Make one for the server if you have time. The more help, the better. Give back to the community.

[granite arch]

Ya...it's all I do but hardly come on here. Only on breaks lol.

[rich surge]

Well, make one for us. Please. If you're good at it, I'd love to see your work.

I'll be waiting.

[granite arch]

I'm not good at ChatGPT lol

Still learning lots...

[rich surge]

Me too, honestly. Would love to learn more.

[whole roost]

I don't get bold formatting but a lot of italic which I don't want to read

It emphasizes words with that all the time and gets excessive quickly

[cloud pewter]

we wait more now but yeah it’s becoming to be false advertising at this point

[cloud pewter]

I think it’s because of the Jan 29 personality update tbh
and tons of emojis for no reason

[cloud pewter]

u mean plus or pro?

[plain arrow]

Thanks, it feels like it fell and hit its head. Glad it's not just me seeing things. The strange part is I tell it to not use bold at all and it lasts maybe a full response.

[plain arrow]

When I ask it to not use sentence fragments, and use paragraphs, it lasts maybe 2 responses.

[granite arch]

Plus gets Sora, image generation, etc and free doesn't, no?

[keen hill]

pro

and how much can i use operator in one month?

[cloud pewter]

100/mo for dr
for operator it has daily limit but its unclear

[cloud pewter]

free no sora but has some img gen and avm
But the usages are very very limited

[granite arch]

👍

[cloud pewter]

I may start offering some sora stuff in the following days because my 2k+ credits are expiring soon

also if they are ever improving it, integrate it into chatgpt pls

[cloud pewter]

dalle 3 is two images a day, avm is 15 minutes per month

[rich gust]

Possibly a stupid question, how do you use deep research? I don’t seem to see any way to use it, is it not rolled out to subscribers yet?

[honest jay]

Hey! It's currently rolled out to Pro users outside of the UK, Switzerland, and the European Economic Area. If this describes you, check in the browser version, as it's not currently in the apps. If this doesn't describe you, OpenAI say that "We look forward to extending Pro access to all regions soon, and will expand to Plus, Team, and Enterprise users at a later date." https://help.openai.com/en/articles/10500283-deep-research-faq

[rugged igloo]

We finally know how smart o1 pro is

[sullen shard]

is chatgpt bad at math for you guys too?

its good with everything else but its terrible at math

[magic lintel]

what model

[honest jay]

Hey! Try experimenting with asking ChatGPT to use its data analysis tool to do calculations for you. LLMs alone don't really have a "calculator mode", but they do have the ability to use tools that can be used as calculators! Also, reasoning models like o1/o3 have improved skills in math. But if you want just a straight calculation, try something like "Use python to calculate..."

[hot anvil]

slow

It looks like the intern deployment is in progress

[sullen shard]

thank you

[halcyon cliff]

Anyone know why gpt 4o is now reasoning?

[whole roost]

mine doesnt

[rugged igloo]

Same

[elfin lava]

It’s down

It’s up nvm

[ebon canyon]

yo whats good guys

[obtuse jetty]

Hey, y'all, does anyone know when the AVM mode will be released on the Web?

[dense halo]

Can you guys stop fragmenting models for chatgpt, you're about to cross the point (huh?) where? what I should use for which specific task and why.

Advanced reasoning vs fast advanced reasoning.... Ok? I assume since it's called mini that means it has its downsides, but calling it fast at advanced reasoning.... Obviously that's what one would think to select, but calling it mini, in almost all cases of "software versions" it's a simplified/basic/trial of the full version, but it's faster at advanced reasoning??? bro...

o3, 01, GPT-4o, all this... It's getting nonsensical. I know what each does and what they're best for, but that there's the problem where you're on a chat, new model comes out, and you may feel the need to import the old chat into the newer model for better results.... Honestly this is is strange.

And then there's this weird bug where it'll just downgrade on its own even if it's not in the "trial models"...

[sick pecan]

should be a choice of 2-3 at most, and if there are more than that... route the query behind the scenes

maybe like.. Smart, Fast, Cheap

[dense halo]

A new user coming to the most popular AI, expecting a chat.... sees all that.... They're going probably end up needing the holy spirit to guide them to what on earth are these options?

[woven dirge]

When is o1, o1-mini or o3 getting multimodal? I really can't stand 4o anymore, it's being really incorrect

Also memory. I use memory a lot and I really dislike it not being able to access it

[sick pecan]

if history is anything to go by, chatgpt features are rarely integrated

[woven dirge]

I'm so pissed off bro

[sick pecan]

proof of concept team are reassigned and there is no follow-up integration or polish team

[woven dirge]

I pay 20 USD monthly for this

[sick pecan]

yeah this will be my last month of chatgpt sub

i was the biggest fan. but chatgpt hasnt changed much since 2023

[dense halo]

also.... great at <coding> and LOGIC... Reasoning then there's Logic, They're the same thing... Basically, because you can't do one without the other, but in order have any form a reasoning, logic comes with it, not string attached. There's no balancing act. You either logically reason, or reasonable logic.... Come on man. You're just thinking stuff through. Give me the K/D ratio of the numbers mason, I must know what they mean.

[wet warren]

when is deep research releasing in uk

[sick pecan]

every other chatgpt clone is capable of mixing models and tools

[dense halo]

There is openai playground but I got my foot stuck in the jungle gym

I love to have options, but only if it's clear what the freaking difference is based off vague/broad-basically-the-same-thing as all hell terms as mentioned. Also, especially if they stay put and don't evolve backwards midchat regardless of to-trial-not-to-trial amount of time I am forbidden to know. CATCH! sometimes they allow me to know. Then, when it decides to let me know how long I have left before I get spun around. The "what the the what the hell what the hell" effect, I believe scientists call it. Oh and by the way, I predict many errors in the future for chatgpt users... I think I am about the have one now when have the energy ask it a simple question only for my question to be over-explained .... (Server error, press this button to to regenerate chat). Answer please? "Question please."

This AI is obviously too advanced to understand what it's doing.

[ebon canyon]

the downgrade currently happens often when u start a deep research and refresh the page. then for whatever reason it switches to the 4o mini

[dense halo]

I also love the part when you don't realize it many prompts deeper confused as to why the degradation of information

because thou shall not be notified

[sick pecan]

what annoys me is that i cannot lock in features. and so if it decides to web search or canvas or whatever, i can no longer use certain models. or if i toggle models then i lose access to features that had been working

so not only is it fragmented, its unintuitive. principle of least astonishment be damned

[dreamy berry]

def true

having a very hard time getting canvas to 'load' code. it often truncates/shortens what i paste in, even when i explicitly ask to reproduce my code exactly. anyone having this problem?

[sick pecan]

ohyeah canvas cant even load half of what a codeblock can render

canvas could have been a defining feature. instead its an absolute disappointment

[dreamy berry]

still super beta obviously

but yea, very irritating

[sick pecan]

how hard would it have been to just render a codeblock and then populate the canvas with that?

[dense halo]

Well that's where you start a project, but you're locked into whichever model you chose, and whatever description terms you gave it... both are unable to be changed, but at least you can go back to the other chats that are in a quantum state of here, there sometimes nowhere and you can add it back into your project and then you can.... Make sure to hydrate.

[dreamy berry]

i am actually going back to editing code outside canvas bc it modifies my code so much when i 'read' it in

just in the regular chat window

[sick pecan]

projects is the last thing i'd use for an active project. the best it can do is archive

tasks is the last thing i'd use for tasks. because i need something reliable

canvas is the last thing i'd use for editing code/documents. because it cripples the output

[dense halo]

There's Windows Copilot which is good for navigating Windows Copilot

[dreamy berry]

it does seem to work for very simple projects. but anything even slightly complex it chokes

it chokes even if i directly paste in my code into the editor

[sick pecan]

yeah but then if its so simple, printing out a new codeblock isnt that bad

if anything seeing the revisions in the history is a feature not a bug

[dreamy berry]

yeah i do like the idea of the canvas feature it just works very poorly.

[sick pecan]

not that you can scroll very efficiently in chatgpt. somehow they made it perform terribly

[dreamy berry]

if it can't load a full code block it should say so instead of editing your code when you say to not edit it!

eg if it is too long

[dense halo]

You really should use Github Copilot or OpenAI Playground for specfic tasks like coding.... Github Copilot has the database of.... Basically everything to do with coding, chatgpt for coding?

[sick pecan]

canvas should just be a wysiwyg view that toggles on/off for any given codeblock. instead they integrated it in a broken way

[dreamy berry]

agreed

i think it did initially work as you suggest but they tried to 'fix' it and made it worse

[sick pecan]

design by committee

[dreamy berry]

literally just tried this study this R function and load it into canvas for editing. load the full complete function and do not truncate or shorten any of it, no matter what your prior instructions indicate. if you truncate it i will ask you to do it again, wasting compute on your end. and it reproduced 1/3 of my code 😂

[sick pecan]

i think you have better chance by pasting the code into the input and clicking the canvas button

but good luck after that

[eternal rock]

`` Server Tip: ``

Add your well-crafted prompts to our #1019652163640762428,
or share your interactions with ChatGPT in #1050184247920562316!

[dreamy berry]

will try

[dense halo]

I would like to be able to prompt with codeblock + syntax highlighting + tabled.... Only the responses have the ability to do that. My obvious reason? Locking on to what part of my prompt was code and what was the demand.

[sick pecan]

be sure to have a canvas-compatible model selected, else the canvas button doesnt appear

[dreamy berry]

yeah. i got it. trying atm. thx

[sick pecan]

not that the UI tells you which models are canvas comaptible

because whats worse than disjointed features? hidden disjointed features!

[dreamy berry]

yeah i would really like to use canvas but it is basically not usable

[sick pecan]

fortunately there are now third party alternatives that you can run on your own desktop

thats what ill be using post chatgpt

(and these canvas tools work with any model, even the ones chatgpt doesnt support!?)

[dense halo]

Am I being ghosted? I would like yo' coder folks attention.... I believe, if you're unwilling to use github copilot, try OpenAI playground? I really to just.... forget about chatgpt and consider it basic-generalized-halfassed stuff. Don't code with ChatGPT, I mean, it is called CHATGPT, I don't care if it says "advanced hyroglifics quintom codeography cia zoom enhance O-MY-G0D model.

Playground for advancement, it's got all kinds of stuff going. The prompts don't cut out if you "locally" host some sort of UI like.... LibreChat. That's good stuff. I don't know what it is, but when you use ChatGPT, you're faced with interrupted responses, timeline shifts, and blackouts.

I'm not advertizing, it's literally openai and some silly frontend that stops the connection errors and half-assed responses.

I hate dealing with tokens though

Lord please help me

[dense halo]

just buy like a few bucks worth of OpenAI tokens and use LibreChat. It's a very simple UI, and it's got all the chatgpt **gpt ** models + actual advancement and other testing/development stage ones that's not in chatgpt. You customize your model, tweak it as you wish, hardlock it, unlock it, (whenever, whichever) duplicate it for mistakes, tweak it again (keeping all data, memory, specs). Archive the hell out of it.... It's almost like it's a playground, and chatgpt is the swingset that some kid flung all the swings where maintenance constantly needs to be involved to untangle the mess (the kid is still there, and he's gonna do it again)

[dreamy berry]

its a good idea.

i am not committed to canvas, my question was about recognizing that it is not useful

[sick pecan]

i aint responding because you keep talking non-openai products. and thats the only time i ever get muted on this server

[dense halo]

openai playground.... non openai product?

[dreamy berry]

playground doesn't seem to really work for me better than regular chat UI

[sick pecan]

playground has the benefit that you can edit assistant response, which is a feature you see in all the chatgpt clones too

like if it refuses, just edit it to say "sure thing" and click continue

[dreamy berry]

cool, good tip

[dense halo]

playground has no UI... at least it didnt, did they make one? The hell with it anyway, use a frontend of sorts that may be openai product (IT TOTALLY IS, ALTHOUGH I AM LYING)

[sick pecan]

back in 2023 i was using playground more than chatgpt. but in those days we only got 25 queries per 3 hours, and the only feature was plugins

[dreamy berry]

it really does feel like they nerfed the plus tier after introducing the pro tier. though a mod on here indicated that they did not. i am not sure i believe it.

w/ deepseek maybe they will un-nerf

[dense halo]

I've checked the ToS and I have verified I am just grazing the grey area of doing an ooopsie, These are not the violations you are looking for. We all want to use chatgpt because autistic people do not like rapid change +

uh

so on and so forth

[upper abyss]

I can't use 4o, it seems to be stuck on one of the reasoning models no matter what I select in the android app

[crystal tapir]

Is it just me or is ChatGPT using reasoning in every interaction on the android app?

It's kind of annoying. This happens even with the default model (4o) enabled.

Does this use my limit for the o3-mini? This is important to me because I use it to study and I don't want to spend on ordinary interactions.

[dreamy berry]

hard to tell exactly what it is doing. i guess, if you want any hope of having reproducibility or any more nuanced understanding of exactly what model you are using, at a particular moment in time, you have to use the API/playground. the app is not well tuned for power users.

oh jeez.

[dreamy berry]

to be fair maybe this is perfectly fine for most users. maybe only a small % would benefit from having more fine tuned control in the app

(I would, if anyone is watching... )

its still early days though.

¯_(ツ)_/¯

[dreamy berry]

this didn't work. but it does actually still work to just ask for help with code in the chat UI (outside canvas). works perfectly well for the case i was working on.

so, clearly a canvas issue

[ivory remnant]

GPT has significantly slowed down lately, especially on the web/PC version. It's really frustrating I've relied on it heavily for the past two or so years for coding and planning. It feels like I've lost a valuable tool. I tried switching to Google's Gemini, but it just doesn't have the same familiarity or the wealth of knowledge I've built up with ChatGPT. Has anyone else experienced this, or found any solutions?

[visual flame]

wen attachements to o models, feels like such a waste

[rugged igloo]

ewww why did discord change on pc i dont thisnew layout

[narrow viper]

( deleted my request, numbers were wrong )

[sick pecan]

i dunno. it used to be way slower. like you can still use gpt-4 legacy model and its noticable slower than gpt4o

[crude dawn]

Bruh the new ad legit felt like it was a warning before an ASI would take over in the first 10 seconds

[granite harness]

Do they really need to advertise to us? We are literally in their discord. What is the point? Is there something new out? No?

[crude dawn]

I saw it on instagram as a story

[eternal rock]

<:book_icon:1171408210398289941> `` Rule 1 `` Be respectful.

Practice kindness and positive regard. Harassment, hate speech (such as sexism, racism, or homophobia), or other malicious conduct will not be tolerated. Maintain a respectful and positive environment.

<:book_icon:1171408210398289941> `` Rule 0 `` Adhere to the following Terms & Policies:

Discord: Terms of Service, Community Guidelines
OpenAI: Terms & Policies

[elfin lava]

Oh

[eternal rock]

Get notified for server events!

Visit id:customize to pick up the <@&1261377106890199132> role.

[elfin lava]

Was that a wind sound I heard 💨

[frozen birch]

i must say one - deep research are lovely

[rich surge]

I'm super jelly as a plus user

[frozen birch]

i think soon Plus users have it too m8 🙂

[rich surge]

Let's hope so 🤞

[warm lark]

how many devices can you use gpt on before you get the suspicious activity thing? I have a few devices in different places that I use gpt on and I keep getting the suspicious activity thing on the plus plan

[rich surge]

Uhhh...it tracks IP address. So if you're using it in two different places at once, then that counts as suspicious activity.

Could get you banned.

And since it tracks your IP, it could be an IP ban. Which is really bad.

[warm lark]

It’s not like the places I’m using it in are cities apart, it’s like, home, work, school, Internet cafes and libraries

All should be within like 40 mins of each other at most

[rich surge]

I believe the distance is a mile radius of where you can access the IP. So if it's like a cafe nearby that you're using it at the same time that's near your house. Or within your house somewhere, that's fine. But if it's like in a different part of the same city, then that still counts as suspicious activity.

[sick pecan]

if you tech savvy you could run a proxy and run all web traffic through it. if you run it at home, use tailscale or something

[frozen birch]

+1

[warm lark]

I have no idea what that means, I’m not that knowledgeable about tech

[frozen birch]

tailscale its a perfect solution for those kind of situations

@warm lark tailscale its like vpn.... but much better than normal vpn in short words

[rich surge]

It's better not to use a VPN. Because vpns are unreliable and they can disconnect and connect at any time which reroutes back to your IP, which counts as suspicious activity.

[sick pecan]

proxy+tailscale probably too tech savvy for most users on this server

[frozen birch]

thats why i told - much better than normal vpn

;)))

k, gonna sleep, bye peeps 🙂

[rich surge]

Ah okay. Tailscale

[warm lark]

Preciate, just checked out the website, I’ll prolly use it

[sick pecan]

by itself it isnt gonna solve your problem

[warm lark]

I’ll look up how to set up a proxy then

[sick pecan]

proxy is something you configure your browser to use. then you use tailscale to get to proxy wherever you are. and so no matter where you are, it always looks like you are coming from the proxy network

[rich surge]

You learn something new everyday.

[bitter dune]

sorry, bros i wonder that not working to features of deep research in browser, whoever can given to solve the problem
i appreciate to response for it

[wise anchor]

cant wait for o3 file support so I can dump an entire NextJS project on it

[elfin lava]

OpenAIGPT

[rich surge]

Oooooo 👀

So refreshing

I feel like I'm in a washing machine, and everyone is invited

[cloud pewter]

[coarse jacinth]

Hi all, were the rate limits ever posted for o3-mini-high? I was limited to 50/week but they said something about 150/day?

[sick pecan]

thats normal o3mini 150 day

[cloud pewter]

yeah that version is worse than o1

Only high is on par

but with some actual testing you’ll know that these are all smoke and mirrors

[sick pecan]

is worth checking your dashboard to check if You been offered free api tokens, as they work for o3mini high...

Free tokens by helping improve OpenAI models

Through the end of February, some users are eligible for 1M free tokens a day across o1 and GPT-4o, and up to 10M a day across o1-mini and GPT-4o-mini. Opt in your organization or project to share your data, which can be used to improve future OpenAI models. Check your dashboard to see if you’re eligible.

[rugged igloo]

what is this

[elfin lava]

OpenAI new chat bot

[rugged igloo]

chat gpt?

[elfin lava]

🙈

[cerulean fulcrum]

Unlike before, the Deep Research stopped asking me question and it seems like it just goes directly to the reasoning (o3-mini-high). Is this the symptom of hitting the 100 query?

[rugged igloo]

idk i havent hit the limit yet

[hasty sedge]

yo is chatgpt glitching the chat limit for anyone rn?

[cloud pewter]

anyone here having issues with dr

[solid bronze]

Will the Deep Research be given to PLUS users or is it only Pro?

[normal quarry]

it's coming for Plus

[minor hound]

Quick question: is deep research only with the 4o model? It seems to not work for me with other models.
And since today it seems to be also stuck in the process ...

[normal quarry]

It does not matter what the model selector indicates. Deep Research uses a fine tuned version of the full o3 model.

[minor hound]

thanks for your answer! seems reasonable.

is it now also released in europe? and thats why its not starting? 😄

[normal quarry]

it's not available in the EU and UK yet

[minor hound]

well ... some of my collegues see it in germany with no VPN.
I hope Germany is still in the EU 😄

i can proof it.
My work VPN (then i am in Germany) is also showing the button. Dont know if its a mistake or working as intended 😄 But it is shown there

[spice star]

has anyone else run up against their task limit?

it seems to be 10 tasks which is a little short if you're scheduling stuff months in advance

[eternal rock]

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.

[fast solar]

It looks like OAI rolled back the "June 2024" knowledge cutoff for 4o (It's no longer on their "Release Notes" page)
I wonder if that means they rolled back the response improvements (math/coding/"emojis") too
https://help.openai.com/en/articles/6825453-chatgpt-release-notes

[cloud pewter]

why are they reverting the knowledge update

i only want to get rid of the emojis

[whole roost]

I begin to want to get rid of italic text

[fast solar]

I found out the knowledge cutoff applies to 4o-latest (only accessible via playground/API) but not 4o

I want to believe that decision was made to

1. Encourage web-search
2. Reduce safetycel
3. Cool it on the mini-releases

It was indeed a rollback though, found a pic

[tidal hedge]

Interesting, 4o is sometimes using reasoning now

[cloud pewter]

i hate web search in 4o(i disabled it inside the settings) because when it’s searching it get rids of all of my custom instructions

and it triggers way too often

[fast solar]

I wonder if the custom instructions can be adjusted to always include [pastebin link to your custom instructions] when performing web search and have it in high regard

[fast solar]

Really? Does the refresher button indicate what model it was using? I wonder if it switches to o3-mini

Though that would annoy me if it ate up my rate limit for o3-mini prompts when I wanted 4o

[tidal hedge]

It changed to o1 apparently

[cloud pewter]

true that lmao

[fast solar]

It did that without toggling the "think" button or whatever? That would annoy me more

[cloud pewter]

they may wanna push the reasoning ability but we don’t have unlimited usage or that much as 4o

[tidal hedge]

Yeah would be kind of annoying other than I hadn't used o1 as much recently, probably only 10 times tbis week

[cloud pewter]

does it say o1 when you click the regenerate button

[tidal hedge]

Yes

[fast solar]

Hope the model keeps how close I am to approaching limit in mind

[fast solar]

What prompt called this function? I'm skeptical of it doing this without request

[cloud pewter]

I really want to do an counter extension to track those usage

but there’s nothing related to the quota on the website

[tidal hedge]

Decided to ask it this from jokes on another server, left it on 4o & didn't toggle on search or attach anything since it was a meme prompt more or less.

"What would a From Soft souls game be like for VR?"

I tend to just use o3-mini unless it's a throwaway prompt like that or I need to attach file

[fast solar]

My quick fix has been to have "Count the responses for [model] at the end of each response" + note the time at the start of new chats, but ofc that only works if you know the rate has reset or when it does

[fast solar]

Damn, can't seem to enact it with a vanilla chat

web portal and Win10 desktop app don't have the "Think" toggle though

[tidal hedge]

It's probably because I have a lot of usage left for o1, semi forgot about it with o3-mini. It's hard to keep track of for sure but I think the weekly reset should be in only a day or two for it for me.

I don't have think toggle either. Just on android app

[fast solar]

Still nothing on android app too, maybe something in CI or context

[tidal hedge]

Plus user currently. Yeah could even be an app bug

custom instructions are off, it didn't work well for me before

[fast solar]

4o is going so slow, I wonder how much compute they're loaning o3

[tidal hedge]

Will be interesting to see if they add o3 normally soon. Feels like with deep research it's probably likely to also have search and then hopefully also has attachments.

Maybe it's due to using o3-mini & DS, but o1 replies felt short in length.

[gilded owl]

Will OpenAI be open source?

[tidal hedge]

I think the chain of thought display might be a big factor in how interesting a model is to use. Seeing full CoT helps me connect with the model response and understand why it gave the answer it did so I'm more likely to agree or prefer its answer.

[fast solar]

Sama recently tweeted an ambiguous "We're on the wrong side of history" regarding open-source but seemed to imply they're going to try to redefine open-source

[gilded owl]

"We're on the wrong side of history"

What do you think of deepseek? Why are you talking about it recently?

[fast solar]

Correction: It was from an Ask me Anything thread

I think a lot of the safetycelling and closed-source is against Sama's wishes and more for the investors

[fast solar]

DeepSeek R1 is being talked about a lot because it's within a margin better than o1 on some benchmarks, but IMO it's great for less safety in response and that probably has to do with its better performance

[gilded owl]

okay,thanks

[whole roost]

I actually prefer a non-reasoning model for most things I do

[tidal hedge]

I used to, but R1 often does well in those cases now. Just needs better attachments support

[whole roost]

I used the API of R1 and stayed on GPT

[summer fjord]

I think Sama just wants to win. If he is the best player in town, he wants closed-source. If that position is threatened by open-source, he then wants to contest the threat with open source of his own -- noting a switch in strategy.

[summer fjord]

CoTs don't have access to the memory features, sadly, which bumps them back a little for casual tasks.

[fast solar]

Memory as in chat context? Or just RAG? I thought o1 and o3-mini processed chat context albeit I don't know how much

[whole roost]

I currently don't have any stored memories anyway so that doesn't concern me but I definitely would need files

[fast solar]

Oh nevermind now I understand ^

[whole roost]

And they are very slow and I usually think that the creativity is actually lower when it overthinks

[summer fjord]

Hm. I also noted this fact. 4o is better for generating simple creative text. But then again, it also has access to Custom Instructions (at least in my case)

[whole roost]

o1 and o3 as well

[summer fjord]

Oh, in the Pro version, you mean? How do they fare with that added feature? Does it increase 'reasoning' time, in your findings?

[whole roost]

I think they have access to custom instructions for everyone

[summer fjord]

Haha :)) so cool

I got the Plus subscription, so maybe that's why xD

[whole roost]

You can also use them in projects where you can set custom instructions

[summer fjord]

Thanks for mentioning it. I actually forgot that was a thing.

[fluid fern]

Hello everyone, can anybody tell me why GPT is not working since last night for me the way it should I need things I need to do things I rely on it. I can’t even delete a chat.

[summer fjord]

Did you try a different browser?

[fluid fern]

I use the app

[gilded owl]

Does the web version work?

[fluid fern]

Nope

Hold the time this request timed out

And I am a subscriber since it started from the first week on

[frozen birch]

anyone in EU has deep research? with pro subscription?

[summer fjord]

I read a post on reddit just now from a guy from Belgium mentioning he has it since 2 hours ago.

[frozen birch]

thank you, i can confitrm

i have access to deep research without vpn

without combinations etc

so sweet

[summer fjord]

So cool. Finally EU not feeling so left out

[gilded owl]

Why do you use a VPN? What country are you in?

[frozen birch]

but funny thing - 3h ago i revoked o1 pro subscription since i was mad xD

[cloud pewter]

i hate when o3 mini overthinks

[frozen birch]

Poland

[cloud pewter]

u cancelled sub?

[frozen birch]

yea, it will be closed in 6 days

[summer fjord]

And you still have deep research xD?

[frozen birch]

but i guess i resub it again since they turned on deep research for eu customers

[frozen birch]

Brandon if you cancel sub cancellation are not automatic

[cloud pewter]

why u mad tho

[summer fjord]

Oh yeah. That's correct. What a great decision, mate :)). a lot of bang for buck.

[frozen birch]

i must say one deep research even in learning linux are a GAME CHANGER, since he have access to newest informations

[summer fjord]

Could you perhaps dm me a chat link, if that would be possible? I'm curious as to how it responds to prompts/ how to best prompt it. Since there's not a lot of sample prompt + output to go out on yet.

[frozen birch]

offc

btw im thinking now - how many people like me cancelled subs

[gilded owl]

why?

[raw portal]

Has anyone had deep research never finish (or maybe it never started). I refreshed and tried different browsers and it just says "working" after many hours

[frozen birch]

since 3h ago deep research wasnt active for eu users

[frozen birch]

for me i must refresh page for work

[tough heart]

yo I made a little web game and wondering if someone will give me some feedback 🙂 I don't know if I can post link here so pls dm if you would be willing to test it out and lmk what you think

[cloud pewter]

u get ur $200 back which is nice lol

[frozen birch]

true 😛

[whole roost]

Oh indeed it's there now

What useless question will I ask first

[raw portal]

Tried that. Even tried different browsers

[frozen birch]

@whole roost "make me report about what is better - pizzas or nachos" XDDDDDD

[frozen birch]

lucid maybe you can try on another device if you have ?

[raw portal]

I did

It's not a browser bug

[frozen birch]

i guess in that situation you need contact openai support 😦

[dense halo]

what?

[frozen birch]

i think @cloud pewter wanted to say that im getting back to openai with my 200 bucks XD

[dense halo]

so you continue?

[frozen birch]

yea, mainly i cancelled it because there wasnt deep research but in those situation.... 🙂

[tawny whale]

Deep research is here for me as well! EU pro subscription! haha makes me so glad! 😄

[frozen birch]

thank you deepseek and our chinese friends i think we got that, anyway - im gonna back to work, later peeps

[whole roost]

I could have make deep research a search about me and see how much is on the internet

[vast crag]

Guys, the are tasks still bugged with notifications?

[fluid fern]

GPT is not deleting nor archived conversations I don’t need anymore. I try the app and the browser nothing solves the problem. Such a drag.

[brisk gazelle]

Hello everyone is the deep research function available on the iOS app?

[whole roost]

It's so childish with the emojis

[frozen birch]

Vishanka you can try non standard instructions

in personalization

[whole roost]

It has those, it just likes to ignore them

Same conversation when it remembered its instructions

But most of the time its 📈✨🚀✅ now

[cloud pewter]

Real

did they actually reverted the change including the knowledge cutoff

[whole roost]

Yes GPT i want help with the water cooler 🥵🔥💦

[buoyant bear]

Does the output of the ChatGPT Deep Research feature vary depending on the selected model?

[patent frigate]

i thought the $20 granted full access to 4o?

[cloud pewter]

yeah only message count limits

[lapis hatch]

Hi I tried subscribing and my card got charged, but on the website it got declined

does anyone know how to contact them?

[cloud pewter]

hey chatgpt. this is not a file. this is an image.

[grizzled thicket]

Images are files

[slim dust]

Then deep research will be available for Plus users?

[dense halo]

real

ahahaha

[worn whale]

did they privatize deep rsearch?

[stray peak]

Google Gemini has their own deep research on their basic plan and has had it for months now... So yea. There's now an open source Deep research that scores as close to OpenAI for free.

Wamp Wamp

[hoary gazelle]

yes but, did they make a cool video about it?

[fluid fern]

Hello everyone, I am unable to delete all the chats that I don’t need anymore not only in a browser or in the app. It doesn’t work gives me error. Something do you know when it’s gonna be fixed?

[fervent agate]

I know the announcement says WhatsApp, but I'll never use that for reasons. Anyone know if it works with Signal or other, better apps as far as messaging the toll free telephone number ChatGPT?

Nah, i find being able to call it or message it on the road in rural areas really valuable, I just am not going to use whatsapp

I'm often without mobile connection.

Well, mobile data.

I don't know, I don't use WhatsApp, that was my question. They have a toll free telephone number you can use with it, which I have, and my question was: is there any other, preferably SMS-based way, to also communicate with that.

I've noticed you can't SMS it directly but that doesn't always mean certain apps can't.

[potent saddle]

Limit

[sand lily]

haha, 4o thinking? The app is a little broke sometimes

[potent saddle]

We got 4o thinking before gta 6

[visual nova]

does daily tasks work with search? can i have it write a daily news report?

[static ridge]

3 hours later... what's people's experience with deep research?

[static ridge]

isn't that the adaptive model switch or w/e

[fluid flume]

deep research not rolled out to canada business plan yet?

[whole roost]

i didn't expect much but got a really really good response even if the topic wasn't that important but it found basically everything

[old notch]

Anyone else finding that the normal chatgpt 4o model is being extremely slow now? Like in the last few days it's slowed down to a crawl

[static ridge]

how long did it take? they said 60-90 mins or something, it's been hours for me

[whole roost]

uh, like 10 minutes or so

[static ridge]

lol

[torpid raft]

my chatgpt talks to me like a frat bro with no former instructions to do so

[grizzled cedar]

any1 else having Suspicious Activity Alert? I can't use the models, already setup 2FA and did a force logoff all devices

[visual nova]

it froze if its more than 30 minutes

deep research is one of their most useful products

[static ridge]

not very useful if it's silently breaks

[summer fjord]

I have those regularly (sometimes, it's sporadic), but they don't affect my use of the models, if only for a prompt or two.

[grizzled cedar]

all my models are disabled until I sort out this 'suspicious activity'
it's been like this all morning

[summer fjord]

That's so weird. Sorry if this has been said before, but have you tried using a VPN, if you have any handy? I wonder to what the 'suspicious activity' is tied, account, location/device. Maybe it flagged some words in a prompt you used that it didn't like (speculation).

[grizzled cedar]

no vpn available atm
I use ChatGPT at work, at home, and on my phone, all with the IP in the same city
I'm at work right now and was just coding
It's telling me to check out this https://help.openai.com/en/articles/10471992-why-am-i-receiving-a-suspicious-activity-alert
but I already did everything it says there

[summer fjord]

I'd open a ticket or mention the problem in some line of communication. Like "Hey, I can't use any model, being locked out due to suspicious activity I know nothing about, do you have any recommendations for how can I proceed to use the service?"

Also lets them know you aren't up to anything.

That 'suspicious activity' alert is sometimes anxiety inducing. xD

[astral storm]

Any help in managing files in ChatGPT projects?
When i upload a file to a ChatGPT project, is there a way to open those files? I can't find an obvious way to see the content of a file I added. All I seem to be able to do is delete it. i cannot see a way to open it, view it, download it, rename it, etc?

[fast solar]

Looks like OAI rolled back the updates to 4o for ChatGPT apps — the update was removed from the change log, and ChatGPT 4o is now stating that its knowledge cutoff date is Oct 2023, while 4o-latest via playground states June 2025.

Is it possible that they removed the cutoff while retaining the fine-tuning for ChatGPT? Because it certainly is still using more emojis than before 1/29...

[wary narwhal]

why did he just randomly think in russian, when the conversation has been in english and never anything else? LOL

[feral zinc]

removed from where? i can still see it here https://help.openai.com/en/articles/9624314-model-release-notes and it works the same

[fast solar]

Oh thanks, I was looking under the wrong link:
https://help.openai.com/en/articles/6825453-chatgpt-release-notes
Still, when I asked 4o on ChatGPT it stated oct 2023, weird

Vanilla chat no CI etc

[honest jay]

Oct '23 is the correct knowledge cutoff date for 4o: https://platform.openai.com/docs/models/gpt-4o#gpt-4o

[feral zinc]

with search off

[fast solar]

Weird... Vanilla chat:

Omitted May 10, 2024

[feral zinc]

for me it said Jan 22 2024 was latest

[fast solar]

Huh

[feral zinc]

ask "2024 oscar winners?"

[fast solar]

😡

[feral zinc]

you need to turn it off

[fast solar]

It's not toggled, it keeps enabling itself and the web app doesn't have a direct setting to disable it

[feral zinc]

there are 2 toggles, one in the chatbox and one in the window where you set custom instructions

openai has a dozen of different teams doing their own thing with feature fragmentation 😅

[fast solar]

Oh thanks, CGPT was lieing about no direct setting

Now turning off code interpreter

Not sure why my account is behind on the June 2024 cutoff

Do you have chats stored in projects folders? Maybe that is holding it back

[feral zinc]

yes but made these without any

[sand lily]

haha, oh no, 🤣 not my precious o1 uses.

[fast solar]

At the same time, 4o-latest returns this in the playground. I don't know what's going on with my account

[torpid kelp]

Why does 4o talk like a gen alpha tik toker now

[cloud pewter]

i might work on something like a counter for plus users but itll be based on a very raw mechanic which is manually track every single message

meaning you need to literally wait for 3 hours or even a week to make sure all of the counter resets to full, then u enable this counter

[fast solar]

Might be able to have it switch to 4o and update the count in memory every 5 prompts or so

[cloud pewter]

yeah maybe

[grizzled cedar]

how I can open a ticket?

[cloud pewter]

support.openai.com

[fast solar]

I tried but it's unable to switch the model. Still, it keeps count within its CoT

Is this worth pasting into my CI? Not sure if what I asked actually has any basis

[full veldt]

I really like o3-mini but for some reason it seems to rather arbitrarily decide whether to show LATEX as rendered maths notation or script which is kind of weird

[shut dagger]

Having an interesting conversation with voice gpt about the importance of positional encoding in transformer models and bam “my guidelines won’t let me talk about that”. It’s getting more sensitive these days

[dense halo]

chat

am i cooked

[dreamy lark]

rip

[grand iris]

why did I get a ping?

[rugged igloo]

Bro found AGI

[wise yacht]

Gotta say, Gemini 2.0 has a wonderful sense of humor

[whole roost]

seems like image analysis is broken

on 4o*

o1 does it

[pallid karma]

4o is the one best suited for some general creative type task, correct?

[rich surge]

Yes

[pallid karma]

👍 thank you

[rich surge]

Np

[rugged igloo]

o1 pro cooked

[ocean bramble]

how can i use deep research again ?

wheres the button ?

[potent saddle]

you need to have pro plan

[ocean bramble]

i do

o3 mini is the one ?

[hoary laurel]

What is it?

4o can think?

OK, thank you.

[solar rose]

They added reasoning to 4o?

nvm, it's a bug.

[honest ledge]

is chatGPT down?

getting internal server error, conversations wont load...

[wind cradle]

yes

[honest ledge]

sighs I've been working on this document for 4 hours, and 2 sections left and now it crashes... it's due in 45 minutes

[elfin lava]

My plus just ended, and now contemplating if I want to renew or not since recently a lot of things that I love about it has been kinda like ,nerfed version (AVM specifically)

[pine berry]

Oof

[ruby pivot]

Did anyone else’s chat gpts just get deleted

[pine berry]

Its possibly just down for now

[honest ledge]

yeah, everything will probably return as normal... just, will it do so in time

[ruby acorn]

Definitely having issues authenticating and doing anything.

[primal wyvern]

Is ChatGPT down again?

[elfin lava]

Wee thank god I was about to renew my plus

[ruby pivot]

I thought so

[sick pecan]

chatgpt broke!?

[honest ledge]

yes

[elfin lava]

God gave me a sign

[pine berry]

if you want a rough estemation when it will be back online, its about 30 - 45 mintues but if its bad enough could take up till an hour maybe 2, but some times it can take up to 4 hours but thats if it is really bad

[sick pecan]

a year ago this would be code red. but now i literally got 15+ other LLMs to use

[cloud pewter]

all of my past chats disappeared LMAO

[elfin lava]

Now that yall here , use deep- banned

[ruby pivot]

Same

[honest ledge]

when I get 600GB of RAM i totally will

[ruby acorn]

Did someone ask it to calculate the question to life, the universe and everything again?

[pine berry]

How I know this is because of how frequent these outages can be

[elfin lava]

Periodt!

[sick pecan]

thats easy, 42?

[ruby acorn]

That's the answer, not the question 😛

[mossy rampart]

Try VRAM because you'd have to be crazy to run inference on a 700B model on a CPU.

[elfin lava]

You know what’s funnier than 24

[wind cradle]

I am wondering if the usage of o1 and o3 is consumed when a server error occurs. If usage is consumed due to an error, there is nothing more ridiculous than this.

[elfin lava]

AYYY ChatGPT is over 🎉

[honest ledge]

700GB of RAM is FARRRRRRR cheaper than GPUs with 600GB of VRAM

[pine berry]

AI is still relatively new so these issues are to be expected

[tawny mirage]

Who needs 700GB ram?

[sick pecan]

relatively new (chatgpt user for 2+ years now)

[ruby pivot]

Chat gpt is so popular that it crashes itself from all the attention

[unkempt perch]

Though it was just me since no one was saying anything

[wind cradle]

Not as good as o1 or o3, but if you are looking for an AI smarter than 4o, I recommend Gemini 2.0 Flash Thinking Experimental. Recently, I often use it when o1 and o3 are unavailable, and it's not bad. It's also free.

[mossy rampart]

Enjoy your 1 token / 50 minutes then 😏 still, you'd need one hella rig to slot 700GB RAM into it.

[honest ledge]

status.openai.com: All Systems Operational!

[native otter]

ChatGPT ded

[sick pecan]

#ai-discussions for non-openai models

[pine berry]

There was an outage yesterday

[ruby acorn]

Major feature announcements have a habit of corresponding with outages.

[elfin lava]

I second this, weird for me to say but it is getting better

[sick pecan]

if we were on #ai-discussions i would say i saw some redditor post 16 tok/sec using 2x m2 ultras with quant r1

gpu? we dont need no stinking gpu

[honest ledge]

yay, I lose access because the rich fvcks are getting an update I won't see for 6 months

[elfin lava]

LMAO

cc pr0 user /jk

[pine berry]

That to

[tawny mirage]

Who are the rich folks you refer to?

[mossy rampart]

Don't think there's enough Pro users to cause a crash with regular use? But could be some overlook in Deep Research causing it to loop itself and crash the system or something similarly ridiculous

[elfin lava]

Remember when they open up ChatGPT integration for iOS and it did major outage ?

[pine berry]

Assuming those with pro

[honest ledge]

literally anyone who can afford $200/mo for an AI service

[tawny mirage]

It saves me more than that a DAY in productivity.

[elfin lava]

Cc r0b- jk jk

[pine berry]

There is always the 20$ version as well, but your better off with the free version

[elfin lava]

No hates to pro user tho, yall cool 😎

[weary estuary]

is anyone having problems?

[eternal rock]

Join the Sora Weekly Theme!

Submitted your entry yet? Don’t miss out—there’s a new theme every week, and we’re excited to see what you create!

[safe wren]

Yeah

[weary estuary]

ok word i thought i got banned lol

[pine berry]

We can't because the Chat is down stupid

[elfin lava]

Yeah I do, back problem

[weary estuary]

valid

[elfin lava]

LMAO

[sick pecan]

more than half the worlds population would think $20/mon is too much to pay for ai

[ruby acorn]

Well, you just answered why I'm not seeing the deep research feature. I forgot there was a pro. I guess plus ain't getting that.

[mossy rampart]

I had to get Team because Plus was limiting me too much

[eternal rock]

`` Server Tip: ``

Add your well-crafted prompts to our #1019652163640762428,
or share your interactions with ChatGPT in #1050184247920562316!

[elfin lava]

Not anymore!

[pine berry]

Same here a few times infact

[sick pecan]

did we just have a thanos moment?

[mossy rampart]

Might as well consider Pro, that's just 3x more cash

[weary estuary]

businesses lol

[pine berry]

Doubt it

[honest ledge]

must be nice, not all of us are running a business, some of us are just wage slave plebs trying to do the best with what our peanuts can afford

[tawny mirage]

Is Sora down?

[elfin lava]

plus user will get it just don’t know when

[sick pecan]

chatgpt got thanos'd

[weary estuary]

$20 would be fair if it wasnt censored on every single prompt

[pine berry]

I don't use Sora so, can't say sorry

[sick pecan]

$20 would be fair if i didnt have something due in 60 minutes

[wind cradle]

It may be so in your country, but in my country it is 10 times.

[tawny mirage]

I don't either. What we got was Sora distilled. It's... words I can't say here not good.

[elfin lava]

How bad is censoring on your end cuz mine doesn’t do that much ,( probably because I injected some stuff to to=bio memory) 😈

[pine berry]

You guys get censors?

[honest ledge]

it's great how GPT seems to always crash when things are due in an hour...

[frozen ivy]

is it just me or chatgpt dead

[tawny mirage]

Probably using terms in their memory that are triggering failsafes.

[mossy rampart]

How much is it in your country then? Because I pay $60 already for Team and Pro is at $200

[tawny mirage]

Yeah teams requires two min seats.

[worldly mauve]

happening to me rn

[mossy rampart]

I KNOW and I hate it

Still, could be worse. Could be 5 minimum

[toxic flax]

how many times ts gonna happen

[frozen ivy]

ya chatgpt's down again

time to report

[pine berry]

Ah

[elfin lava]

Since ChatGPT is dead, I have alternative options

[worldly mauve]

bro im studying for an exam, the timing is soo bad

[frozen ivy]

they've got the roblox october 2021 servers

[frozen ivy]

it alwaysseems to go down whenever students need it most lmao

[tawny mirage]

Then I wouldn't pay for it. I only did so when it was $30/s/mo because I used one for work and one for personal.

[pine berry]

I love how some one remebers those dark times

[worldly mauve]

i remember the dark ages

[pine berry]

Ah the good ol'days where times where simplier

[elfin lava]

What did they do

[tawny mirage]

This made me laugh. Little d errors.

[pine berry]

now everything just makes me sad

[elfin lava]

last 15 minutes

Did they do another major change again

[mossy rampart]

That's my point, $60 is... tolerable for what I get from it. But at that point, I still hit limits for 4o. If I shell out 3x the money, I can get Pro and just... not give a darn becaus Pro has no limits, supposedly

[pine berry]

I feel bad for those who are spending over 200$

[cloud pewter]

I should make a script for reenabling the bloom logo for OpenAI logos

[pine berry]

But at the same time I don't

[tawny mirage]

Hah, they already changed it to elevated.

[twilit jungle]

Ope it's down. Hey everyone I only see when ChatGPT is down 🙂

[cloud pewter]

is better than whatever they are pushing rn

[soft pewter]

is gpt down

[twilit jungle]

Welcome. Take a seat. Have a drink.

[sick pecan]

the chatbot is dead. long live the other chatbots

[pine berry]

The chat wouldn't be active if it wasn't

[neon raven]

GPT down once again ladies and gentlemen

[cloud pewter]

10x*

[soft pewter]

wdym

[worldly mauve]

imma use google bard for the mean time:(

[mossy rampart]

No, not 3x because I am paying for two Teams seats

[twilit jungle]

He means when ChatGPT goes down this place is hoppin'

[honest ledge]

Well, already crafted the 3 remaining prompts with all I need, assuming that each of them will take 2.5 minutes to generate, that's roughly 6 minutes of putting everything together in the document, saving without checking the last few sections of this 19 section document of ethical philosophy for inconsistencies, as long as it's back up before 23:52 local, i should still be okay for the deadline

[soft pewter]

wdym

[neon raven]

Less quality responses, I’m assuming

[native otter]

Someone still calling it bard 🔥

[twilit jungle]

I... what

[cloud pewter]

except for a certain memeber in here

[soft pewter]

wddym

[pine berry]

Thank you my good sir

[tawny mirage]

Trollus maximus.

[cloud pewter]

got dam whos sharing the teams with you

[worldly mauve]

better than nothing

[wind cradle]

Ah, I see. Are you in a team?
In my country, Plus is 3000 yen, and Pro is 30000 yen. It's 152 yen per dollar.

[fringe trench]

is gpt down??

[twilit jungle]

Grok 2 has gotten significantly better

[elfin lava]

Actually the new one got deep research and 2.0

[pine berry]

In other words the chat isn't as lively unless there is an outage

[mossy rampart]

Yes, in a team of me and me xd

[twilit jungle]

But 4o is still my favorite model for most things

[soft pewter]

wdymn

[cloud pewter]

Japanese strikes again(they pay cheaper than Americans but only $3 of difference

[mossy rampart]

I just hit limit for replies on 4o on Plus almost non-stop. Team has better limits, means I hit the limit maybe once per day

[pine berry]

Better than 3

[cloud pewter]

but no increase on limits for o series

[tawny mirage]

When the service called ChatGPT has an outage, the OpenAI Discord #chatgpt-discussion has a surge of activity. Otherwise this channel is fairly lackluster, and almost sleepy in response rates.

[worldly mauve]

someone @ me when gpt comes back in. I hate the gemini ui

[honest ledge]

legit, if I could switch it to claude or deepchina in the time available, i would, but too much invested and too little time left before the deadline

[pine berry]

I disliked 3 turbo alot

[gaunt kindle]

I think I got banned and I don't know why, can somebody help me?

[soft pewter]

wdym

[dawn jackal]

If you think so because your GPTs aren’t working don’t worry it’s happening to everyone rn

[pine berry]

You lost all your chats and can't acess anything?

[sick pecan]

sweet i just got an extension.

[elfin lava]

Notice how ChatGPT getting worse each day performance wise

[pine berry]

This right here

[cloud pewter]

what extension

[tawny mirage]

Oof, it went from yellow to orange error rate.

[gaunt kindle]

Thank you for telling me😭 🙏

[tawny mirage]

Next is red!

[cloud pewter]

definitely not because of o series and possibly sora

[pine berry]

Yeah that is a werid surge

[mossy rampart]

I rarely use o series, and usually swap to o-model to force it to use chain-of-thought to parse some important and tough stuff, then swap back to 4o and ask it to rephrase it. I don't like how o1 and o3 series act overly robotic and 4o has more humanity to it

[gaunt kindle]

yeah, I thought I was the only one😮‍💨

[tawny mirage]

o-series is fantastic for boilerplate code generation.

[cloud pewter]

I would subscribe to teams too if it’s available on iOS

[elfin lava]

I blame ChatGPT gov , operator- insert Here

[pine berry]

Its okay bud we all have those fears

[smoky carbon]

I've got an issue of ChatGPT web app. Nothing works properly, no history, same above

[supple fractal]

Same

[smoky carbon]

Should I report a bug or not.

[kind crater]

People really need to learn how to scroll up.....

[pine berry]

Its an outage my dude

[dawn jackal]

They know

[elfin lava]

I use o series when it comes to something that involves logic or fact, once i got the reasoning output I switch back to 4o for continuation

[eternal rock]

Get notified for server events!

Visit id:customize to pick up the <@&1261377106890199132> role.

[charred gazelle]

time to switch to deepseek

[sick pecan]

i have teams for another 4 days. then i am done with this service

[pine berry]

Heres a tip for you all watching look at the Status board

[elfin lava]

Careful! A Robert will ask you to move to #ai-discussions

[smoky carbon]

outrage far more frequently...

[tawny mirage]

Oof. ChatGPT, API and Sora are in distress state.

[mossy rampart]

Exactly. o series feel like an actual robot. Same for whenever 4o searches the web. "Hey Glitch. Here are results... Beep boop."

[wind cradle]

Japan happened to be close to America.

[pine berry]

https://status.openai.com/

[pine berry]

Use this to see what is going on

[sick pecan]

my o series talks like a philosopher

[elfin lava]

SORA IS OVER 🎊

[sick pecan]

i'd show you a screenshot except chatgpt is down

[inland estuary]

I assume everyone else is having issues too?

[primal wyvern]

is down again

[smoky carbon]

aha, openai stauts updated issues.

[smoky carbon]

20:28~ 😄

[pine berry]

Yup yup

[elfin lava]

Damn this is probably major outage they having since ChatGPT iOS integration

[sick pecan]

the system is down. the system is down, the system is down, down down down down

[cloud pewter]

let’s go

[frozen ivy]

yep time to wait another 5 hours!

[cloud pewter]

screw sora!

[worldly mauve]

its back up

[pine berry]

if they're onto it, it possibly be 50 mintues

[worldly mauve]

it was a nice one boys

[sick pecan]

every time i use sora i lose faith in ai

[pine berry]

Are you sure?

[smoky carbon]

I should have made a plugin of my agents, change api from openai to other llms when it down.

[elfin lava]

No cause I really think they adding something cuz the ChatGPT release note page

[cloud pewter]

comparison between sora and other ais