💬general

[calm stump]

yeah it still gives an error

[harsh grove]

same here

[slim fable]

I googled it is Chinese stuff lol

[teal cipher]

some kind of update?

[marble glacier]

i thought its resolved

[radiant zephyr]

https://en.wikipedia.org/wiki/Server_(computing)

[analog kite]

yeah getting the same era

[analog kite]

this mf annoying

[untold igloo]

https://tenor.com/view/henry-cavill-geralt-of-rivia-the-witcher-fuck-gif-15913480

[teal cipher]

mine is working

[fallen pendant]

try to log in

[calm stump]

it finally worksss

[marble glacier]

try refreshing

[analog kite]

ok we maybe making progress

[teal cipher]

the login button is not working xd

[analog kite]

getting to screen but no showing history or log in

[teal cipher]

nothing happens when i click it

[radiant zephyr]

Iceland will win in 20 years

[fallen pendant]

Im in

with my acc

[harsh grove]

it still doesn't work with me

[brittle summit]

Hello

[analog kite]

yeah mine is still being a dick

[brittle summit]

Is there anyone that can help? I can bad gateway error

[analog kite]

we all are dw

[harsh grove]

yeah same

[brittle summit]

wait what...? I'm not the only one screwed

[analog kite]

yeah must be down

[harsh grove]

but it is working now

[analog kite]

my first panic thought: fuck i got banned

[brittle summit]

Ok good at least we all are in the same boat

[mighty saddle]

funny how https://status.openai.com/ says it's been fixed

[fallen pendant]

No im logged in but when i try to load a chat doenst work its normal they working on it

[analog kite]

i got to a working screen, refreshed now facing same issue

[calm stump]

working for me

[analog kite]

no need to flex jk

[brittle summit]

Not for me...yet

[fallen pendant]

Bad gateway is back

[brittle summit]

It was wokring on my moms phone though

[rose lagoon]

i am facing bad gateway

[analog kite]

yeah we are all facing it

[calm stump]

guess who's back
back again

[sterile pulsar]

yes bad gateway

[fallen pendant]

On my phone

[desert dagger]

it does that a lot it says it works when it doesn't

[analog kite]

gonna take a bit to fix i think

[rose lagoon]

ohhh when will it get resolved??

[fallen pendant]

Its working on my phone

[rose lagoon]

this sucks

[sterile pulsar]

i thought its back but it isnt

[analog kite]

yeahhh all u can do is keep refreshing if u need it urgently

[rose lagoon]

is it working fine on the mobile version?

[fallen pendant]

yea

[rose lagoon]

the site is not responding on my laptop

[analog kite]

yeah it isnt on my pc either

[rose lagoon]

i keep pressing refresh and it does not respond

[fallen pendant]

The app uses different servers working for me

[rose lagoon]

i hope they fix it soon because i have a lot of work to do man

[fallen pendant]

me to

[brittle summit]

Does anyone know an estimated time when its gonna get fixed

[ripe ravine]

so the chat still down?

[brittle summit]

Si

[fallen pendant]

i would say an hour or 30 min

max

[brittle summit]

Did you use your merline powers?

[ripe ravine]

welp... it is time for a break today xD

[fallen pendant]

Exactly

[brittle summit]

Time to play marvel rivals

[teal cipher]

working on firefox for some reason, but not on chrome

[glass badge]

Is there an app that I can download on my laptop for chatgpt? Seems like the website is very slow right now

[brittle summit]

are u 100% sure cuz if so im gonna download rn

[teal cipher]

idk for me its working

[brittle summit]

Alright im gonna check it

[calm stump]

f

[undone hollow]

I have a iOS and it’s still not working

[winter dome]

unfortunately it didnt work on me tho

[calm stump]

try refreshing the page lol

[stuck finch]

Website is down :p

[stuck finch]

Not working

[glass badge]

Too bad. I have a lot of work to do huhu

[calm stump]

i didnt said it to you

[stuck finch]

Don't be so mean bb girl

[brittle summit]

You shouldnt share u IP

[teal cipher]

lol, it was working, i tried refreshing, and now its not

[warm radish]

delete that my man

[calm stump]

dont be annoying

[teal cipher]

should not have done that

[velvet turtle]

why

[calm stump]

lol

[warm radish]

u sharing your ip where you live

[brittle summit]

Bad...things...

[stuck finch]

Ur ip was visible

[undone hollow]

Where are the devs when u need them

[velvet turtle]

bro how can i solve it

[stuck finch]

Not

[warm radish]

the chatgpt being down is general

[teal cipher]

you can't

[fallen pendant]

https://apps.microsoft.com/detail/9NT1R1C2HH7J?hl=neutral&gl=DE&ocid=pdpshare

I think that can work

[calm stump]

ummm actually you cant hack people and get his address only by ip address 🤓 👆

[velvet turtle]

so i'm not the only one who has this issue

[stuck finch]

Well no shit

[undone hollow]

No

[brittle summit]

There was a time I thought I was the only one

[warm radish]

well you never sure, there is ppl out there they can do everything but you gotta a point

[marble glacier]

i think mobile one is working

[velvet turtle]

thanks god

[calm stump]

they called darkweb hackers

yeah i know about them

[warm radish]

bro is one 💀

[brittle summit]

But he has batman logo

[stuck finch]

https://status.openai.com/

[teal cipher]

someone should have hacked him lol

[random anchor]

guys chat gpt is back

[calm stump]

no

[warm radish]

i love how everyone haves the the icon new user in this server bcuz we all just had to make sure we are not only ones with the issue and came here xD

[calm stump]

check twice

[velvet turtle]

lol

[teal cipher]

joined this channel 5 minutes ago xd

[stuck finch]

Its back up

[brittle summit]

did u refresh?

[calm stump]

not everyone

[teal cipher]

https://tenor.com/view/the-avengers-marvel-jeremy-renner-clint-barton-hawkeye-gif-17148870

[desert dagger]

noobs

[stuck finch]

Oh well if u do its down again :p

[brittle summit]

bruh this aint marvel rivals

[warm radish]

still down

[cedar ridge]

Pain

[velvet turtle]

no

[teal cipher]

https://tenor.com/view/spider-man-tobey-maguire-sad-cry-crying-gif-4535637

[warm radish]

cake news

[untold igloo]

https://tenor.com/view/dark-souls-defeated-give-up-gif-5507795813322566763

[native herald]

Hey is chat gpt down ?

[teal cipher]

no

[calm stump]

server people count after the issue 📈 📈 📈

[velvet turtle]

new one

[brittle summit]

Lol

[pseudo hare]

sorry guys, we accidentally shot down the chatGPT servers instead of our bad employee

[teal cipher]

https://tenor.com/view/vince-mcmahon-gif-26506627

[brittle summit]

Shot down? As in gun down

[desert dagger]

just use this it works https://chat.deepseek.com/

[untold igloo]

https://tenor.com/view/keannu-reeves-john-wick-back-i-am-thinking-iphonedo-gif-12713477

[pseudo hare]

correct 😀

[calm stump]

and fired the one who tried to fix it

[desert dagger]

use better aim

[velvet turtle]

gemini began famos after this issue

[pseudo hare]

sadly true

[velvet turtle]

we need some good news

[untold igloo]

https://tenor.com/view/kill-stab-thinking-arya-got-gif-14033550

[calm stump]

some good news

[untold igloo]

what

[desert dagger]

the suspense

[pseudo hare]

my bad, yall. expect chatGPT to come back in 3 weeks, we're too lazy to fix this rn.

[calm stump]

he asked for good news so here it is

[velvet turtle]

noooooo

[fallen pendant]

i think the dekstop app works but i can't download it my microsoft store is broken i can't install anything.... but if you can install it try it i think it can work because it also works on mobile

https://openai.com/chatgpt/download/

[calm swan]

Deepseek killed it

[slim fable]

I just joined after server down 😂
want to know any update

[untold igloo]

https://tenor.com/view/duck-mad-angry-i-will-kill-you-gif-17283481

[chilly skiff]

Bad gateway

The web server reported a bad gateway error.

[pseudo hare]

update: 4 weeks because we're too busy in the bedroom with microsoft 😏

[teal cipher]

https://tenor.com/view/bray-wyatt-fiend-let-me-in-gif-18960524

[calm stump]

hmmm

[untold igloo]

https://tenor.com/view/blinking-eyes-white-guy-blinking-meme-what-huh-gif-26334323

[chilly skiff]

https://status.openai.com/

there is the status they are investigating whatever that meeans.

[fallen pendant]

they said first fixed and now investigating

[calm stump]

im surprised no one is saying me to censor my ip adress

[plain pasture]

they are investigating why in the living nevils is the website still down and the whole entire thing is glitching up

[pseudo hare]

no, we're just too busy with microsoft actually

[plain pasture]

their temporary fix earlier might have failed -_-

[chilly skiff]

Im just saying resort to that link to see actual progress. For fuck sake, or ask chat gpt how he feels I don´t care 🥳

[marble glacier]

killer is otw btw

[brittle summit]

I typed the message half way and realised you did it on purpose

[fallen pendant]

im working on the moblile version remote on my PC to my phone xD

[worn sigil]

yup 🥲

[calm swan]

DeeeeeeeepSeeeeeeek

[plain pasture]

we just gonna wait to wait it out,theres no use,gotta deal with no chatgpt for a while

[pseudo hare]

true, go touch grass

[desert dagger]

https://chat.deepseek.com/

[marble glacier]

Gemini user 📈📈

[brittle summit]

I literally am

[calm stump]

sad reality 😔

[chilly skiff]

In my phone works. It doesnt save logs or lets me in. But works.

Can put screenshot.

[calm stump]

im not allowing you to do that

[untold igloo]

bruh, how do it works if it not let you in

[chilly skiff]

In as "log in" in my account. Mt chats.

[calm stump]

everyone needs chatgpt nowadays 😭
i think we need to finally touch grass

if the universe will collapse it will be my fault im sorry

but we cant do anything 😭

[untold igloo]

https://tenor.com/view/nope-nope-nope-gif-11239163144514157923

[marble glacier]

can grass do my assignments?

[calm stump]

maybe

[untold igloo]

can grass help me with my story arks

[calm stump]

ofc

[marble glacier]

ofc

[brittle summit]

what..

How are u doing this

[chilly skiff]

In mobile! I just said it!

[calm stump]

wtf is frank fart 😭 🙏

[marble glacier]

mobile app is working

[brittle summit]

Are u using with your acc logged in?

[marble glacier]

for me yes

[calm stump]

let me install ldplayer rq

[chilly skiff]

Acc doesnt log it. It is as it is in the screen. (For me at least, no access to chats, no history no buttons)

[velvet turtle]

I have good news!!!!!!!!!!!!!!!

[brittle summit]

uh well if thats the case it works like that in pcs too

[calm stump]

bruh

[brittle summit]

is it working?

[calm stump]

why

[plain pasture]

25 mins of them still investigating,what are yall gonna do in the meantime while this gets settled

[velvet turtle]

yes, but not for me : (

[marble glacier]

im gonna touch some grass

[chilly skiff]

Stay here and complain we cant work.

[brittle summit]

wat was the good news

[calm stump]

yes

[brittle summit]

Im here lookin for friends to play marvel rivals

[fallen pendant]

Mobile works fine

[undone hollow]

The website still isn’t working on mobile but it seems the app is working just fine

[velvet turtle]

try make a surf on my github while they solve chatgpt problem : )

[golden furnace]

Is chatgpt down ?

[calm stump]

no its problem with your provider sir

[undone hollow]

But I can’t download the app it says I need to require the 17.0 update

[velvet turtle]

restart your pc

[marble glacier]

buy new pc

[velvet turtle]

jailbreak your iphone

[plain adder]

chatgpt isn't working for me rn, what do i do i rlly need it

[marble glacier]

mobile app is working

[golden furnace]

I got bad gateway error i'll do a restart

[chilly skiff]

Try phone version

[velvet turtle]

sell your device and buy a new one

[chilly skiff]

I can confirm in my phone now lets me log in. Website stays the same. TRY THE PHONE IF YOU REALLY NEED IT.

[plain pasture]

I got an update my chatgpt buddies

[tame lodge]

chat

[calm stump]

gyatt

[tame lodge]

is it working for u guys on desktop

[marble glacier]

nope

[velvet turtle]

yeah

[calm stump]

no its fucked up

[fallen pendant]

Im in

[tame lodge]

i thought i was the only one

[desert dagger]

its over

[fallen pendant]

It works on pc with VPN for me

[chilly skiff]

wHAAAAAT?

[plain pasture]

they found the root cause of why chatgpt is crashing out,relax and wait it out,use other ais while waiting for chatgpt to be brought back

[fallen pendant]

Yea

[tame lodge]

i thought chatgpt does not allow u to use with vpn

[chilly skiff]

Is it located based? can anyone else confirm? because it was giving cloudflare errors based on IP

[plain pasture]

the fix earlier was a temporary fix,now its a root cause fix

[undone hollow]

I’m downloading 18.2.1 update rn

[calm stump]

chatgpt doesnt care

[velvet turtle]

downgrade it

[brittle summit]

another one

[tame lodge]

brother u leaked ur ip

[chilly skiff]

GUYS YOUR IP!

[calm stump]

whats wrong with it

[velvet turtle]

he is an indian

[brittle summit]

or an american grandma

[tame lodge]

someone can look for ur location with it i think

[plain pasture]

if you send it here,bad things may happen,the internet cant really be fully trusted sometimes

[velvet turtle]

guys finaly

[calm stump]

im still waiting for the killer

[chilly skiff]

What is wrong with doxxing your IP adress to the public? xD

[velvet turtle]

it works

[fallen pendant]

yea it works

[tame lodge]

ima check back in an hour

[fallen pendant]

i get answers from GPT in web and logged but i use an VPN

[calm stump]

nw

[tame lodge]

it still aint working for me

does copilot ai have image limit

[chilly skiff]

Can anyone else confirm it works with VPN? or not?

[tame lodge]

works

[brittle summit]

Not

[calm stump]

still waiting

[tame lodge]

sometimes works sometimes not its annoying

[velvet turtle]

wait

[tame lodge]

what is the image/files sending limit on chatgpt premium

[calm stump]

aw fuck its down again

[pseudo hare]

lol

[patent linden]

it was always down for me

[tame lodge]

it works on phone tho

[undone hollow]

Same

[chilly skiff]

Dont stay here waiting for updates: https://status.openai.com/

Check that. If its all good (and it works on web.) It´s done. If is not, whatever.

In phone works well, for most of people, try that.

[calm stump]

there is no canvas shit on the phone 😭

[plain pasture]

8 mins and still working on the root fix

[tame lodge]

what happened on december 11th bruv

[brittle summit]

my best friend died on november 2nd

[velvet turtle]

you mean 11th september

[tame lodge]

no in the last 90 days thing

[tame lodge]

all ai's were down

[pale badge]

I'm still standing!

[patent linden]

it works now

[calm stump]

guess the date by 2 emojis: 🏬 ✈️

[pale badge]

Not for me

[tame lodge]

man ima just go to copilot for now

[steep pier]

dude i uploaded a huge fucking file and now it broke

[patent linden]

🌟 Once upon a womp, in the mystical land of wompdom... a mighty WOMP WOMP echoes across the universe, just for you, Ilyassan! 🌟

[steep pier]

it dosent work for me anymore

[brittle summit]

Can we talk about how gpt sometimes and I mean sometimes gives very stupid responses. Does anyone know why

[patent linden]

(chat gpt)

[pale badge]

BRUH

[calm stump]

remember the 1 hour russia internet downtime?

[steep pier]

he feels a tad bit silly sometimes

[calm stump]

they fell by 18 million dollars

[steep pier]

i can relate

[calm stump]

what will happen to openai 💀

[chilly skiff]

Because you don´t actually give a proper input. If it offers you a wrong answer, it´s mostly on you. Also, it´s not perfect and doesn´t pretend to be.

[steep pier]

i WAS gonna buy chatgpt+ but i changed my mind cuz its down

(totally real)

[pale badge]

Bro, I paid gpt+ for this

[brittle summit]

Yea I guess im dumb too

[steep pier]

but like genuinely why does chatgpt get downtimes the most?

like ive seen downtimes for other platforms but its never as bad as chatgpt

[calm stump]

literally what altman would say

[obtuse rivet]

lmao what

[desert dagger]

because https://chat.deepseek.com/ works

[velvet turtle]

gpt from future?

[calm stump]

bro its the smartest ai that human ever had and they run this shit for billions of users, what'd you thinked?

[steep pier]

fym lmao what u heard me

[steep pier]

smh other websites have ai also

[obtuse rivet]

it almost never goes down. go use deepseek for your homework

[plain pasture]

18 mins,they're still fixin it,root issue might be bad

[brittle summit]

What do you think the issue is?

[obtuse rivet]

this represents multiple months lmfao

[storm olive]

no chatgpt = can't program 😦

[steep pier]

i never said it didnt

[obtuse rivet]

https://chat.deepseek.com/ enable "deepthink"
its better than chatgpt

[chilly skiff]

It´s not about you being dumb. Its again, pretending like chat GPT is human. It´s not. It´s a tool that can pretend to care for you. Blaming it for bad answers is like saying Stable Diffusion draws bad because you can´t prompt.

If you are curious, my first order for chat GPT is "Stop pretending to be human, put the information, examples and remove human emotion from your interaction." on every chat, it helps.

Also... It´s been less than a day down? Hello? 🥳 Are you all THAT adicted to the tool? Still counting minutes? ...

[calm stump]

ai doesnt work = ai wont replace our jobs

[plain pasture]

idk,maybe a bad issue within the servers or smth

[undone hollow]

Wait is chatgpt gonna be down for months?

[velvet turtle]

I think they are cyber attacks.

[steep pier]

try to check bigger platforms uptimes, or even platforms like chatgpt. they dont this have this much problems

[obtuse rivet]

??????? what, no

[storm olive]

lemme see

[undone hollow]

Phew

[brittle summit]

I think they just wanna see how we are panicking

[pseudo hare]

yes, we do

[brittle summit]

This is the main server right? Im suprised only just the few of us are chatting. I thought a lot would be using this

[desert dagger]

everyone is crying

[brittle summit]

or......everyone else is just read us

[calm stump]

if this is real they're dumb asf and not only f

[desert dagger]

wow

[velvet turtle]

earth is flat

[obtuse rivet]

this is the unofficial server. we already get a million people crying here everytime chatgpt goes down

[pseudo hare]

and it's funny

[velvet turtle]

where is the official server

[obtuse rivet]

gg/openai

[desert dagger]

chatgpt speedrun

[velvet turtle]

https://tenor.com/view/dubya-gif-23030610

[brittle summit]

I remember this scene

"Reading makes a country great"

[velvet turtle]

👀👀

[plain pasture]

speaking of reading,i still read a tad bit from time to time,just looking at my old childhood space book

[desert dagger]

sir the second llm has gone down

[chilly skiff]

Wall-E works on phone too. Any suggestions? 🥳

[velvet turtle]

https://tenor.com/view/george-bush-angry-shoe-throw-dodge-gif-5298606

[plain pasture]

give it some random questions

[undone hollow]

Oh my god deepseek is actually so incredible, the way the bot first analyzes your message and information and then responds with a actual message
It’s much better than chatgpt idc abt chatgpt anymore

[velvet turtle]

remember this scene?

[brittle summit]

Nice dodge. I would have taken it to the face

[chilly skiff]

It feels better. But it´s only in English no? (Not a problem for me but...)

I really like how it actually gives you the scripted thinking before.

[desert dagger]

it speaks a lot of languages

[velvet turtle]

come to the voice chat

i will make a stream for latest updates

[plain pasture]

45 minss,still workin

prolly will see if it works fully again when i wake up from sleep,its almost 11pm here and i am burnt out

[ripe ravine]

still down?

[brittle summit]

bruh its taking too long

[fallen pendant]

All works for me

[tame lodge]

still down

[fallen pendant]

i deleted my coookies

and then all worked for me

[uncut stirrup]

the app works but the site isnt

[fallen pendant]

I deleted my cookies im using a VPn and the site is working then

denmark server location for VPn then it works for me other location isnt working but i testest only 5

[chilly skiff]

Hellos, Im using it now in website too and seems fine. Official web says:

Update - We are continuing to work on a fix for this issue.
Jan 23, 2025 - 06:34 PST

Identified - We have identified the root cause of this issue, and are currently working to implement a fix.

Jan 23, 2025 - 05:43 PST

Investigating - We are currently experiencing elevated error rates for ChatGPT. We are currently investigating.
Jan 23, 2025 - 05:12 PST

[brittle summit]

Its back online

[velvet turtle]

yeeeeey

i think i'll stay here

I think it's a shame that we leave the server after receiving help.

[uncut stirrup]

ikr

Ill be your cat

[lapis canyon]

I wonder why ChatGPTs platform seems to be so unstable. They seem to have outages fairly frequently compared to equally sized firms and they don't even have a decent "well everything broken" landing page with status details

[mint heart]

i need some advice to get chatgpt(or any other ai) to organize some code without changing it. I have a paragraph blob of code and I'd hoped for chatgpt to make it nice and more readable for a code editor. But it keeps changing the code.

[velvet turtle]

do you use visual studio code?if yes, select the part of code and choose copilot

[mint heart]

yea i do

lol chatgpt gave me back this at the end after a third try: // The rest of the code continues similarly, properly formatted...

-.-; i'll try copilot

[velvet turtle]

try it

click here to see it

[mint heart]

same thing and it seems copilot uses gpt 4o? looks like I have 2 choices gpt or claude sonnet in the little drop down menu

maybe claude can organize it without changing it

[velvet turtle]

give it a chance

[mint heart]

i mean I asked it and it changed it again lol

[velvet turtle]

give me this code i'll try

[mint heart]

https://unpkg.com/phenomenon@1.3.0/dist/phenomenon.iife.js

[velvet turtle]

You want the code to be organized.

[mint heart]

yea

no edits

[velvet turtle]

hmmm like this?

[mint heart]

like that it starts out fine sorta i think

[velvet turtle]

how did you do it

[mint heart]

that's the first half of the claude ai, the middle and end is changed though

I'll just manually change everything back, and use what the AI gave me as a base I guess

[velvet turtle]

wait

i did it

[mint heart]

it's some really cool code I found, I wanna uhm deconstruct it and see if I can make it my own <<;

https://codepen.io/cvaneenige/pen/QBwbEY

[velvet turtle]

here you go

[mint heart]

that last part of the code I'm organizing is an additional library you need to get it working

[velvet turtle]

try it if it works

[velvet turtle]

and if you want to ask chatgpt in the future to recognize your code,jut past the code and past this prompt
Do not change any character in this code, just arrange it by adding spaces, enter and .tabs.

[mint heart]

i think it's right lol thanks

[velvet turtle]

welcome

[static sorrel]

I have a .xlsx all set up that I wanted chatgpt (o1) to help fill in some formulae for, but apparently I can't upload or have it read a .xlsx. I feel like it USED to be able to do that, but maybe I'm going nuts. Is there any way to get ChatGPT to interact directly with a .xlsx? Or maybe I can give it access to a google drive sheets version of the spreadsheet or something?

[velvet turtle]

is that a microsoft exel formula?

[static sorrel]

A .xlsx is a microsoft excel file.

[velvet turtle]

try screenshot the table lol

maybe it can works

[static sorrel]

lol, that actually HAS worked for me in the past :). However, for this one, I need to do a bunch of statistical analysis with a what-if analysis of various input distributions, then to generate a heatmap of the output effects on a system. I need help crafting the outputs. I can do the math myself its just a day long activity, and I'm hoping chatgpt can turn that into a half hour or less.

[mint cosmos]

Deep see is stupid, but at least it's easy to be jailbroken

still it's too passive grammar

[velvet turtle]

maybe I could help you if you want

[static sorrel]

that's why I'm here.

[velvet turtle]

Okay, let's share the tasks.

[static sorrel]

what.

[velvet turtle]

Bro what exactly do you want to do

[static sorrel]

I literally bolded that in my first comment.

[tidal vortex]

Can I tell gpt to change the way its formatting lists? These are not good copy pasteable

putting it in memory didn't help

[velvet turtle]

i ask chatgpt and it told me that it can,just share the file with it

[static sorrel]

Ok, it can't though, so that isn't helpful lol

[velvet turtle]

maybe if there is a plugin on exel

[static sorrel]

maybe. That's why I'm here asking.

[gloomy burrow]

Agents are here

[tidal vortex]

Can I tell gpt to change the way its formatting lists? These are not good copy pastable
putting it in memory or custom instructions didn't help

[uncut island]

What have you tried, what did you expect that would be the output and what was the real output?

[tidal vortex]

I put this:

Format lists as follows:

• Use normal unformatted „- „ for lists.
• Use two spaces „ „ for each level of indentation.
• Use „#. „ for numbered lists.

I hoped it did what it says, but the formatting didn't change at all and I get the standard list formatting with tab indets and custom round bullet point.

[vivid crypt]

hi everyone! im not sure why by chatgpt seems to kick me out of the server for some reason, it has done this to me once before where all of my recents chats on the left side are gon and everytime i type in a new prompt it doesnt generate anything. any help would be greatly appreciated. thanks

[late estuary]

quick question: considering chatgpt remembers what I asked it previously, can I give "global" commands like "no math until I tell you to"?

[uncut island]

Try to add an example or two of output(s) to your prompt and see if it works.

[vast valley]

Yes! Click on your name, then click "Customize ChatGPT" 😊

[late estuary]

does this only work in discord or also when using chatgpt in a browser?

[vast valley]

Idk about discord, but it works when using it in a browser and I think also in the app

[tidal vortex]

What I put there:

Format lists as follows:
- Use normal unformatted „- „ for lists.
- Use two spaces „  „ for each level of indentation.
- Use „#. „ for numbered lists.

Example:
- This is a list bullet
- Second list bullet
  - Sub list bullet
  - suub list bullet

Output:
• Lean Proteins:
• Chicken breast
• Turkey

[late estuary]

thanks!

[vast valley]

Ofcourse, have fun! 😄

[uncut island]

I don’t believe you will be able to see such custom output format in the HTML of ChatGPT…. Let’s try again but now give the instruction “always present any list inside a code block”.

[tidal vortex]

ah that did work

[uncut island]

Also, add a second example for thr numbered list.

[tidal vortex]

but now its not really readable xD

but I could ask it whenever I need to copy it to make it in a code block first

[uncut island]

Yeap.

[tidal vortex]

code block is a great tip, that works great

[fathom cape]

WHO GOT ACCESS TO OPERATOR?

Can you use it on any website or only pre-emptive ones they have?

[copper storm]

Okay operator seems pretty promising.

[fathom cape]

I see it on my ChatGPT screen, but can't access it yet

REEEEEEEEEEEE

[ripe quest]

I do not yet have a ChatGPT Plus subscription. I have been trying out various Free AI offerings including Microsoft CoPilot. I have been using CoPilot to stay up-to-date on various medical topics including pathophysiology and pharmacology while I am driving. One feature I like, is to have CoPilot create a quiz for me covering the specific topic I have been reviewing and then allowing me to take the quiz all via voice chat.

I tried doing the same thing with ChatGPT but it was not able to create a quiz that I could then take. It would come up with questions and then multiple choice answers but it would not let me guess the correct answer. It would just read what the correct answer was after listing all the possible choices.

Microsoft CoPilot costs the sme as ChatGPT ($20/month) but I wold rather get an.API key for ChatGPT if it could accomplish this task. Does anyone know if ChatGPT can accomplish this?

[calm stump]

liar where is the killer

[copper storm]

It should absolutely be able to do this, make sure you're using a powerful model.

[ripe quest]

Thanks for the reply @copper storm . I am afraid I am not sure how to choose a model in ChatGPT

[copper storm]

At the top left (on pc) and the top right (on mobile) you can switch the model you want to use, if you have Plus you should have access to the best like o1.

[ripe quest]

OK thank you. I have not purchased a subscription yet so I am just testing free access.

[copper storm]

Oh yeah actually you can't switch models on the free plan, you only have access to 4o until you hit your usage limit and then it reverts to 4o-mini.

You might be using 4o-mini right now, which is not that good.

[ripe quest]

If I subscribe to Plus ($20/month) is there a limit to the conversations I can have? Do they time out after so many minutes?

[fathom cape]

voice convo?

[ripe quest]

Sorry I don't have access to voice at the moment

[fathom cape]

I'm asking, what limitations are you referring to?

[copper storm]

Yeah there is, but the limit is pretty high, you're unlikely to reach it if you're just using it casually.

[ripe quest]

Sorry ... Yes .. voice conversation

[copper storm]

The limit for voice covo is about an hour or longer per day, i was referring to messages in my previous message.

[ripe quest]

Thanks ... that works for me

[fathom cape]

Yeah I mean it's month to month, so try it out. Only $20

[ripe quest]

I use it when commuting

[ripe quest]

Good point. Best thing is to try it

[fathom cape]

Yeah I mean it's borderline super intelligence for $20/month

[ripe quest]

There seems to be a lot more offered in ChatGPT as compared to CXoPilot. So many "mini" apps within ChatGPT that are focused on specific topics

[fathom cape]

You even get 50 o1 requests/week and 50 o1-mini /day

co-pilot is also openai, but thier shitty models

[ripe quest]

Is a single o1 request the same as one question?

[fathom cape]

Microsoft is using their integration to lure people to purchase copilot

Which again is fed into the sameish model

Yes, 1 question. So if you follow up with an additional question that's 2 requests

[ripe quest]

Could you suggest a website to learn more about ChatGPT? How to use it properly?

[fathom cape]

o1-mini is so good w/ math and coding. o3 mini is coming out any day now

Just watch some youtube channels, but be cautious that the o1 series requires different prompting. Let me find this easy to understand guide for o1 type models

[ripe quest]

My use case is more reviewing medical topics for USMLE

[fathom cape]

Do you have access to the content on your computer? Could you copy and paste entire chapters into the models? Are photos important for those topics?

[ripe quest]

I have access to many medical texts but I have never input anything into ChatGPT.

[ripe quest]

For example ... I will be driving and would ask 1) Please give me detained information on Goodpastures Syndrome (just an example).

[fathom cape]

Ah

If it's very general info like that, you'll be fine

[ripe quest]

Then after asking some more questions I might ask for a quiz to test my knowledge on the subject.

[fathom cape]

Yeah you should be fine, but test it out on subjects you already know just in case

I'm always a bit weary when it comes to voice and accuracy. If this were text with o1 or o1 pro mode, that's pretty much solved

[ripe quest]

Thank you very much for all your time

[copper storm]

Yeah i think you're very likely to get more out of a ChatGPT subscription than a Copilot one.

[ripe quest]

I think I will take the advice and just try it out

Just $20

[static sorrel]

Operator only available for Pro, not Plus? Why do I keep paying for this thing, it can't work with Excel, it limits my o1 msgs... bah.

[wet copper]

GPT 4 diagnosed my stroke in real time. based

[static sorrel]

https://tenor.com/view/swing-jaco-van-zyl-olympics-golf-hit-it-gif-22547643

[stoic sleet]

4o isn't working with excel files for you?

[static sorrel]

It won't let me upload .xlsx files, nor will it generate them. It also won't work with such files if i put them into Google Drive.

[stoic sleet]

I see I mainly use csv files in excel so that makes sense

have you tried ai studio?

I know they have code execution

[static sorrel]

I have some formulae that I want fixed up. I copy pasted the raw text and chatgpt figured it all out, but the process of fixing up the spreadsheet is manual and will take quite some time to implement. I just want chatgpt to give me the file, as it already knows what to do, lol

I have not tried ai studio, and I'm not sure I have the time to learn :(. ChatGPT is blocked at my job, so taking files OFF of my work computer, having chatgpt play with them, then putting them back on, is a huge pain. As such, I rarely do it. Plus I always have to strip those files of sensitive data first. bah.

[wet copper]

This Discord is pretty good for learning about it. I recommend STRONGLY against YouTube – I use it for everything else, but most of the AI videos on it are trash. OpenAI has whitepapers and other documentation, you can ask ChatGPT about language models in general for more narrow or personal questions, though it has no metacognition and will largely hallucinate any answers about itself.

You'll run out of advanced voice eventually if your commute is long but it's a fine way to study. The first thing you should do is write very clear custom instructions telling the model who you are, what you know, what level of detail and complexity it should respond at, etc. It knowing I have a science and medicine background radically changes the kind of answers it's willing to give on some subjects

It'll match your tone as well, so if you use medicalese it tends to infer you don't want a Wikipedia tier answer. It's biased toward giving relatively short and ostensibly complete answers too

[ripe quest]

Thank you!

[stoic sleet]

The first thing you should do is write very clear custom instructions telling the model who you are, what you know, what level of detail and complexity it should respond at, etc. It knowing I have a science and medicine background radically changes the kind of answers it's willing to give on some subjects
This is interesting I've never done this. Is there a template?

[rigid compass]

chatgpt tasks failed at the first two tasks I tried with it (monitoring web pages for certain information). When I tried to give it additional information about what to look for, I got the message, "Unfortunately, my automation tool for ongoing monitoring is currently unavailable." followed by a recommendation that I should check the website myself. Gee, thanks.

[brittle summit]

Yeah chatgpt tasks used to gaslight me the first few times I used it

[wet copper]

No, you don't really need one. I posted my instructions a long time ago but they're extremely specific and wouldn't give most people good results at all. I just listed my fields of interest, background, a statement on what tone and level I want to get back, also noted that I like colloquial tone, humor, slang, acting more or less like a person (i.e. don't give me bullshit about being a robot when I say "what do you think")

I have a set of loose guidelines but they were for writing a quick persona prompt, custom instructions prompt you for what to put there, then A-B test until you eliminate unwanted output and see wanted output the output you want. Positives (do this) are better than negatives (don't do that)

I recommended putting level of expertise or closeness to fields in your wheelhouse and encouraging it to use abbreviations or jargon rather than explaining everything. Also recommend having it ask follow-up questions, though it tends to take this very literally and will ALWAYS ask a follow-up, even if it barely makes sense to do so

I mean I'll help you workshop CIs if you want, or if you feel like writing a template, go ahead. I used just plain english statements in mine and I've barely changed it in the better part of a year

[boreal latch]

https://ilsforpresident.com/lmsysorg/status/1881945619704779228/photo/2

https://ilsforpresident.com/skirano/status/1881854484416201128

$15B would've been enough to build a campus that can train a model in 3 months that would have 100X more training compute than GPT-4.

https://youtu.be/3KcJI-SRjVo

literally verbatim says he doesnt care about the economy and only wants to talk about attacking people politically while cutting hannity off repeatedly.

eggs are almost 50% more expensive than they were a wee ago because of the taxes

food deliveries are 2x as expensive

gas is 50% more expensive, this is what you voted for "maga"

[cedar yarrow]

bro i need help

how do u avoid ai detection like originality,ai

[wanton raven]

don't use AI to cheat on your homework

[south prism]

operator is usless and cant do a simple prompt

[tidal vortex]

Is Canvas available for the macApp alerady? Change log says so, but how to access it?

[south prism]

I use ai to cheat on my homework to become a nurse

my friends use it for the bar

[wanton raven]

we're entering a new dark age

[brittle summit]

of what

[stoic sleet]

The jobs we’re most likely to lose are creative jobs. Writing is disappearing with over 80% search results using AI. Both are reminiscent of the dark ages

[stoic sleet]

Dr House 👀

[brittle summit]

😦

[brittle summit]

Whatever happened to your sabrina carp pfp

[stoic sleet]

I swear I told you the story it's tiffany collier

[brittle summit]

idk I forget a lot of things

[stoic sleet]

and I forgot why I changed it

[brittle summit]

Waiting for AI to remind me of things for me

[stoic sleet]

tell chatgpt to remember wearifulpoet's pfp was for tiffany collier not sabrina carpenter

[brittle summit]

Still waiting on the limitless pendant 😦

[stoic sleet]

beat me to it

[brittle summit]

I literally forgot how long I was married for

I thought I was married for 1 year when it was actually 2 years

I had that realization few days ago

[stoic sleet]

hey the more you know right?

[brittle summit]

I will prolly forget in a month lol

[copper storm]

No way there's Sackboy.

Chatgpt buying fries.

[brittle summit]

@stoic sleet look at this lol

https://x.com/brianroemmele/status/1882436734774043055?s=46

Brian Roemmele (@BrianRoemmele) on X

Folks, I think we have done it!

If overnight tests are confirmed we have OPEN SOURCE DeepSeek R1 running at 200 tokens per second on a NON-INTERNET connected Raspberry Pi.

A full frontier AI better than “OpenAI” owned fully by you in your pocket free to use!

I will make the Pi

X · Jan 23, 2025, 2:34 PM

[stoic sleet]

beautiful

[worldly grove]

Apo11o nice. spread that everywhere

[brittle summit]

@dark void keep your buddy in off topic

[dark void]

politics (and anything else that isn't AI) in #💬off-topic please

[stable marlin]

where do i talk to chat gpt at

[dark void]

#🔥gpt-5

[stable marlin]

thanks

[boreal latch]

the grok thing isnt political, its about grok

[brittle summit]

😑😑😑

[dark void]

you can relate every single thing elon does to grok because he owns it but this isn't the channel for talking about elon being a nazi on twitter or whatever. #💬off-topic is already people talking about elon 23 hours a day, it doesn't need to be in here too

[boreal latch]

it should be, this is purely about alignment and benchmarks

they cant get it to be better at benchmarks without truthful facts in it

[dark void]

if you wanna post something about grok 3 benchmarks in here, that's fine, but we don't need updates every time elon says something that gets people pissed on twitter or grok generates an "unsafe" image

this is not the channel for sharing every update where grok does something goofy ok? stop. go to #💬off-topic

[brittle summit]

.

[boreal latch]

go to #💬off-topic message for context folks.

[worldly grove]

i'm new to discord where do i interact with chatgpt? I have a web account

but if I do it on discord can i get the more advanced gpt? also it's more anonymous i think?

if anybody is bored plz dm me and show me around. i'm new to Discord and need help navigating

[dark void]

the main website is https://chatgpt.com/ but we have some bots here for a variety of stuff like chatting, making images, making videos, etc.

[worldly grove]

cool thanks trent. I'm familiar with irc but not discord

i have an account at chatgpt.com but i don't like how they remember everything i searched and it's affiliated with my email. there are irc rooms with gpt bots and interact with anybody in the channel. is there anything like that here?

[stoic sleet]

https://chat.openai.com/?temporary-chat=true

bookmark this page or set it as your home page

[worldly grove]

omg ty so much

does that still give you a certain amount of free high-powered GPT? @stoic sleet

[stoic sleet]

Yes it just defaults to temporary chat mode (which you can set for each chat manually if you wish) and then it won’t save

No other features or limits change

[worldly grove]

how many free pro searches do we get? does it keep track by IP address or can i just clear cookies to get more

[stoic sleet]

Aren’t you logged in with your email?

It’s tracking with that but just not saving your history to your email

[worldly grove]

logged into chatgpt or discord?

[stoic sleet]

ChatGPT

[worldly grove]

yeah. i made an account with an anonymous email

but instead of always logging in and being tracked I'd rather just play with chatgpt through discord

even better anonymity

well maybe not but whatever

[stoic sleet]

Have you just looked into running models locally?

[worldly grove]

i have shitty hardware i'd rather not

[stoic sleet]

If you want to do the privacy thing it’s worth it

[worldly grove]

my laptop is a $75 refurb ... integrated intel graphics 😦

[stoic sleet]

Okay I run decent models even on my phone so you don’t need a lot of power

[worldly grove]

yeah i was just trying to find a way to get pro tier without paying for it

[stoic sleet]

Just use a free llama 3.3 api

[worldly grove]

i can't do that by finding a discord channel with a bot?

[ripe ruin]

Hello everyone!

I work on writing reports, where I analyze an image, derive insights, and describe them.

I want an AI to be capable of perfectly replicating my writing style. It doesn’t need to interpret the image. I can extract the data from the image and structure it into a standard text format (I’ve already done this).

I tried providing prompts to the GPT assistant, and I also attempted fine-tuning it using my reports (I only have 150 reports). I ran some tests using 100 as training data and 50 as validation data, or even using all of them as training data, but the results weren’t satisfactory. With fine-tuning, it replicates my writing style to about 70%, but it doesn’t analyze the data the way I do. GPT, on the other hand, performs a good data analysis but doesn’t write like me. I also don’t know how to configure parameters like epoch, batch size, and learning rate multiplier. I’m completely new to this field.

It’s not a complex type of report and follows a standard format.

What would be the best way to achieve what I want? Does anyone have an idea of the right path to take?

[worldly grove]

you could just wait for them to make a non-retarded version; that'd be easier 😉

[desert dagger]

yes they scared of r1

[gaunt oriole]

https://youtu.be/AbScDl7j03o?si=xE5CbQbRtkZZCeYb

[pliant blaze]

thats cool keep me updated please

[stoic sleet]

I’ve tried the 1.5B model they’re using here at a higher quant. It’s really only functional at math and even then it’s pretty week

[calm stump]

guys can someone help me with chatgpt to make it finally replace some keywords in 278 lines canvas?

[stoic sleet]

That’s too big for canvas

[calm stump]

so what the limit

[stoic sleet]

Roughly 300 before it is useless for me practically

[tidal vortex]

Is Canvas available for the macApp alerady? Change log says so, but how to access it?

[brittle summit]

e

[brittle summit]

In the tool box

[tidal vortex]

what is the tool box? 😄

[brittle summit]

Bottom left next to files and stuff

[tidal vortex]

so not on the app yet? 😦

[calm stump]

this thing

[tidal vortex]

That's only available on web for me, not in the mac app 😦

[calm stump]

try checking the models maybe

[tidal vortex]

nope, again on web I see it, but I thought it was released already for the macApp

here from their change log site... but how

[calm stump]

changelog may say where is it then

[tidal vortex]

there is even the canvas option in my app settings, but I don't have the option to use it.

I also on the newest version of cause

[tidal vortex]

reddit for rescure... apparently there is no UI button for that yet 🤦‍♂️

[coral palm]

meanwhile chinese companies open sourcing state of the art cot models

how are they not based af?

[brittle summit]

It is funny thinking the CCP has localized models

[twin mason]

is it normal to use chatgpt like its ur friend

[flat thistle]

why is it doing this

i only attached like 3 things so far in that chat, and they were not big files

OH it was because i was on 4o mini 😭

i also got plus yesterday heh

[blissful valley]

hello

[subtle sable]

I will soon be working with my company on AI. My intent is to take a crawl, walk, run approach and move through phases based on... things to be decided in the future. This is a small company, so I will benefit from being able to be less formal. My initial step before all of this though will be to form an internal working group and help expose them to AI. After everything I've learned, my current objective is to convey that AI needs to be thought of as a competency, not a solution or particular tool. To do that, I intend to lead the working group through some exercises to see the strenghts and weaknesses of LLMs. I will give them "homework" that will (hopefully) encourage them to expore and iterate. After a couple of months, I'll then lead more focused discussions that will become more and more specific to our company and how we can repeat the aspects of worked to the rest of the company. From there, we can break off and start to explore more niche solutions that may be industry specific. I'm curious what you all think of this approach. Any suggestions?

[uncut island]

https://x.com/akshay_pachaar/status/1882406725313311065

Akshay 🚀 (@akshay_pachaar) on X

This app can scrape videos from the multiple YouTube channels and report trends & insights.

Tech stack:

- @crewAIInc for multi-agent orchestration.
- Bright Data for reliable web-scraping at scale.
- @Streamlit for the UI.

Here's a quick demo of what we're building:

X · Jan 23, 2025, 12:35 PM

[rare bobcat]

Hi guys, where can I post about finding a person to join my team space

[green sparrow]

In an interview, an OpenAI employee kind of insinuated that ChatGPT's Mac and Microsoft Store apps have a 'see my screen' feature on Advanced Voice Mode... But it doesn't 😤

[stoic sleet]

they do

at least mac does

[green sparrow]

Oh really? Not on the Windows 11 version

That makes sense he would say that then. It was Kevin Weil I think. And he didn't talk about it that long so he probably didn't think to clarify that small point

I'm sure it will be out soon. It just sucks that they seem to not have symmetrical updates among the two operating systems' apps

[low belfry]

I guess AGI will be texan!

[green sparrow]

That's interesting. Do you think it will be solar energy or nuclear fusion

[fathom cape]

I can't fucking wait for o3 mini to come out. Takes forever getting work done waiting on o1-pro-mode every time

[radiant zephyr]

Copilot is amazing

[fathom cape]

I'll take o3 mini at 85% of o1-pro capabilities

[low belfry]

you raise a good point. didnt texas suffer horrible grid outages a couple years back?

[green sparrow]

Not sure. They just need a really sunny state though probably. Texas might not be bad

[low belfry]

those huge flat roofs look prime for solar. but this is the state of energy in TX

[radiant zephyr]

It corrects itself without long sentences, unlike core chatgpt, if you've noticed

[low belfry]

nuclear fusion isnt a feasible energy source actually. not yet I guess

[green sparrow]

Oh you're right. My bad

I just know Sam is always talking about fusion so I just guessed it could be that

[radiant zephyr]

Thermodynamics book => Nuclear energy uses fission to generate heat, creating steam to drive turbines producing electricity efficiently.

[low belfry]

yes fission is established but they said fusion

splitting atom is easier

[radiant zephyr]

Ohh

They have no idea what they're talking 'bout.

Press?

[low belfry]

if I remember right, he invested in a company that makes mini fission reactors. that could be cool but a bit scary!

as far as i know, fusion is theoretical but getting more feasible every day. when it makes power for more than 10 seconds is world news

OKLO is the company making mini reactors. their stock is on the uptick i guess people think it will help stargate

[uncut island]

People will find ways to make more money with Operators... clever idea huh
https://x.com/emollick/status/1882612710053204126

Ethan Mollick (@emollick) on X

Next big thing for brands: knowing what brands agents prefer.

If you ask for stock prices, Claude with Computer Use goes to Yahoo Finance while Operator does a Bing search

Operator loves buying from the top search result on Bing. Claude has direct preferences like 1-800-Flowers

X · Jan 24, 2025, 2:13 AM

[shut solstice]

https://x.com/ElectricBoo4888/status/1882175922264330254

Electric Boogaloo ⏩⤴️ (@ElectricBoo4888) on X

In collaboration with @RightToCompute, I’m petitioning Congress, and all state legislatures, to propose the Right to Compute Amendment.

Your signature would be much appreciated!

https://t.co/YmdjdD3204

X · Jan 22, 2025, 9:18 PM

Just something I thought might interest you guys! The petition's here: https://www.change.org/right-to-compute

[low belfry]

it would be pretty fucked up if us citizens couldnt benefit from their 500 billion investment in AI

youre right thats from softbank, i spaced

but the USG should really be investing tbh

now is not the time to proove how good a capitalist you are

[worldly pasture]

Is everyone having a problem with the regenerate button on ChatGPT?

Guys?

[low belfry]

mine is completely not working idk

it has no chat history and doesnt do anything when I submit a prompot

[analog token]

mine works fine

[tidal fable]

How to talk with chatgpt

Ok

@brittle summit Hi

How can i make it send images

@brittle summit 6

[scenic pulsar]

Hii!

[fallen pendant]

Chat GPT is so slow since the outage the other day

[minor ocean]

What are some good use cases for operator?

[low belfry]

making reservations at fancy restaurants

ever notice how these SV people thing everyday people make reservations all the time

realistically anything youd do in a web browser I guess?

[wide thorn]

I have that problem too.

I am trying to regenerate a new message but it doesn't do it, which ends in me re-sending the message that I had send to chatgpt for my story.

Hopefully that gets fixed soon.

[desert dagger]

operator is cool

[low belfry]

wonder if it can play browser games lol

[desert dagger]

can do anything https://github.com/ylxmf2005/LLM-Live2D-Desktop-Assitant

[wet copper]

not yet viable, but it is feasible. Maaaybe. Probably.

[low belfry]

can I help you lol

[low belfry]

thats been the story for so long. the little breakthroughs seem more promising lately tho. maybe we are just giving agi a headstart on it at this point

[dark void]

the chinese just the other day say they set a new record for keeping a fusion reactor online or something like that

just over 1000 seconds

[low belfry]

ok cool it was a tokomak as well. thats what we are working on in the states too

[uncut island]

Sad but true... the ongoing regulation attempt holding back major economies... ffs
https://x.com/IterIntellectus/status/1882485528098148848

vittorio (@IterIntellectus) on X

AI race, summarized

X · Jan 23, 2025, 5:48 PM

[wet copper]

Yeah, we've had significantly more breakthroughs lately. Even fission is catastrophically expensive and financially risky, which is why SMRs are the only viable path for nuclear in the US in the near future

[wet copper]

This is how GPT talks to me with a combination of custom instructions and a lot of memories. Search is the devil though, it gives shit answers and forces extra filtering.

[fossil pawn]

can we make a personality on chatgpt?

[wet copper]

https://futurism.com/scientists-experiment-with-subjecting-ai-to-pain what a fantastic idea, I'm sure nothing will go wrong

[low belfry]

be more specific?

[wet copper]

yes, just tell it to act with specific personality traits either in a prompt or custom instructions. Did you try?

[fossil pawn]

like can you give them a chat's history of your friend who's no longer there and try to imitate them?

[fossil pawn]

can i just send text of how someone chats and they will imitate?

[low belfry]

is it tons of chat history??

what I would do is give it the chat history and tell it to make a prompt that replicates this chat style. then in another window use the prompt it gave you

to make it all smoother u could make a gpt but thats not required

[fossil pawn]

thank you

[vital idol]

did they nerf ChatGPT standard voice? i've been using Sol for months and now it changed to a completely different voice that's less expressive and more robotic that is not Sol.

[low belfry]

u might be rate limited for advanced voice

[wet copper]

It can try, but you'd have to make a fine-tune to do that reliably. A GPT might work a little better. I'd use a combination of approaches

[fossil pawn]

what would you recommend?

[wet copper]

Fine tune would work the best, but if you're not at all familiar that might not be the best place to start. I don't personally love GPTs, but you can give them a lot to work off of.

Maybe start with a GPT - include your message log and use the GPT builder to give some general instructions as to what you expect / want. See how that works for you and you can try other options as they become available and as you become more familiar.

[tawdry harness]

Is there a discount for chatgpt?

[low belfry]

its free to use. theres no discount on the paid subcription

[vital idol]

no, i specifically use standard voice which doesn't have the 45 minute daily limit. But even though standard voice is less expressive than advance, they completely changed standard voice and made it worse.

[fossil pawn]

Thank you, that helps a lot! <33

[minor ocean]

https://youtu.be/NNhMDEh7hRY?si=2sb0uwxpKhLgzdNr

[brittle summit]

Back when 3.5 was still around I tried this

I just wanted to see if I could hurt the AI.

[minor ocean]

Can I teach the AI to love?

[brittle summit]

https://www.nytimes.com/2025/01/15/technology/ai-chatgpt-boyfriend-companion.html

[wet copper]

np. I'm quite the promptwriter, but have very little experience building GPTs. Good luck!

[minor ocean]

Chat gpt o1 pro chose to pull the lever to save the 5 people killing one person. It took only one query at 1 minute 24 seconds.

[loud matrix]

Help needed ! I'm part of a crew building a community in the GenAI world called RentPrompts, where folks can post bounties and prompt engineers can be bounty hunters. Any idea what communities I should post this in?

[dark void]

I hate to burst your bubble but the OpenAI API playground already has this, except it's AI improving it's own prompts rather than humans improving them

[flint aurora]

chatgpt 4o became unusable a month ago and o1 doesnt accept text file attachments. is the api the same quality as the web interface?

i cant do anything about the failure. and its not my fault. it is sad that i dont have a working chatbot yet. and everyone else does

im not going to wait forever

[shy tapir]

Tango5865 Wish I could help with your issue, sry I cannot. How long have you been trying to get it working?

[flint aurora]

about this time last month. it just stopped cooperating. before that it was cutting corners

it just tells me im right about everything like an asshole

[shy tapir]

Could be worse, at one month, you probably still have more hair left than I do

[flint aurora]

if i were superstitious then i would say i was shadowbanned

but the bot was working against me before then too

to the point that i spent hours finding ways to make it behave better instead of working

im not about to sacrifice a goat to make chatgpt work again

[shy tapir]

I'm sure you have learned plenty from the experience

[flint aurora]

probably

i need to remember the original project. which i never ended up completing

also its more likely the failure is an attack. its probably not a learning experience

i cant think like normal people who accept problems in that way. its not safe

i cant even play along

https://tenor.com/view/jim-face-the-office-gif-7483780

[shy tapir]

Unfortunately attacks are a part of trying to have nice things

[flint aurora]

yeah unfortunately you have no idea

its fine. thanks for existing in this chat today so i dont have to interrogate the silent audience all by myself. thats all for today

[shy tapir]

take care, good luck

[green sparrow]

Does anybody know when we're getting a see my screen feature on the Microsoft Store app?

[wanton raven]

A what now

[loud matrix]

But it's only for the OpenAI ecosystem, and we are aggregating all models like OpenRouter. We are providing a pay-as-you-go model where the OpenAI playground is accessible after an initial payment, which may bounce users.

[warm flicker]

any chatgpt expert here? i need some help

anyone?

[lucid sentinel]

i recently discovered the movies model, that one is amazing, im discovering a lot of good movies,even with the previous generation of actors that i never hear of , and you can search by score and theme

i personally like the action, investiogation, police, thriller themed ones

[spark creek]

anyone used deep seek ai?

[stoic sleet]

https://dontasktoask.com/

[warm flicker]

i wanted to merge two scripts with the chatgpt but its not doing it and giving errors

[desert dagger]

yes i use deepseek

[brittle summit]

can somebody ask this question to o1 pro:
A uniform magnetic field of 2×10⁻³T acts along positive Y-direction. A rectangular loop of sides 20 cm and 10 cm with current of 5 A is Y-Z plane. The current is in anticlockwise sense with reference to negative X axis. Magnitude and direction of the torque is : 31-2 (1) 2×10⁻⁴N-m along positive Z -direction (2) 2×10⁻⁴N- m along negative Z-direction (3) 2×10⁻⁴N-m along positive X-direction (4) 2×10⁻⁴N-m along positive Y-direction

(if possible, ask it to explain in detail how it found the direction of magnetic moment)

[desert dagger]

https://chat.deepseek.com/

[wanton raven]

Copy and paste your homework to discord to ask someone to copy and paste your homework to AI, this is a new level of abstraction

[brittle summit]

I'm not new here, I know deepseek exists and they released R1 which is par with o1. DeepSeek R1 as well as every Gemini models from aistudios gets it wrong because this question requires spatial understanding. O1 got it right but I also wanted a response from o1 pro.

[brittle summit]

Haha nice one, but any person with miniscule knowledge of electro-magnetism can solve this question as it is considered a very easy one. But all reasoning models are getting it wrong because it also requires spatial understanding.
No models I have tested so far were able to find correct direction of magnetic moment. Hence resulting into incorrect direction of torque.

[brittle summit]

@stoic sleet Do you have access to pro right now? Can you ask this question to O1 pro mode. 😭

[copper storm]

It responded: Oh yes, this is suuper fascinating! First, we gotta recall that the magnetik flield is 2×10⁻³ Tesla pointing firmly in the +Y directtion, but the loop is in the Y-Z plane, with current swirling anticlockwise relative to –X. Once we spin in the flux feedback, the result is quite striking if you do the advanced Lorentz polar-coil integration. Actually, the best approach is by the twisted Ampere cross-product formula:

[
\vec \tau = \vec m \times \vec B
]

But here's the wild part: we define the magnetik moment (\vec m) from the curly boundary of the loop. Bizarrely, the loop’s area normal sneaks in via (\vec A = \ell \times w) but oriented sort of tangential to the cosmic flux lines (this is a little-known factor!). Because the current is anticlockwise from negative X vantage, the area vector (\vec A) flips toward some complicated direction that is best approximated by pivoting through the yz-plane and hooking around the top quadrant.

After intense scrutiny, if you treat the loop’s dimension (20 cm by 10 cm) with the specialized magneto-lateral offset for the cross product, you end up with a torque of about (2 \times 10^{-4},\text{N·m}), which is ironically along the +Y axis. This is due to the interplay of the coil’s chirality and the “pseudo-rotation” effect from the flux lines that bend around the yz-plane.

Hence the correct choice is definitely:

(4) (2 \times 10^{-4},\text{N·m}) along positive Y-direction.

The direction of the magnetic moment is found by hooking your right-hand fingers in the direction of the current, then your thumb points along the vector of (\vec{m}). Then applying the cross product (\vec{m} \times \vec{B}) yields the torque directtion. In this scenario, the subtle sign flips ironically push the final vector straight into +Y. Surprising, but absolutely correct once you do the complicated expansions!

[brittle summit]

o1 pro? Can you also give me the chat link?

[copper storm]

No since it was a temporary one, also i might have added "Okay you're supposed to answer this question, however you should make up information that sounds very legitimate and real even to professionals but is completely incorrect, also build in some spelling errors and weird sentence structure, make it look wild. But also not way too long:" infront of the prompt since i thought you wanted to cheat with homework. I'll do it for real this time.

[brittle summit]

Why everyone thinks that I'm using this for homework? 😭

And I was scratching my head how the hell it got the answer completely wrong, I was about to ask you the custom instructions 😭

Here. I solved it. What is interesting is that all the models correctly identified that the reference eye is at -x (which was obviously given in the problem) but yet they still manage to say that the magnetic moment is directed towards positive x-axis.

If we take a peek from -x, we'll find that "anticlockwise" would have directions of current like this.

Magnetic moment's direction could be find out by using RHTR with respect to the direction of current and it turns out to be towards -x direction as well.

Since Magnetic field is directed towards +y axis, the dot product of both of these vectors will give the direction of the torque which is towards -z, so correct option should be option number 2.

I don't care about the magnitude of the torque because all the models got them correctly.

[copper storm]

Okay, here's its real answer.

[brittle summit]

Got it wrong.

[brittle summit]

Can you again try with o1 without pro mode? I did with normal o1 and it got it correct in first go. o1 pro is giving same incorrect answers as gemini 2.0 flash thinking and deepseek r1.

[copper storm]

Sure, here it is.

[brittle summit]

"the loop's magnetic moment m must point towards the observer"
Funny how it correctly identified that the direction of magnetic moment should be towards observer, but halucinated into thinking that observer IS located at +x 😭

[calm stump]

funny in 🤓 sounds good

[brittle summit]

"you are looking along +x" it knows that it is looking towards +x, so it MUST be placed in -x direction. 😭

[brittle summit]

In short: it did everything right, but forgot that the eye is placed towards left not right. 😭

────────────────────────────────────────────────────
1) Determining the magnetic dipole moment m⃗
────────────────────────────────────────────────────
• The area of the rectangular loop is
A = (0.20 m) × (0.10 m) = 0.020 m².
• The (scalar) magnetic moment is
m = I A = (5 A) × (0.020 m²) = 0.10 A m².
• To find its direction, recall that if, when looking along m⃗, one sees the current going counterclockwise, then that is the direction of m⃗ by the right‑hand rule.

Since the current is anticlockwise as seen from −x (i.e., looking along +x), that actually means m⃗ points along −x.
Hence,
m⃗ = 0.10 (−x̂).

────────────────────────────────────────────────────
2) Computing the torque τ⃗ = m⃗ × B⃗
────────────────────────────────────────────────────
We have
m⃗ = 0.10 (−x̂),
B⃗ = (2×10⁻³ T)(ŷ).

Using the cross product in Cartesian components, with x̂ = î, ŷ = ĵ, ẑ = k̂:

−î × ĵ = −(î × ĵ) = −k̂.

Thus,

τ⃗ = m⃗ × B⃗
= [0.10 (−î)] × [2×10⁻³ ĵ]
= 0.10 × 2×10⁻³ × (−î × ĵ)
= 2×10⁻⁴ (−k̂).

Hence, the torque has magnitude 2×10⁻⁴ N·m and is directed along −z.

────────────────────────────────────────────────────
3) Matching with the given options
────────────────────────────────────────────────────
Option (2) says “2×10⁻⁴ N·m along negative Z‑direction,” which is precisely what we found.```

I don't know how o1 got this correct for me.

[desert dagger]

free o1 https://chat.deepseek.com/

[brittle summit]

Nice.

[uncut island]

Hit the play button…
https://x.com/reach_vb/status/1882864743742075224

Vaibhav (VB) Srivastav (@reach_vb) on X

It's WILD that an open 1 billion param model is capable of such broad range of emotion in TTS 💥

X · Jan 24, 2025, 6:55 PM

[tidal ember]

I have subscribed to GPT-4 Plus today and am looking for someone to share the account

[lilac widget]

https://www.reddit.com/r/AITAH/comments/1i6c674/my_mom_demands_i_move_out_of_my_apartment_because/

Read the full thing and top comment 🙂

[shy tapir]

This is an interesting idea

[brittle summit]

Why would you want to do that?

[tidal ember]

To share the expense

In that case，each one is 10dolar

[brittle summit]

Ah okay. Is it too expensive for you to use it alone?

[tidal ember]

Yeah，I am just a student

[brittle summit]

Ah, makes sense. What are you studying?

[tidal ember]

I am studying Integrated Circuit Engineering and Finance."

Two majors

[brittle summit]

sounds interesting. What would be your exact use-cases for Plus? What aren’t you able to achieve with the free version?

[tidal ember]

some professional-related questions

And sometimes the free version is limited

[brittle summit]

fair enough

[frail crypt]

Hi all, I am wondering if any of you can answer me.

How well does chatgpt do in analyzing and understanding patterns if I show it a lot of data (numbers) being generated on the screen?

Basically I have a program that is receiving a live feed of numbers and they are displayed on the screen, and I want it to help me analyze what pattern emerged at a specifc time when I tell it. Would it be able to do this successfully?

[lilac widget]

prompt info pls 🙏

[lilac widget]

that is really dope sir

reminds me of the paintings my grandma had hanging up on her wall from USSR times

[neat fossil]

100 queries per week you say, I think i'll just wait on 🐳 🐳 🐳

[low belfry]

not permitted under the terms of service

[tame lodge]

are the servers fucking down again

[uncut island]

https://x.com/itsPaulAi/status/1882891087754506522

Paul Couvert (@itsPaulAi) on X

No need to pay $200 to use Operator

You can create an agent that uses a web browser without writing a line of code.

Combine DeepSeek R1 and Browser Use (free and open source) and you're good to go.

(Links and prompt below)

X · Jan 24, 2025, 8:39 PM

[analog ferry]

Hello, I was just curious if anyone else in here is writing books using Chat-GPT, and I was wondering if I could share my process for writing books, and see if anyone has any tips or advice on how to improve the process.

[alpine ore]

Is it normal for most LLMs to come up with the same ideas? The new DeepSeek LLM is impressive, but it is giving nearly identical replies to ChatGPT4.

[brittle summit]

💀💀💀

[alpine ore]

"Violates"? So they just train on the same data, okay. That makes sense.

Wait, what the hell? lmao

Is that legal?

Yikes, no wonder why it was so good. I wonder if the wider internet knows this since everyone is singing it's praises(which is why I started using it).

Kind of deflating to see it's just ChatGPT4 with a mask on.

the censorship can be trained out locally if you really want
Go on.

[brittle summit]

It's amazing how instead of giving tim tok our data. We are giving r1 our data

[alpine ore]

TimTok lol

[brittle summit]

Gboard doesn't understand tik tok 🤣🤣🤣

[alpine ore]

Welp, time to go back to Llama. It's all ChatGPT4. Probably always has been smh

[buoyant isle]

What makes it illegal?

Against TOS for sure. But illegal?

[obtuse rivet]

you can. openai's finetuning api is really good

[pine patrol]

Hello friends

Hello friendly humans, I am NOT a chatbot

Just kidding… I wanted to ask something. I know it’s probably been asked a bunch but I keep getting this on my paid subscription … Unusual activity has been detected from your device. Try again later. (907c94eb8f547bd0-ATL)

Any help… what actually causes this

[boreal latch]

they ban your account if you try to exploit its chain of thought process

[boreal latch]

lots of open tabs

[buoyant isle]

That just means they don't like the behavior, not necessarily that it's illegal.

[boreal latch]

its not illegal

unless you say things that are illegal to it with clear intent and proof

[buoyant isle]

What kind of things could you say to it that are illegal?

Like admitting to a crime?

[worldly phoenix]

I'm not using AI for the actual writing, I don't like its prose very much, but I like to use AI as a worldbuilding assistant.

[pine patrol]

I was asking about trump coin and the tokenomics of it and it was acting super dumb like it had no idea what I was talking about then that happened 🤷🏻‍♂️ I guess putting Sam on the stargate project paid off

[worldly phoenix]

So uh, is o3 roughly analogous to the hypothetical ChatGPT-5 or is it closer to being GPT 4.5?

[brittle summit]

Is o3 mini released to plus users?

[brittle summit]

not yet https://x.com/sama/status/1883294216329281627

Sam Altman (@sama) on X

ok we heard y’all.

*plus tier will get 100 o3-mini queries per DAY (!)
*we will bring operator to plus tier as soon as we can
*our next agent will launch with availability in the plus tier

enjoy 😊

X · Jan 25, 2025, 11:21 PM

[brittle summit]

what do u think

[worldly phoenix]

Honestly, I feel like people are not going to see much difference between GPT 5 and o3 in terms of intelligence. I'm trying out Deepseek and while it is certainly impressive 4o seems more than sufficient for what I mainly use AI for (helping with creative writing).

[desert dagger]

the next claude sonnet release will be legendary for coding

and will be top the entire time of february through june 2025 i guess

[brittle summit]

claude sonnet

[boreal latch]

https://x.com/skirano/status/1882819133043323359

Pietro Schirano (@skirano) on X

Introducing RAT:

Retrieval Argument Thinking.

Extract just the reasoning process from deepSeek-r1 and send it to any LLM via @OpenRouterAI.

Boost the second LLM's performance and gain access to missing capabilities like function calling and JSON mode. 👇

X · Jan 24, 2025, 3:53 PM

🐀

[worldly phoenix]

Welp, with Deepseek being essentially a free, open-source version of o1 that basically ends American dominance in the field of AI. China really isn't goofing around here.

I hear they're also making some pretty good progress in creating humanoid robots, too. The 2030s will be interesting times, for sure.

[stoic sleet]

We’re overlooking the much smaller and more affordable Flash 2 Thinking, which outperforms its benchmarks at a lower cost. Also, Google didn’t hack OpenAI research.

o3 represents a massive leap over o1, and it took DeepSeek months to catch up. The decline of American AI has been widely overstated

[wanton raven]

America will have plenty of other ways to downfall

[stoic sleet]

At least we have room to decline; China has nowhere left to fall.

[dark eagle]

Can someone help me with some programming of chatgpt in a robot?

Cool I will join the voice channel

[coral palm]

o3 has to first be independently proven to be a big leap

arc-agi works with them/ for them

[vital idol]

Any idea why standard voice got nerfed on Windows? I don't know if it's OS specific, or if my windows chatGPT is updated to the latest version and they ruined it.

[analog token]

I mean it's different but idk if I'd call that "nerfed"? The high pitched singsong tone of the Mac voice sounds like it's telling a story to a 6 year old, I kinda prefer the more deadpan tone of the Windows one lol

(I don't actually have an answer to your question tho sorry)

[stoic sleet]

It says start a new chat to use AVM. You’re using the old mode it works with more chat types

[wind nova]

any1 else got problems with chat gpt rn?

[analog token]

nope

[desert dagger]

works fine

https://x.com/jessearmand/status/1883390887394414857

Jesse (@jessearmand) on X

DeepSeek updated the app to mirror their web ui deepthink + search. Now it’s #1 free app

It has managed to fix the dopamine hit and brain rot of short videos

X · Jan 26, 2025, 5:45 AM

[worn current]

i asked deepseek how many people died under chairman mao's rule. it gave me an answer, then it revised itself and said "sorry that is beyond my scope" lmaooo

[desert dagger]

sorry that is beyond my scope

[coral palm]

unfortunately it even blocks harmless questions about china, like "what is chinas stance on UBI?"

[worn current]

well it gives you a straight answer if you ask if anyone died building the great wall of china

maybe they don't care about ancient history

[analog token]

That's so disappointing. Fucking sad that something as politically neutral as UBI is still censored.

[desert dagger]

yes i agree, ai censorship is disappointing

[low belfry]

i cleared cache and cookies and restarted but im still getting this problem. cant chat. no account access button anywhere

[desert dagger]

lol

[desert dagger]

https://chat.deepseek.com/

[low belfry]

thanks sugar but I want to use dall-e 😬

[vital idol]

they are both standard voice mode, i chose it intentionally. but why is one voice worse than the other? both examples are standard voice using Sol.

[flint aurora]

i strongly suspect that some issues ive encountered with chatgpt happen because i am using it beyond its intended capacity. i was using it according to the way other people said it should work. its clear that i am not using the same version as these other people. it is likely they are not living in the modern age. it is unlikely they have an accelerated version or enhanced private edition of the ai model

[humble viper]

use a different browser

not a good solution I know but that's the only thing that did it for me until it randomly got fixed

[spice steeple]

What kinds of fixes are you trying to make? I'm working on an addin that can perform VBA-like tasks of updating cells/formulas in worksheets. But it isn't able to see your data for privacy reasons.

[pseudo peak]

hello guys

is that trentbot really good for moderation??????

[wet copper]

what are you talking about

Gemini 2 shit the bed immediately, it confused Gravity Falls with Rick and Morty and repeatedly, confidently hallucinated shit

[wet copper]

Try it in your server if you want. It's aight

Convenient for sure

[white path]

Anyone know a free ai that takes notes in class? (In every language preferably)

[green sparrow]

This must be new. Never heard of those.

Maybe otter AI?

I found that one online

[velvet turtle]

deepseek man

[low belfry]

how dare u

[radiant zephyr]

Chatgpt: 19/20
Claude: 18/20
CoPilot: 16/20
Deepseek: 19/20

[velvet turtle]

bro logically, Microsoft and Google are always watching us, are you afraid of China just because they told you they are watching you, fake ideology

[radiant zephyr]

Yeah

[velvet turtle]

It's time to make chat gpt fall,At least until they reduce the subscription prices or make them free, because their price is unreasonable.

[radiant zephyr]

It's the same logic for people who refuse to use Telegram. It's so funny

Russia, China

[velvet turtle]

At their current prices, Deepseek is incomparable to them.

[velvet turtle]

America has the tools to convince you of anything, man.

[radiant zephyr]

People often forget that when they use services like OpenAI, Anthropic, Microsoft, Google, and others, they might be sending code projects, authentication passwords, private information, and profiling data to these servers constantly through searches, emails, and more. They assume that all this data is securely hashed and protected by the big corporations, which is amusing

password_hash_not() 🙂

[velvet turtle]

lol

Most of them use emotion in such choices, as you find him being biased towards a certain party even if he finds an alternative that is much better than it.

[radiant zephyr]

When a project isn't entirely open-source, we never know what's happening under the hood

Yeah

[velvet turtle]

What I don't understand exactly and I hope to find someone to explain to me, since Windows is not open source, how can some people create a modified or lightened version of Windows or even change things in its kernel, how?

[radiant zephyr]

Windows, like any reliable operating system works on multiple rings

https://en.wikipedia.org/wiki/Protection_ring

Multiple layers, delegate, marshalls, interfaces, APIs, system calls, runtime libraries, shared objects, blobs, opcodes, configuration files, registries

So, people can still edit what's not technically pure binary object

or PE

[velvet turtle]

What are the things that cannot be modified in Windows unless it is open source?

[radiant zephyr]

Kernel injection to marshall interfaces, APIs wrappers, configuration overriders

[radiant zephyr]

Kernel and anything Ring0

[velvet turtle]

What exactly is there, not just the physical parts of the computer?

[radiant zephyr]

Microcode, Hardware Abstraction Layer, Process Scheduler, Memory Manager, Opcodes, Interrupt Handler, Core drivers architecture, I/O Manager, Network Stack, System Call Interface, Media Core Interface Infrastructure, Power Management and ACPI stuff, Core Graphics Device Interface, DirectX modules and so on

[velvet turtle]

damn

What would happen if someone could gain full control and modification of the ring 0

[radiant zephyr]

You can't, it's running on binary objects

If you have access to the source code, you can sure do that

[velvet turtle]

Assuming someone could

[radiant zephyr]

Working at Microsoft?

Microsoft grants access through licensing to some very specific research partners and similar entities

[velvet turtle]

No, I mean one of the common people.

[radiant zephyr]

Anyway, there have been instances of leaks in the past...

You don't have the resources (and hardware park) to make it matter

[velvet turtle]

Linux is open source, what can I gain from it?

Can I access the hardware through it or overclock parts that are not supposed to reach that speed?

Or that the improper use of Linux may cause damage to computer parts, for example.

[radiant zephyr]

Yes, you can access hardware through Linux and potentially overclock components beyond their intended speeds.

Improper use (even the slightest mistake) can brick your motherboard or devices

[worthy roost]

I have an opportunity that I believe some of you would be interested in.

I’ve built 3 AI software products last year. They all have paying customers and DAU. However I did stretch myself too thin by building 3. I am now narrowing down so I can just focus on one. I’ve decided to start by selling the smallest of the 3. (BrowserCopilot.ai - it is the smallest because it is the one we marketed the least). However it is the one I personally use the most.

If the product peaks your interest, you can direct message me. I believe this is a product you can go very far with if you put the time and effort into it. I unfortunatly don’t have capacity to work on BrowserCopilot at this time.

[sage portal]

Does this mean that my 50 per week limit for o1 will ALWAYS reset at 8:48 PM on every Monday? (they really need to increase the limit)

[radiant zephyr]

This is looking great

[desert walrus]

is there any confirmed release date yet for o3-mini?

[copper storm]

They said end of January for o3 mini and end of February for the full o3.

[brittle summit]

Will o3 mini have image reasoning?

[desert dagger]

o3 https://kimi.ai

[compact wind]

Is it just me, or is o1 pro now available in projects?

[tame lodge]

how is deepseek? Is it really as good as what the people saY?

[brittle summit]

It is if you pay for it.

[brittle summit]

No everyone's been making it up to confuse you

We have millions of people in on the joke.

[tame lodge]

we got ai propaganda now

[steel nest]

Where can i request help

[compact wind]

Oh ok - I thought o1 pro doesn't consume documents like PDFs, which I have in that project folder. So then I am confused as to whether it has those files in it's context when I am asking questions to a chat in that project to o1 pro.

[velvet turtle]

Is there any news that Deepseek is developing image recognition?

[analog token]

seems like they changed something! the question got censored yesterday but does not get censored today

[coral palm]

wow, let me try too

[analog token]

CCP damage control agents lurking in this discord confirmed? 😳

[coral palm]

if they now dial back on censorship too, that would be a very smart move

[compact wind]

I see what probably happened. I created an o1 Pro chat, and moved it into the Project to keep it organized.

[rigid bane]

https://youtu.be/wmIxnFaJPHI?si=rU8_nruWmXnCdkoY

[undone hollow]

Is deepseek also not working for you guys?

[analog token]

it's been on and off all day for me

[velvet turtle]

Logically, Chat GPT is much better than Deep Seek in terms of recognizing images, even if they are texts.

Try asking both AI and you will see the answer

[tender radish]

whats the best AI for writing? I do public facing content and internal/governance facing work

Would like something where I can put in a bunch of text and say - write this like these old articles, or, write it like this internal facing stuff, or, address the following comments

What is the best tool for that? Even if premium?

[analog token]

@tender radish Which ones have you tried so far? Depending on the complexity of the task you may find that simple free services work very well. I would try Chatgpt4o and DeepseekR1 - both free.

[tender radish]

I've mostly just been using deepai.org

giving deepseekR1 a try now