#💬│general
1 messages · Page 192 of 1
Potato
😭
Duck
Goose
is it accurate?
AND WHY AM I NAKED
No iam 185
I know right.
Wtf
No nsfw in chat
imagine being the same height as a potato
Where is my clothes
Jokes on you pato came from potato
Good family photo finally
That diamond icon means 100 credit one.
Ohh
@cobalt vale
How does this bring your own key work can I use bring any key even if the model isn't from open ai?
Api yes
No i think the models need to be defined there
Like opus and o3 pro and so on
Atleast in t3 chat
It would be nice to have some boths in chat here
To check how many messages one has sent in this server and so on lol
it shows the total messages when you search for someones messages (doesn't work on mobile)
Cheers
It doesn't?
But have to check by channel lmao
Mobile shows 100+ always
I think it searches all channels at once though?
nvm 1000+ it was
1000+
in: #channel
yeah if you want to specify one channel
Add that and it filters a particular channel.
rip plus users
Lmao stupid premium model
They really need to redo how premium credits work
Interesting tgat they made it available for team users
Lemme try it out
I was right
I'm team and don't have it :(
Is there pro and plus in teams users
Or just the standard one
not gonna hope 🥲
???
Thats not too bad
No
massive improvement
Maybe should check web
From o1 pro being 600 usd
On android and don't see it for teams
@short narwhal @minor shard
Where did @gloomy cipher go D:
hi iam doing something illegal
That's what I'm suprised it removed
why did i become a cat
That's jawn
Let me fix it
Jawnze is in Starfleet?
aww @timber niche looks like a baby
Gemini deserves all the hate.
Better than not existing like you lol
And you're a king in there
oi u play league aswell?
Almost 20k messages
how about the duck?
nahh, I downloaded it for a bit after arcane though
Still have it on my pc
Cuz of how awful Gemini has been, I'm thinking of making a dedicated thread just to make my criticism prompts more hurtful and degrading.
Mere 1250
iam the youngest
I suggest how to make a function in Claude when the code writes there was a separate window.
I feel very tired today.
I love crows.
How comes we cannot attach images to spaces? Like pngs etc
@gloomy cipher
The translation of the prompt into English is now available, download it from here:
#1381764428982390917 message
thanks ❤️
I have encountered a wild crow at my location that perches on a traffic sign and requests peanuts from me.
I find this very positive.
At 10:28 PM, there is a path that splits into two directions; on one path, I placed a mutant XXL peanut, and on the other path, a normal small peanut. The crow selected the XXL peanut.
What's special about this mutant peanut?
@gloomy cipher
its gigantic.
Awe ❤️
2x the length of normal unroasted peanut
Gang
Nice.
What's the price difference?
what food next
its good to see my clothes 🫶 but iam still short 😠
I read it as bang first.

also need kirai and systemzx as our grandfather since they do all the hard work
In the pic?
hmm
what food next
what food next
what food next
what food next
what food next
Today Gemini will get bullied.
What can you afford?
Lol. Okay.
Hopefully you can afford mental health support too.
omg why u insluted
@timber niche @distant aurora look at my hand 😭 😭
what did i say
In Harvey's underarm.
Look at u hugging that cat like it's ur pet
My avatar is so messed up as well lol
Wait. There are two ducks.
Feed it salmon. Most nutritionally comprehensive food afaik.
Is that from Playground?
ok i am not rich
Blud went from affording my family to peanuts.
/s /s /s /s /s
Awwwww yeeeeeeee
:annoyed:
Where
Under
Api? Idk
Yes. Api.
Hmm
Open ai playground I think.
Idk why my teams acc doesn't have o3pro yet
yes
Yeet
Sup
Been a while d and t
300g protein
Everything good on your end?
they said starting today, probably staggered rollout
Followed by 300 minutes diarrhoea.
If it was nutritionally viable, I would eat chicken shwarma all day.
Pplx or
yea
Devs ain't very fast.
yeah lots of trouble at work lol
Yeah you can from web only
What is the context size of o3 Pro?
Should be same as o1 pro afaik
Hopefully you didn't get in similar trouble like before.
No lololol. Upper management is trying to cut down my staff head count
That's terrible
All while I'm fixing a huge mistake of theirs
Upper suckers
Lay offs?
peak SOTA
Yeah
This is o3 Pro itself answering
It's probably same as O3.
Lmfao
200,000 context window
@cobalt vale Is the o3 model decent for advanced narratives that require precise calculations?
Gemini told it?
Ah
I’m not trusting a word that comes out of that things mouth
Lmao
So essentially
O3pro is o3
But bumped tf up
Yeah same as o1 pro basically
More compute thrown at it or something like that
I don't know... 😭
atleast normal o3 is 80% cheaper :)))
May 31, 2024 knowledge cutoff
Outdated
That's all it's about
That’s my #2 goat OpenAI
I'm broke
Betraying claude
I can't afford chatgpt.
😭 noo i could never
why is it so expensive
Claude is the side guy.
would be funny if they made o3 shitty and o3 pro is just old o3 😭
Literally did that to your bestie
Lmao, fr.
Just be happy that it isn’t 600 per 1m output anymore
Exactly
dont doubt the ability of open ai
dude, gemini 0605 is 10$ per 1m and its wrecking a lot of ais
Fixed it
Hmm
deepseek is also lowcost
Betrayal stays can't edit feelings
I hope the increase usage quota for Plus users
apple user detected
We been telling
real 💔
LOL they actually labeled it w the date 😭
@valid charm is open ai playground pay as you go?
Sorry…
dont be sorry
nooo I didn’t mean itt
yea 03 not 03 high
At least someone’s got my back
yeah with free tokens daily if you allow your usage to be used for training
Will private account do?
And outdated
I thought it needs company account.
Ew
nothing comes close to OG o3-preview
Betrayal should not be encouraged
am i the only one who didnt read that as an apology? i tho u were asking me a question 😂
I have a personal accaunt with verification. All I did is take a selfie and flash my ID
Where's my boy r2?
It was both
Company id?
What is Pareto?
Nahhh my regular ID
Is that necessary? What for?
It says Company verification but it just asks for normal personal ID
It enables detailed summaties of the CoT
Oh.
Suppusedly can be used to crate jailbreaks
👀
wtf why did they cap it to 1k reasoning tokens
did they increasethe output?
what was this for
100k ouput is crazy i wonder what they limited down to in perplexity
i was about to say nothing cuz the models not in pplx
but it is 😭
when did they remove o4m
whats sonnet limited to?
no reason to use o4-mini when o3 is there
which one
there is a lot of version of sonnet
@valid charm am I supposed to select purchasing as business?
i couldnt find the thinking model but maybe its 2in1?
i dont think it matters
u just go to data controls and enable sharing
i think
Which one? Also, what does that 7 free weekly evals mean?
You need to pay first for it to show up
are they even offering the thing anymore
i jus read its til april 2025
🗿
They extended it
again?
also I think it's 250k tokens on o3 unless you pay $100 for API tier 3
yea
u have to put in atleast $5
guys, i just tried the o3 reasoning model in pplx. it takes about 30-45 seconds to generate an answer and then the entire answer just appears instead of being slowly typed like other models (token by token)
is this a bug?
headsup virtual/prepaid/paypla/google pay are not supported
I don't have to select purchasing as business though, right?
Nah you don’t
no additionally you can invitie other into your organization 🥺
did pplx just get o3, or has it had it for a while?
I’m 19 and grew my dropshipping store to $3K–$4K a day in just two months. I started with zero experience — just drive, focus, and a willingness to learn. If you’ve been thinking about starting your own journey, I’m offering free advice and mentorship to help you get there. No gimmicks, just real help. Shoot me a message and let’s get started!
just got it after today's price drop
also t3
thats normal
i believe
ah ok
no "new" thing is weird then
i think that should not happen. it doesn't happen in any of the other reasoning models.
also the other reasoning search the web while this one just stays blank for 30 - 45 seconds
lmao
it just doesnt stream the response...?
not that big of an issue
ah
i see what u mean
maybe this is a bug
yeah...
didnt happen last time
bro is meditating
but I do have web turned off for this prompt soooo
turn on web and it still won't search most of the times (i never got it to search)
tried other models with the same query, they do search
whats that?
the first thing i always ask a model
just to make a calculator with as many features as possible
okay correction. it does search when you explicity tell it to. but it is very less likely to search compared to other reasoning models
mine searched just fine
🤷♂️
i worded that prompt so weird
😭
hmmm, it works
guess mine was a one off
but i think they have done something at the backend to make it call search less often or something
@short narwhal @cobalt vale nice
does it stack
orr
Wdym?
sadge
nice, do you happen to use the API a lot?
I'm wondering how to track token usage.
You can ask perplexity itself 😂
should say it on site
🤷♂️
I don't see any option to use o4 mini.
For O3 it's asking verification of organization.
Are you busy today?
Extra thinking.
yay
the model is o3 mini and the reasoning is set to high
It just allows the model more tokens to reason
aka better response
i see thanks
How do I give the model internet access on playground?
Also, can it access tools too?
@cobalt vale
idk...
Have you never tried it?
does the pplx api (that comes with a $5 credit every month for pro users) also have the TTS pplx uses?
@valid charm @short narwhal
I got the month of ChatGPT Team instead
Nah
No clue, never used the playground
it does NOT know what buttons should be on a calc 😭
Yeah
itll cost you money
-0 🙏
Bruh.
is there any way i can access pplx TTS altogether for my mini project? i was looking for a good free TTS but all of them are paid options (the good ones).
The best i found was google ai studio which is also limited via the api i think
🥲
host it locally 🤷♂️
wdym? host what locally?
Could have used it on open router instead.
The TTS
What are you trying to use
o3 pro?
tts
I think pplx uses/used eleven labs for TTS
also yeah openrouter is amazing
I needed strong reasoning model with web search for research.
I want sesame to release their speech already !!!!
why not just use the 5 free DR
yeah yeah i get that but host which TTS locally? i mean ik these are some stupid questions i am asking but i really don't know much about TTS
or something like exa api
the voice quality is so 💩
Cuz no one told me.
I think theres also ways to integrate google API using open web ui
Since I bought I code online how long will the code be active before I lose my pro. I’m using it for a project so I want to prepare
I'll suffer in silence now.
plex it next time
its okish, compared to the open source ones i found, this was far better
it doesnt feel as good as eleven unless it was an old version
Probably old
also this is allegedly part of the pplx voice system prompt
use blackbox they have good models and its 3 month free
or alternatively use openrouter with your fav ai
not sure why, since iirc it can't do accents
wdym
not if u account for cost of reasoning
but its crazy how close they are pricing them
Its accounted for in the per million token price
is it?
Should be otherwise it would be free reasoning tokens lol
they charge for the CoT tokens, even though we can't access them
In any case if the price reduction is enough that even pplx can use it then they can definitely up the limits allot...
oh u mean on the chatgpt site it self
so they are technically free to make up a number if they wanted
pplx is using it...
Because you all asked: we're going to double the rate limits for o3 for Plus users. Rolling out as we speak. Now go do awesome stuff with it!
o3 limits doubling again
@green crystal 🗿
openai or perplexity?
openai
openai
there is no Plus plan on perplexity
so 200k?
huh
rn 100k is output token
what are you talking about
oh this is for subscribers nvm
Ahhhh
Dammit not even 50 per day
i tho its for the api
It was originally 100 uses a week of o3 on the chatGPT plus plan

now its going to be 200
i got confused with the api
it resets weekly
or atleast 25
was originally 100 a week
Only a few models can get the internet search tool
I use 3 or 4 because I'm using it sparingly lol
bruh only double
Amazing, o3 and Gemini 06-05
what if you use "bring your own api''
you can do that with something like openwebui
and link it to googles own api
He might aswell write "here take these bone scraps and shut up!" Lol
@distant aurora https://platform.openai.com/settings/organization/data-controls/sharing enable free tokens here
its already much better than the old 50 a week
Definetly works
meanwhile t3 users getting 100 a month
True but with teams getting o3 pro and the price being reduced by so much it should atleast be 50 per day for plus users
lol
Hopefully people are pissed off enough they increase it more
tbh o4 mini high > o3
I dont even use o3 on the Pro plan lol
Nah there's a clear difference when having it do different tasks in a prompt that work towards a single goal
why?
overall intelligience is the same, o3 has a higher budget for tool calls
o4 mini high often performs better and much faster
@distant aurora explained the difference nicely
its mainly just speed
💩
so its o3m and o1 all over again?
And it matches with what I experienced
o1 sucks compared to o3 mini high
yea
In any case let's hope people toss enough pitchforks at OpenAI
i heard that they were planning on putting 2.5 pro to a prem model as well
For more stuff
Why o3 better o4-mini?
peak SOTA gemini
LOL
Definitely worth the $250 /s
meanwhile 4o
its not, its slower and they have about the same level of intelligence
now 200
oh i thought it was 50 b4
so like 28-29 queries a day
honestly 30 a day aint that bad
So why pplx set o3 instead o4-mini?
pplx just replaced o4 mini with o3. surely o3 might be slightly better?
just bc pplx did something doesnt mean it was the smartest thing possible 😭
also the same reason to why they added gpt4.5 and removed it like 3 weeks later
Cost or speed?
but there must be some reason behind it? 😅
4.5 just sucked 😭
o3 is more expensive then o4 mini high
haha yeah
forgot about the model until earlier td
nice

@obsidian umbra source for leader board?
Do you know any good leader board for creative writing?
And longform writing as well
honestly not sure
there is eq bench or something?
but writing is very subjective
🤷♂️
Yea i cant get my head around eq bench
The result from eq looks suspicious
Opus 4 being top 2 doesn't sound legitimate
@distant aurora So I paid for another $1 month of ChatGPT Team, not sure if it'll stay up, but send me a ChatGPT account email and I'll invite you
Woah mobile has it now too
Deepseek r1 528 is worse than older versions apparently, gemini 0605 feels like it should be higher @obsidian umbra and opus 4 is for coding. Deepseek v3 may make sense i heard it's good for brainstorm
That's right sorry for ranting on you
The Elo solver shifts all the scores around whenever a new model is added. To counter this, on the leaderboard we anchor the scores such that DeepSeek-R1 has a score of 1500 and ministral-3b has a score of 200. You may see intermediate scores shift around a bit; this is just a quirk of relative rating systems.
r1 maintains a score of 1500
Iam just lost on which is better and lab is telling me sonnet 3.5 is the best
Hi guys, is perplexity doing well as o3 model is present now?
what
can pro subscribers let me know if perplexity is useful after o3 or not
anything improved?
usefulness is subjective
what was wrong with it before that caused you to not want it?
i am asking about improvement
ok leave it buddy, let some one who understood me answer
yea, but what improvement
u ask a broad ass question then refuse to give specifics 😭
If you like writing yes o3 has a very high writing skill and soon they may increase the output which will also be useful in its own way. But if u are a coder definitely no use
you are asking ass follow ups, when somthing new comes up i am asking general question how is the improvement, and you are irritating with some arguements, you just ignore and dont respond to me
It's also friendly and short great for every day use
o3 was added
thats it
if u see that as an improvement
thats on u
it is definitely an improvement imo
Anyone know when the next business fellowship is?
0605 2.5 pro was also added
hi
id rather have o4m ngl 🫢
can I make vídeos with perplexity pro?
no
hi bro, definetly im not a coder. But i want more message cap limit of O3 and what to use it for conversations and troubleshooting tasks. do you think its good? i just want to stop chatgpt subscription and come to perplexity, do you think i can use perplexiy in place of chatgpt?
yea
@haughty birch a great improvement imo i would say is if they add r1 0528 because it's one of the leading model rm and they also added gemini 0605 which has been crushing so far
as long as context doesnt matter for ur task, pplx is fine
but the pro versión doesn’t include chatgpt pro which can make them?
nice set of stickers you got there. Lol
@obsidian umbra
im not sure if it's even accessible through api
Why do u need gpt in the first place is the question buddy you should know agar you are looking for uf it's for every day use i would say pick the cheaper one
chatgpt plus and perplexity, both are 32000 context window, so want to know in reality, which is good, chatgpt plus or perplexity pro,
both are same price bro
chatGPT context window is higher
pplx has alot of other limits, why dont you just try out pplx for a month and see if it fits your taskes?
but their page says plus gets 32K context
I think perplexity is more
wait lemme check once again
yeah sorry, i was mistaken. plus is indeed 32k.
i think i might have saw pro plan
(or a plus plan way back if there was more than 32k context window)
in that case, yeah i would say i find pplx better than chatGPT plus for my use cases
plus plan is never more than 32k bro, i can confirm that being a chatgpt user since it is born
good bro
i just need more O3 model messages with less cap limits, and i want on top of internal knowledge (like in spaces in perplexity)
in chatgpt plus i get only 200 o3 messages per week, that is the limitation
i think you should get 300 every day with pplx pro. But i think it would be a slightly stripped off version of o3
In terms of reasoning effort, pplx is actually higher than ChatGPT
not sure, but i dont think perplexity guys wont strip it off, because o3 api price is now cheaper than gpt4o model
wait what? damn.
o3 better on pplx than on open ai's own service?
Could you point me to a source where it says so?
it is his personal experience bro, he shared
see perplexity uses its own mechanism for searches, these models are an extra intelligence on top of perplexities searches. so thats why he meant pplx is more good at reasoning may be
@royal ingot discussion here
Source (note that the source appears to be outdated for ChatGPT app after I did extensive testing): https://github.com/asgeirtj/system_prompts_leaks/tree/main/OpenAI
o4-mini-high is indeed 512 on ChatGPT, and o4-mini is 32 instead of 64
should i try for a month, or go for an year plan as getting discount. that discount tempts me
in perplexity
def go for a month first
I wonder if the free month for 7 consecutive days on Windows app offer is still active
The advantage of Perplexity is that it has more models, which openai is not able to update its databases to date...
got it, but i need features of chatgpt, many of them are present in perplexity. and i love openai models performance, and now they added my favourite model o3 in perplexity
thanks for this matured suggestion bro, it really helps me to get an opinion after using for 30 days. thanks for saving me for paying for an year, that offer is really tempting me.
doubt
I... got it...
@timber niche
welcome to pro lol
@left stream would be wary of you.com these couple days, they recently changed their pricing page and moved the original "unlimited" for teams to 2x of Pro
not liking the changes that they have done recently
"Limited" (Poverty Edition) 😭
ah lmao
wonder what the limits are for team
That's certainly one of the responses of all time
the thinking time has been halved compared to o1 pro
jesus 😭
not sure what "tasks" means exactly, but it sounds every bit as overpriced as Manus
and I doubt it's as good as Manus
atleast manus has really cool features
mhm also flowith
I could not get used to flowith's UI
yeah also it uses a shit ton of credits
for the simplest thing
btw have you been experiencing any hangs on o3 pro?
seems like a flaw with all of these "agents"
no, but I'm only on my second query
ah, been trying to see if theres a limit
used like 25 prompts so far
meanwhile plus users with their cheaper than gemini 2.5 pro model only getting 200 per week

pro
i hate how they stopped showing the amount of DR left on chatgpt
I got new UI again
pplx ui changing is infectious
@cobalt vale this has been disappointing
time to fire all the devs bois
ai gonna takeover
woops forgot /s
everything is going backwards
we need another deepseek

haha
with the right prompts
yeah
if an engineer tells AI use this library integrate this here like that etc..
it will do a senior dev's job
Will o3 be added now as unlimited model as price is now even lower than 4o?
o3 is the same limit as everything else
nothing is unlimited
and because its a reasoning model, you cant just compare pricing to 4o
its in
i used it
but not unlimited
99% of users will never hit the limit though
indeed
why i love pplx
support team sucks tho
this has been so goddamn annoying
Hi guys, I'm having trouble logging in on the browser using my apple login
how gud is o3 compare to sonnet 4 thinking ?
Wdym
Not sure but chatgpt recently being way more hallucinating than it used to be
With o3 model being around 30% hallucination during a long answer and reasoning
And roughly 50% if i remember it correctly for short q&a
The worst for o4 model being 70% for short q&a
Seem like o3 beats sonnet but Claude beats o3 in coding and math
40% for long and reasoning answer
Although OpenAI said that o4 was designed to respond faster but still we should double check before use any information of it
So better use claude bois
even on my heaviest days i dont think i even used 300 messages
ah
Team sub?
yea
yeah but you wouldn't input the questions one by one
for things like that you gotta do it in batches
I had to cuz online exam had one on a page.
I’m 20 and making consistent sales through my dropshipping store!
Super grateful for how far I’ve come — it’s been a lot of learning, testing, and late nights, but it’s paying off. 🔥
I’m always happy to share tips or answer questions if you’re just getting started or thinking about launching your own store.
💬 Feel free to reach out — I’m here to help!
70 pages total.
admitting to cheating is lovely
😭
What's lovelier is that there were questions with answers absent in study material.
At that point, I gave up.
After studying a subject, I figured that it won't help.
ah..
@distant aurora he awake?
I don't want it.
yes, not all of them though. rollout
you can make a new account, and only you can see your messages
As expected
Somehow I was charged extra money when I selected card as default payment method when I selected verify organisation option but no credit was added to my account.
So I'm 10 dollars down now.
No thank you.
I'm keeping my distance from open ai playground.
Contact support
This you're free from being charged bro
Just enjoy it
Lol
Fuck you bro. You and your lols.
anyone got good sys prompts for discord chatbot
ChatGPT Team is entirely separate from the playground... you could make a burner account without payment info
Lmao salt
Just be rational and contact support
Yes. Salt. that was worth a month's electricity bill.
oof
Anyway good luck gotta go back to the mines
I'm trying to cope with it while people have been making fun since last night.
I didn't make fun
I removed my card.
Last time i raised this concern others told that's not how API works
Just ask them what happened and why it was this way
I'm not adding it back to risk making it minus 15 dollars.
Pretty sure they can refund you
Don't have to add it back
Can just have support refund to the bank with details later
if the extra $5 wasn't even credited, they will 100% refund you
That bottom one. It wasn't added.
And I didn't even select any recharge option.
different amounts?
Yeah
It ain't even showing in billing so wtf was I even charged for.
Playground is a scam. Gonna tell people to stay away from it.
What happened
Unusable crap to hog 5 dollars.
Why did u got charge for the extra 5 bucks
No idea.
I just hate myself for everything that's been happening since last night.
Could have put that money in open router instead.
Or saved it for later.
you are going to resolve it right?
Bank portal isn't working.
The fraud and dispute section says contact administrator. So it's not functional for now.
Can u press view
That's from 5 hours back but sure.
yep they literally can't charge you without your permission and without crediting it
I just sent a hi
Or someone in ur team did
Last night.
Did u invite someone else ?
Do you have access to o3pro btw
@distant aurora use the support on help.openai.com and make sure to specify that you want an actual person or else its just going to be AI
its not
Teams also
team users get limited amount
Mine plus
Pro and Teams, Enterprise and Edu coming shortly
You have teams sub
Yes
Mine is team plus
Is that count
should?
Potat had same as you and got it
There is no Team plus, it's just Team
its still rolling out
You'll get it shortly
just wait a bit then i guess
you might need to wait
Yeah same
I see
also the amount of hangs ive been getting with o3 pro is insane
Can u give some reviews
on...?
o3 pro

nothing super crazy from what i've seen
o3 being 80% cheaper was the major change
yes, all three
Noice

16!!!!
i ran 6 queries in a row
and all of them failed
must be perfect cube, not just fill in the blanks
Do o4 mini
Nice
i ran the same prompt with all openai models
O4 mini > r1
and claude
Can you do grok?
not surprising
🗿
looks like it got stuck up in the controls
Gemini failed I think.
and just failed to make the actual "game"
Sweet.
grok3 mini on t3
Lmfao
2.5 pro on gemini
Gemini seems to be better
Damn. So it does well with code with sucks with simple questions?
Than grok and o4 r1
its been a hit and miss for me
Cusor actually good
I was talking about Gemini.
Oh mb
4.1
Try 4.5
just a copy pasted vsc 🤭 but i wont go there
haha
is that new google gen
surprisingly well ngl
flux dev
too bad the model still sucks 🥀
the balls eventually escaped
online
nice
is kontext dev out?
never heard kontext dev
its not
Kontext is the new Flux series that outperforms GPT Image
oh shit fr
should be?
