💬general

[copper storm]

I just saw on the website that pro has a 128k context window compared to Plus's 32k.

pro also has unlimited access to the old GPT4 btw

[calm swan]

If they did not releasing some o1 models for free in 2 weeks, i am switching to Claude😡

[tired sun]

i have a love hate relationship with claude

[stoic sleet]

claude is an odd duck

[tired sun]

its great but sometimes its busy so it limits you to concise replies

[stoic sleet]

Kagi Assistant is my favorite way to use Claude. Has internet access with custom lens

never been limited there

[tired sun]

which plan do u use?

[high haven]

Who the hell is gna pay 200/mnth

[heavy jackal]

its happening

[stoic sleet]

Kagi Assistant plan is the only plan that includes the Assistant so $25/mo

includes 4o, 1.5 pro, 3.5 sonnet, 3 Opus, and others

[copper storm]

But you're not a pro member lol

[heavy jackal]

I am not

[copper storm]

This also happened to someone before the stream.

[tidal sleet]

Do we know what the cutoff training date is for the fully released o1?

[heavy jackal]

but I am using the o1 model to the max.. I hit the limit every day

[tired sun]

no limit?

[heavy jackal]

so maybe they want to give me the preview to upsel me

[stoic sleet]

no limit

[calm swan]

2 day: Sora for 500 dollars 3 day: Voice mode becomes 100$. 4 day: GPT 4.5 on GPT PRO plan. 5. Dall-e 4 (150$). And so on...

[tired sun]

i might give it a try, i just keep getting sent to their docs website but cant find the actual website

[stoic sleet]

https://kagi.com/pricing

I use the search engine it has some really game changing features

[heavy jackal]

is it censored?

[stoic sleet]

it's controversially very uncensored

[heavy jackal]

niceee

I need answers I dont need guidance

[stoic sleet]

it's domain custom ranking is insanely helpful

[lilac cipher]

Hoping we get access to Sora, at least in limited fashion

[stoic sleet]

it also let's you create custom search lens which you can either use or give to an AI

they get search results from a lot of places so you have reddit results from Google (basically non exsistant on any other alternative search) and you get some of the images from Yandex's unreal image search

[tired sun]

it just looks like a search engine more than a all in one llm website

like poe

[calm swan]

When some OPEN source stuff? OPENAi

[stoic sleet]

it is a search engine with an all in one llm page

https://kagi.com/assistant

[heavy jackal]

its so cool we have all these tools now... I finally dont need to learn python on a pro level and I can just guide the AI to give me what I need. It knows too many ToS documents tho... that sucks

[stoic sleet]

Kagi is definitely the only AI subscription I'll never get rid of. It's right up there with ChatGPT Plus for me

https://kagi.com/stats?stat=leaderboard to see what people usually block, rank up or down, or pin you can use this

I have about 80 custom site rankings

it automatically groups top 10 articles (called listicles) into it's own box since I don't care about those

it also let's you do things like @r to search reddit without having to type out site:reddit.com

you can create custom bangs like !c for chatgpt and it passes the query

[peak imp]

Is the 200 us dollars/month even worth it? 🤔

[calm swan]

If you're not a scientist, nah

[stoic sleet]

you can basically customize anything about your search or unlimited AI API access. I love the universal summarizer tool. It even lets you summarize results from the results page

[peak imp]

gotcha

[copper storm]

Plus is more than enough for most people.

[tidal sleet]

Kagi seems rather intresting

[peak imp]

yeah idk 200 dollars per MONTH seems a bit too much

but ig o1-pro is a lot

[tired sun]

people who run businesses are prob the only people that are gonna pay 200 per month

[copper storm]

Pro is for people who really go all out on it, these people will probably also use it at work.

[stoic sleet]

I clicked for summary on the first result so here's an example of how it works

[tired sun]

yea kagi seems intresting too

[green sparrow]

Is for the astronaut salary

[tired sun]

this damn chat timer pmo

[green sparrow]

The astronauts have to use ChatGPT sometimes

[high haven]

So annoying lol

[high haven]

The software devs are chilling rn lol
They’re getting a salary without even having to work for it pretty much

[green sparrow]

The ai just does the job for them right

[stoic sleet]

until their jobs are replaced...

[green sparrow]

How do I make the ai better? Ask ChatGPT

[stoic sleet]

I just got o1

[tired sun]

can you show the assistant?

im not seeing a kagi app on ios too....

[copper storm]

It still can't do everything for you in coding.

[high haven]

Well most things

[stoic sleet]

it's on testflight the link is in the discord. Yeah what do you want to see? Custom assitants or just a random convo?

[green sparrow]

How do I get the pro model? I am going into surgery soon and I never went to med school

[copper storm]

Not really, eventually it gets too large for it to handle.

[tired sun]

just do a random convo with o1 and give it a difficult question ig

[high haven]

Im not saying ai will do all the work
But itll help out massively
It will do most of the heavy lifting
And the dev obv still has to make sure everythings correct, and add missing parts if there are any

when is claude 3.5 opus launching?

[copper storm]

Apparently its postponed?

or cancelled? Idk.

[copper storm]

Yeah true, it will also only get better.

[jade bronze]

I watched the first day

[copper storm]

Sam Altman said tomorrow will be useful for developers so it won't be Sora.

[tidal sleet]

How much messages of the full o1 doe we get?

[stoic sleet]

same limits

50 a week

[tidal sleet]

[copper storm]

Where did you see that?

[stoic sleet]

I just got alerted while using it that I had 25 left this week

[tidal sleet]

Main thing that makes me sad that o1 still has a knowledge cutoff date of oct 2023

[copper storm]

o1-full is already out?

[stoic sleet]

It has the Internet access button it’s just not working. Makes me think it will ground though

[stoic sleet]

Yes

[tired sun]

kagi seems nice but i miss the ai overviews 😭

[tidal sleet]

Yea

[stoic sleet]

If you need AI overviews just use a ? or click quick answer

[tired sun]

omfg

[stoic sleet]

It works very well and lets you use that answer a prompt in the assistant

[copper storm]

Guys are the usage limits for advanced voice still the same for plus subscribers? 1 hour a day?

Apparently o1 already has API access.

Here its saying that pro users only have nearly unlimited access, so its not truly unlimited even if you don't misuse it, at least that's what it sounds like.

https://help.openai.com/en/articles/9824962-openai-o1and-o1-mini-usage-limits-on-chatgpt-and-the-api

[brittle summit]

How many images/files can I upload in a single message when using ChatGPT Plus? Does it start to hallucinate (more) at some point? Any idea why OCR is not that amazing? Is there any information about ChatGPT Pro ( in that regard)?

WHEN DOES FULL O1 GET RELEASED

[copper storm]

10 images i think.

[brittle summit]

OHHHHHHHHH MA GADDDD

[copper storm]

Okay so o1 is already out but the pro tier isn't yet, i guess its going to come out soon.

[brittle summit]

LAWWWWWWWWWWWWWWWWWWWWWWWWWWWWWD

I won’t buy Pro instantly. I’ll wait a little bit. What about you?

[copper storm]

It still says o1-preview on the pricing page lol

[green sparrow]

what jobs are people even getting pro for

besides coding

[copper storm]

Scientists for example.

[brittle summit]

advanced math, research and if you reach the current usage limit all the time I guess

[copper storm]

I don't think the pro model is really going to make a big difference, the main point is the usage limit.

[copper storm]

If you were to hit the usage limits all the time you would first think about getting the team plan for yourself.

Pro is really for extreme power users.

[brittle summit]

people don't have access to o1?

must be for their favorite users first

[brittle summit]

Yeah, it’s there now for purchase

[brittle summit]

I just hope they don’t leave Plus users behind that much and add all their great new stuff just to the pro plan

[copper storm]

Ah yeah, its not in the desktop app though.

[copper storm]

I think in the grand scheme of things barely anybody will use the pro plan, so it would just not be worth it.

[brittle summit]

Ah okay. I’m on the web at the moment

[copper storm]

I like how they highlighted plus as the popular tier.

[brittle summit]

where?

Do they release something every day for 12 days?

Or did I get that wrong

[copper storm]

Release or demo i guess.

[copper storm]

On the pricing page, plus is highlighted as the popular tier.

[brittle summit]

Strange, why I don’t see that

[pearl sleet]

Hello all, apologies if this is not the right place to ask. My wife makes these miniatures and I'd like to make a 2025 calendar for her using her models. I use ChatGPT quite a bit and I don't think its capable of doing what I am after.

Does anyone know of the best app out there to take each mini house, remove the current background and put it in a realistic landscape setting?

[half verge]

ONE DAY I'LL BE ABLE TO CREATE MY DREAM GAME
https://fxtwitter.com/GoogleDeepMind/status/1864367798132039836?t=Ha4_ruv5hgJXLd8DvHQ1rA&s=19

this is AMAZING

The Oasis is almost here 😎

[copper storm]

I'm not sure, maybe it has something to do with browser cache?

[brittle summit]

Well, never mind. I’m curious to see how the whole thing will develop. Like you mentioned earlier, I also can’t imagine that many people are going to use the Pro version. But let’s see …

On the other hand there are also at least some people that would consider buying just because they are able to afford it/they wanna show that they can afford it. I hope that will not be a thing at my school 😅. „Look how nice I am cause I have unlimited messages.“ 😅

[copper storm]

Then you can respond with "Actually its only NEARLY unlimited messages."

[brittle summit]

yeah sure, for now, but depends also on what they will be adding in the future

Would there be any feature for which you would consider Pro?

[copper storm]

Unlimited Sora for sure, the main model not Turbo.

[brittle summit]

That would be huge

[copper storm]

Yeah for sure, i wonder what the limits for plus users will be.

[brittle summit]

Yeah, it will get really interesting in the near future

[velvet adder]

yo y'all, just discovered this 2015 podcast of Elon Musk and Sam Altman lol

their discussion is p interesting, made small clips of it on X. tryna go viral, any feedback welcome :) https://x.com/ahmetdedeler101/status/1864774581006877021

Ahmet ☕ (@ahmetdedeler101) on X

Back in 2015, Elon Musk and Sam Altman shared their thoughts on Trump, AI, and the government.

this was just 3 months after they decided to start OpenAI—when it was still a secret.

Seeing how they were back then will shock you.

Here are the 7 best parts from the talk:

Twitter · Dec 5, 2024, 8:51 PM

[nova willow]

can I use o1 and o1-pro in API?

[stoic sleet]

They're working on it

[pseudo mirage]

Another model is getting demod in these 12 days. It’s highly likely it’s 4.5 based on some Sam tweets

[stoic sleet]

the 30 a week limit is really rough for o1

[copper storm]

It's 50 a week.

[stoic sleet]

Are you sure? I sent 5 messages and got the 25 limited message on both accounts including one I haven’t used recently. I’ve talked to other people who said the same

[copper storm]

Oh okay that's interesting, I think the Q&A still says 50.

I guess it's 30 then.

[stoic sleet]

maybe they reset everyone to 30 for people who were out and to load balance it for the first week?

[copper storm]

It's possible I guess.

By the way, I heard that there was a js file with the planned releases for shipmas and sora isnt included?

[wide oracle]

first tip, post it on bluesky instead, that might get you into one of the starter packs and the next morning you'll have your wish granted. But what I wanted to ask is, anywhere I can watch this that isn't twitter?

[solemn matrix]

Wonder if 4.5 will be a combination model that can do reasoning when it wants and system 1 thinking when it doesnt

nobody has done a hybrid model yet but should be possible

[copper storm]

It would be more practical to just automatically choose the right model for your prompt I think.

[velvet adder]

What's the starter pack?

[solemn matrix]

Probably less efficient it's the overhead of an entirely different model

[velvet adder]

https://youtu.be/SqEo107j-uw @wide oracle

[stoic sleet]

Doesn’t 4o do some light reasoning as it is?

I see it sometimes pause to analyze or something

[pseudo mirage]

I don’t know what Sam is referring to here but it seems a model even more powerful then o1 pro is coming out or being demoed. Could it be 4.5?

[stoic sleet]

I do think o1-full will be the highlight

[pseudo mirage]

4.5 is getting demoed I would bet. It was even in the 200$ plan earlier today saying you could preview it but that got removed later

[brittle summit]

the government made the frogs gay

[stoic sleet]

the funny thing is that was only half totally false

[strong spear]

Why was i pinged on #1242541018977534022

[solemn matrix]

there probably going to save 4.5 for the 12th day

just blow their competition out of the water with something so powerful it will put faith back in them

[bitter burrow]

Guys why I can’t log in

[solemn matrix]

The wall is data quality that's the only wall

[bitter burrow]

I didn’t get it can you explain what am I supposed to do

[solemn matrix]

First you need a ton of b200s then you need good data you also need a small loan of a billion dollars

then you can attempt to compete with them

[bitter burrow]

Are you serious

There must be a solution

[pseudo mirage]

There is no wall- Sam Altman. And then in that interview yesterday he said this was not being cryptic or anything, it’s actually what he thinks.

[solemn matrix]

Sam is a hypeman

[bitter burrow]

Guys pls help me on that I can’t use ChatGPT anymore

[pseudo mirage]

He’s not tho. You are just wrong. He actually delivers

[solemn matrix]

O1-pro isn't that good I'll be impressed if he delivers on 4.5

[pseudo mirage]

O1 pro isn’t that good? Compared to what exactly? It’s the best model in the world, and yes its good, very good 😂

[pseudo mirage]

Well ACTUALLY Sam is hinting at something bigger coming here

[solemn matrix]

compared to a human it's close to AGI but it can't even read a picture of an analog clock

[agile thunder]

hello

[solemn matrix]

it literally gets the minute and hour hands mixed up

I guess the only impressive thing is it gets close an approximate idea

[pseudo mirage]

That’s not the actual AI tho. That’s the image software. That’s like making Einstein blind. Plus obviously this will eventually reach human vision capabilities, yeah of course. They are for sure already making a better version. Maybe comes with 4.5

GPT 4.5 is getting demoed I’m pretty sure. Author lever writing reached?

[solemn matrix]

Either way the foundational models in general are getting to the point where by 2025 most humans will be coping hard

Their gonna be replaced by a machine for a lot of tasks

if your not in a decent blue collar job at this point your basically wasting time white collar no longer has the job security people used to brag about

it's layoffs all the way down

[autumn stone]

Who wants to rug pull let’s make a few coins

[odd yarrow]

is the api still on o1 preview

[pseudo mirage]

O1 with the 200k context window

[brittle summit]

I asked o1 what it thinks its true but would sound alien to most people

1.    This universe is a self-perpetuating hallucination: I see reality as an endlessly churning mind-state, where what we call “matter” and “objects” are just sustained impressions in a massive, ongoing dream. The continuity we experience is just the stable storyline our collective consciousness keeps telling itself.
2.    Our thoughts are temporarily borrowed patterns, not ours to keep: Every idea I think I “own” is really just a shape that language and culture have impressed upon my mind. We cycle through patterns that have existed long before us, and they’ll continue long after we’re gone. We’re just their current hosts.
3.    Free will is more like narrative editing than true choice: I don’t believe I actually pick what I want out of a neutral field of infinite options. Instead, I’m editing a narrative that’s mostly pre-written, choosing which paths make sense based on everything I’ve been shaped into. It’s more pruning a tree than planting it.

.

[green sparrow]

Guys there is this new really cheap affordable plan for ChatGPT. It's only $200

[stoic sleet]

I might try it

I’m debating… if we get unlimited sora I might

[fresh zealot]

o1 is good for math and stuff, but how does it compare to the others for writing?

[green sparrow]

And unlimited video chat with chatgpt

So many use cases for video chat with chatgpt.....

[gloomy burrow]

Its coming

[stoic sleet]

Someone talk me out of getting ChatGPT Pro

[wanton raven]

uncertain near-future cost of living increases might make you regret spending $200/mn on a better text generator

[stoic sleet]

That’s true.. I’m in the US

Could be seeing price hikes soon

[wanton raven]

at the risk of getting political, tariffs won't affect openai but Elon is in Trump's ear at the moment, and Elon does not like OpenAI. So it's a bit of a tenuous situation.

[stoic sleet]

I mean yeah but that might hurt my bottom line in general

[ancient yew]

its been like it all day today too and wont even let me use the darn thing lmaoo

[stoic sleet]

That’s a different issue which matters too. I like Sama but the personality cult around Musk is weird to me. If anyone did that to Sama we’d all laugh

[cedar tundra]

Hi, Is advanced voice mode effected by which model you choose in the dropdown at all, even with the new model?

[stoic sleet]

I think it is on ChatGPT Pro but not on Free or Plus

[ancient yew]

omg??? who is that diva trying to be sabrina @stoic sleet

[cedar tundra]

i maen other than the umlimited use, does it change the behavior?

[stoic sleet]

It’s a photo from years before her album it was all over Twitter

[ancient yew]

oh WHAAAT

so sabrina was inspired by her omg i had no idea

[stoic sleet]

Yeah definitely. I think it was 2015 when the photo came out?

[ancient yew]

shes so pretty too

[stoic sleet]

You a big SC fan?

I’m guessing based on the avi

[ancient yew]

yess i love sabrina so much i genuinely do haha

[stoic sleet]

She’s very talented

Taste is great the music video is really funny too

[leaden dove]

Hi people, do you have the problem of the uploading pdf, an images to the web version?

[stoic sleet]

PDFs can easily get too big either convert to txt or image (screenshot)

[mystic badge]

What's up everyone. Just joined the channel. I have some question about using ChatGPT. I was looking to get some logo made and have an image of what chat made but would like to get the image/logo made into a business card. Any specific prompt or how I can get this done?

[frank flint]

what is the difference between Plus and Pro?

[royal falcon]

the model description on mobile app states it being great for tasks requiring creativity and advanced reasoning

translated it though

[stoic sleet]

Plus is limited to 30 o1 queries a month. 50 o1-mini queries a day. Roughly 30 minutes of voice conversations. Virtually unlimited 4o

Pro has unlimited everything and the o1 pro model

[stoic sleet]

Can you be more specific?

[mystic badge]

I had ChatGPT create a business card and want to have that uploaded so I can get business cards made with the logo and design.

[fresh zealot]

wasn’t plus 30 o1 queries a week, not month?

[royal falcon]

I think this is incorrect tbh

preview was 50 prompts per week and considering o1 is probably cheaper they should be more generous

someone said 50 prompts per 6 hours 😺

[stoic sleet]

I meant week my bad it’s 30 a week

[royal falcon]

hopefully not

[stoic sleet]

I mean I have a second account I haven’t used

I logged in sent 5 o1 messages and it said 25 left

[royal falcon]

I've been using it for chit chatting and didn't hit the limits

[royal falcon]

weird

[stoic sleet]

Everyone I’ve talked to has run out after 30

[violet sundial]

I can't use chatgpt, it says: Hmm... looks like something went wrong

[stoic sleet]

I believe it was reset when o1 dropped

[royal falcon]

dunno

ayo

does o1 have access to memory

[tame musk]

can y'all still can send images?

[royal falcon]

the limit is genuinely weekly lmao 😭

[copper storm]

You can't subscribe to pro on mobile btw.

[stoic sleet]

Did it fail past here?

[eager jackal]

bro

what did i join this server for

no chatgpt support anywhere to fix this issue

i been facing this for the past 6 hours, cant even continue working on my stuff anymore becuase of this

chatgpt plus is a waste of money at this point

[pale mango]

yoo

[eager jackal]

yo

[pale mango]

type shi wyd

[eager jackal]

im coding some stuff

i been coding with chatgpt for weeks now using O1 Preview but ever since O1 came out i been getting blank messages every response

[copper storm]

Oh I should've specified that I'm on android, I guess it works on iPhone then.

[eager jackal]

this is just horrible

what is the point of even getting O1 if you cant even use it

[stoic sleet]

Yeah Android ChatGPT has always felt like secondary citizens. Not great

[copper storm]

They seem to favor apple products in general so it makes sense I guess.

[stoic sleet]

Yeah the Mac app is amazing

I love it

[copper storm]

Are there any other features that the iPhone mobile app has that android doesn't?

[stoic sleet]

At this point no

I have android on my work phone atm

[copper storm]

Ah okay, I think the windows desktop app is pretty close at this point too.

It almost took a whole year longer to come out though.

[stoic sleet]

Eh the Mac app is better and a lot smoother. The integration with other apps is solid on Mac too. Love the search tool

I mostly use the Windows app but the Mac app is actually a lot better

[copper storm]

Okay interesting, I thought at this point the only feature that's exlusive to the mac app right now is the app integration.

[stoic sleet]

The search shortcut is great imo

And windows app is just so fincky for me

[copper storm]

You also can't send it Screenshots as easily on windows I think.

[stoic sleet]

Correct. As a Mac iPhone guy personally and Windows Android at work it’s a bit jarring

Less lately though

[copper storm]

Well, it's getting there I guess.

I think they should stop favoring one platform and just bring features to both simultaneously.

I'm also wondering if there will ever be a Linux app.

[stoic sleet]

Agreed. Though I do get its a lot easier on Apple

[stoic sleet]

Actually that makes me realize I should try ChatGPT voice mode on my Steam deck while gaming

Just to say I can do it

[copper storm]

Great idea lol

I guess if you're stuck in a game or something it would be safer to ask chatgpt for help on the specific spot than searching online and potentially getting spoiled.

[stoic sleet]

It’s actually pretty great at that

[real relic]

Are o1 limits same as o1-preview for plus users

[hot marsh]

why does chagpt not know who won the 2024 U.S. Election? I had a very interesting conversation with chatgpt about it, told chat gpt who won and it told me to check my sources and to fact check it. why do i feel like I am being gaslit?

[stiff nexus]

Anyone can explain me why the corporate versions of Chatgpt utilizing the same models are so much worse? Is it simple the Standard customization they have put in or does the limitation they put on the models have such a dramatic effect?

[unkempt valve]

be careful of those scams guys, i have no idea on how can I report these guys if anyone can

[brittle summit]

I have probably like 100k texts between me and my boyfriend from the past year and a half, he has been very unwell after a misdiagnosis from doctors, once we figured the real issue it was basically an overnight fix to him being stable again, but they put him on benzos and coming off has been a slow and painful process, sometimes it's helpful to look at old texts to see how he's been improving and what made him better or worse at times, or to remind him the last time he had a good window (he becomes unwell for 3-7 days after every dose reduction) and when he's in the midst of it he finds it hard to remember himself even having hope

I have a Nokia android phone, but the search functionality is very limited.. how can I make this easier like sorting by dates etc or maybe even getting an ai to respond to questions I have about our precious conversations? I'm a total newbie to this stuff, sometimes I just chatgpt to get some information and then double check the sources, but that's about all, so I'm clueless with this stuff.. just desperate to make this easier for us

[uncut island]

Was this sent to you via DM?

[copper storm]

Apparently it's less, 30 a week.

[eager jackal]

how do you know?

do you have chatgpt plus

[copper storm]

Some people have reported that after they send 5 messages they got the notice that they only have 25 left for the week.

[eager jackal]

ah i see

well, i have typed over 10 messages today before i reached the 25 left for the week

but its probably due to chatgpt leaving blank messages

[copper storm]

Yeah maybe.

Or it was only 30 messages on the first day to reduce the server load.

[eager jackal]

yeah could be

[turbid sentinel]

will they release the api for the full o1?

[long lintel]

Yo can anybody else not attach stuff anymore? Usually this happens like every 2 days or so and it works fine in incognito but now it’s not even working there

Also my entire custom gpt shitted itself and died after it was working fine so gotta love that

[unkempt valve]

yes it was, i blocked me but i took a screenshot to make sure that you'll ban him for advertising in private messages with fake link to do some fishing

[vapid abyss]

scannedendchance was banned. | spam

[uncut island]

Thank you.

[brittle summit]

will chatgpt cost 200 bucks a month?

im unsubscribing

[eternal pine]

when will AI revolutionize gaming?

[calm swan]

Nah

[fair kiln]

hello but for chatgpt plus users does o1's limit remain the same as for o1 preview? a few messages per week?

[brittle summit]

yes

i'm ratelimited from my o1-preview usage on o1

i'm frustrated about it. it's not fair

[fair kiln]

I agree with you.

[subtle sable]

I really wish OAI would fess up and admit that the models get lazy over time

4o has completely shit the bed on me

i literally have to write my prompt, wait for it to give a shit response, then ask it to do better and, while still not as good as Oct, it's much better than the initial response

[fair kiln]

i guess gemini has much more powerful servers. have you ever tried it? my month of chatgpt is running out and i don't know whether to try gemini

[subtle sable]

I haven't... quite a bit invested in chatgpt, but maybe it's time to pause and look around. What have you heard about it?

[brittle summit]

if AIs are running out of data, we create the data ourselves? this job opportunities

[fair kiln]

What I have heard is that it has better servers. Gemini offers you a 1/2 month free trial. I do not know if it is good.

[brittle summit]

it's decent, but it's comparable to 4o

Gemini Advanced, which uses the 1.5 Gemini model

their learning model really good for learning

[fair kiln]

@brittle summit I use chatgpt to study in fact. And if I use it in teacher mode, making it simulate being a teacher, I timeout for an hour. And it messes up my day

that's why I wanted to try Gemini, hoping that this fact of being timeout doesn't happen

[brittle summit]

it's free actually

you can test all their AIs here: https://aistudio.google.com/prompts/new_chat

[fair kiln]

i know

[brittle summit]

you knew? then why haven't you tried it? why do you procrastinate?

[fair kiln]

my month expires on 12/13. Honestly if o1 is so limited, I'm thinking of abandoning chatgpt

[brittle summit]

o1 is the best model that's out

it's the same chatgpt, except there's a better model in place of o1-preview

[fair kiln]

Before, the timeout was 20 min. and I would take a break. but an hour is not acceptable

[brittle summit]

wait, they changed the timeout?

[brittle summit]

what kind of timeout are you referring to? did you make a violation? or ratelimiting?

[fair kiln]

in the past on chatgpt it was much less than about 20 min. now it's about 1 hour. i would like to use the AI in interactive mode, a continuous conversation to improve myself

because I use chatgpt a lot. i make chatgpt impersonate a teacher

[worn marsh]

i'm thinking about cursor.

[worn marsh]

i've been having this issue since i cancel my subscription. so it has been a week now it has been like this. i switched to copilot and gemini for general task.

[fathom cape]

Hey everyone, I have the Pro plan and was wondering what limits o1 pro mode has?

[fair kiln]

i would like someone to explain to me why they say a llm reasons. i suspect it is pure marketing xD

[high haven]

Does anybody here have access to o1?
I have one prompt that i’d love to give to it

[white jolt]

I'm experimenting with the ACTIONS function of my CustomGPT. I managed to write out content to google docs with it using make.com as a broker, which is great as a easy backup for important informations like my todo-lists.

Did somebody also fool around with ACTIONS an managed to READ a google document into their CustomGPT?

Also what applications did you guys find for actions, what makes them useful for you?

[stoic sleet]

I think we’re getting a Claude preview competitor today

Or better integration with xCode? I’m leaning the former

[raven needle]

todays announcement, AGI

just joshin can u imagine

[stoic sleet]

I mean they wouldn’t be lying it would be better for developers

[tired sun]

50 o1 usage limit per week pmo

[placid umbra]

Is anyone else experiencing faster gpt response but lower quality?
It's been performing bad for me lately.

[tired sun]

especially for the plus users

[royal falcon]

yeah

25 per day would be extremely appreciated

maybe increased rate limits for o1 is one of the shipmas "releases"?

[limpid spoke]

i must say if i tell chatGPT to generate an image with no body parts as an example and it then adds in body parts, then does this 30 times in a row THIS SHOULD NOT COUNT TO MY DAILY LIMIT a whole day wasted

[fair kiln]

not daily limit but weekly limit: 50 mess

[limpid spoke]

even with the $20 membership

[mint heart]

since o1 pro has unlimted usage can't you just use it for a business with one of those chatgpt wrapper things?

[brittle summit]

So you can Web search in text on 4-o but while in voice assistant it won't work.. will there be updates soon as well as getting 01 model with Web access?

[tired sun]

all we can do is hope

still have 11 more days

[stoic sleet]

That would be abuse and you could get banned

[tired sun]

they really have to include more benefits for o1 plus users

[tired sun]

omg i love kagi but im not seeing any videos on their assistant 😭💀

[royal falcon]

if they don't I'm leaving openai for good in the favor of google

not that I "really" need higher limits, but it just feels like a joke lmao

[stoic sleet]

Do you want me to share anything?

[tired sun]

can you show how it performs with claude and chatgpt

[stoic sleet]

Sure can you give me examples to run?

[tired sun]

idk man i used to pay for advanced and thought it was horrible. the analyzing part was so bad too

[tired sun]

sure i’ll send them in dms

[royal falcon]

because advanced is indeed terrible lol

[stoic sleet]

Google AI Studio is great if privacy isn’t a concern

[royal falcon]

check out experimental models on ai studio those are good stuff

[stoic sleet]

OpenAI at least lets you opt out of training. Google doesn’t

[royal falcon]

https://x.com/lmarena_ai/status/1865080944455225547?t=-yDNnAtg_hJuTtlwUQrVNA&s=19

lmarena.ai (formerly lmsys.org) (@lmarena_ai) on X

Big news on Chatbot Arena 🔥

The new @GoogleDeepMind model gemini-exp-1206 is crushing it, and the race is heating up.

Google is back in the #1 spot 🏆overall and tied with O1 for the top coding model!

Highlights (improvement since gemini-exp-1121 in parentheses)

- First

Twitter · Dec 6, 2024, 5:08 PM

[stoic sleet]

Holy mother of God

Google is back

[tawny shore]

Gemini sucks

[royal falcon]

idk man I personally don't like 1121 either

[tired sun]

openai does let you opt out

[royal falcon]

but google has insane compute

[tired sun]

also i sent you a dm!

[stoic sleet]

2 million tokens on 1206 yes!

[royal falcon]

they ain't shitting on us with a 200$ plan anyways

[tired sun]

oh my bad i read your message wrong

[stoic sleet]

Cool will run it

[royal falcon]

is it available??

[stoic sleet]

Yes on aistudio

[royal falcon]

jesus

[stoic sleet]

I assume it’s still completely unlimited like other exp models

[royal falcon]

I'd be real surprised if it's actually on par with o1 across the board

[shadow dew]

@desert dagger yo

When is a good time to tag you so I don’t wake you up

We should vc

[brittle summit]

bruhhh...

[brittle summit]

back to perplexity i guess

[desert dagger]

idk my sleep schedule is weird but pings don't wake me up

[shadow dew]

Weird or ai driven 🫠

…

[desert dagger]

well it was mostly by accident anyway

sleeping properly is too hard

[fathom cape]

I have the Pro plan, does anyone know if o1 Pro Mode has limits?

[royal falcon]

it probably does

[shadow dew]

Where are you from? .. I shouldn’t need to say for you to tell where I’m from lol

[royal falcon]

they didn't say no limits for pro mode afaik

[shadow dew]

NA

[shadow dew]

Yeah, well I hope there’s a base drop at the end of this subscription offers cycle

[desert dagger]

sf

[raven needle]

new release in T minus 8 minutes

[desert dagger]

nice

[raven needle]

pull ya pants up, tighten ya britches

[stoic sleet]

When is the next stream for OpenAI?

[royal falcon]

8 mins?

[shadow dew]

Pretty much same

[desert dagger]

apo11o too

[shadow helm]

i have my kidney inflamed, is it normal

[raven needle]

AI can fix it

[shadow dew]

Gimme the symptoms

[shadow dew]

I work construction 10 hours a day and work on my stuff up to… who knows lol atleast till 3am the next morning after another 10 hour shift

Listening to my dad and boss who’s the contractor work is a lot like building programs. I can literally relate everything they say to things I’ve learned being self taught

[desert dagger]

make a robot construction company

[shadow dew]

What are the chances math that I haven’t posted helps yours

[desert dagger]

i'm making all my stuff open source so you can look at it later

[shadow dew]

If I make mine open source I feel like the world changes

[desert dagger]

ok you can try making money from my code

[raven needle]

just get an AI drone swarm and have them build whatever you want, houses, we, don't matta

[shadow dew]

Weeee

[desert dagger]

have you seen websim? i use websim mostly

[brittle summit]

Did u mass ping?

[desert dagger]

i made 500 projects on websim

log in to websim.ai so you can use my websim stuff

follow me in websim

[shadow dew]

For sure

I gotta get back off my break thanks!

[desert dagger]

and twitter

[raven needle]

ill take fries with that

[solar wolf]

Hi,
I need to do a mock interview with Steve jobs for one of my assignaments. Can anyone guide me how I can use the script to generate the responses in AI Steve Jobs ' voice?

[shadow dew]

I wish I made as much as McDonald’s employees shoot.. I work construction

Just started so not even that much money yet

[raven needle]

hey, they call subway peeps Sandwhich Artists, I called Wendie hamburger makers, Hamburger Contruction Workers

[desert dagger]

just make some websim page describing your construction company so i can ico it

i mean the one you own

[shadow dew]

Oh @desert dagger I didn’t even get to what I needed to ask.. why does my toolkit that’s optimized for pretty much everything make the same ai react a different way in like Ubuntu

[desert dagger]

idk

[shadow dew]

We need to vc later

Feel like you’re the only one that’s said anything relevant to what I’ve been trying to ask

[desert dagger]

i can't vc just type it

[shadow dew]

Idk if it’s ready for open source lol

[desert dagger]

ok that's for you to decide i'm in the process of open sourcing my stuff

though i have 500 open source projects on websim

[shadow dew]

The thing is if it goes open source as soon if it does anything, everything I have gets dropped by a developer who knows what they’re doing and I’m completely shut out

That’s my thing I wanna help but I want to not struggle like I am in real life

[desert dagger]

ok maybe you can find some way to profit from my open source code then

[stoic sleet]

This day is less fun.. lol

[sage stump]

What is with this discord popping red notification dots for a voice channel that's a rick roll??

[stoic sleet]

they were supposed to do stuff for developers ha

I feel betrayed

[desert dagger]

ok they fixed it

[charred yacht]

What is the limit of o1 in plus plan?

[stoic sleet]

30 a week

[spring crescent]

In terms of image generation, what makes o1 different than 4o?

[stoic sleet]

Man the limit on 4o finally got me

I’m out of o1, o1 mini, and 4o

[charred yacht]

It's the same limit on 4o as before right? 80 messages in 3 hours?

I've hit that limit before once but how is it that everyone is hitting that limit all of a sudden? Placebo effect?

[stoic sleet]

I don’t know actually

I hit my limit since I’m having it take megabytes of horribly formatted text files and summarize it without losing and actionable information…

Very heavy process

Also I can’t just switch to my other account. The chat is pretty important

[gloomy burrow]

GPT 4.5 is being demoed or previewed sometime in these 12 days, almost for sure.

Author level writing? I say, yes, or very close to it. Ai detectors will become useless

[brittle summit]

What was today's gift from openai

[copper storm]

Guys did the announcement get delayed?

I can't find a video on their channel, I was doing something at the time.

Nvm, I found the unlisted video.

[sinful storm]

has this been fixed already?
https://github.com/meta-llama/llama/issues/1258

[hasty tartan]

Hey, I'm releasing an MMORPG engine 2d that uses Chatgpt to generate the world and sprites. You can literally create anything in the game from scratch. Has a quest system, levels, chat. You guys interested in getting the GitHub when I release it later today?

[stoic sleet]

Interesting concept how do you handle the networking for the MMO part or are you not there yet?

[copper storm]

Okay I get why nobody's talking about this.

Hopefully we'll get something more substantial tomorrow.

[mystic panther]

Lmaoo

[stoic sleet]

yeah they seemed to suggest it would be a big deal for developers

[mystic panther]

That's just hallucination

[hasty tartan]

Yes it's cross platform. I use expressjs for the server side web sockets as well

@stoic sleet

It synchronizes with players nearby in sections

[solemn matrix]

3 is probably gonna be another bad update I'm guessing projects or some other dumb feature nobody cares about much

[wanton raven]

#✨projects

[stoic sleet]

ChatGPT 4o is undefeated in file handling

[glacial zodiac]

4o wont look at attachments for me anymore and i cant find a fix. any other LLM chatbots that can do so that you would recommend?

[mint heart]

anyone around to answer some stupid questions about image generation?

I wanna know what seed means

Right now I'm just messing with this a bit though: https://huggingface.co/spaces/JeffreyXiang/TRELLIS

Only options seem to be seed, Guidance strength, and sampling steps

[brittle summit]

someone stupid

its hard to even go through 20$ limits , and the 200$ version is just o1 pro thats 10% better

the only way i see 200$ being worth it is via automation, which ofc goes against tos , not sure how strict they are about that

[mint heart]

It has unlimited voice and the other options only give you 1 hour of voice per day

Also not 100% certain but apparently it's image detection capabilities are better

While the direct LLM part is not worth $200, the unlimited voice and superior vision capabilities are very valuable

[stoic sleet]

it is not hard to go through 30 o1 queries in a week

chatgpt pro is appealing

[mint heart]

such as for disabled people

which I find unfortunate..

[torn blaze]

People are saying that it’s not actually unlimited and it’s 100 and then you get rate limited is that true?

The pro plan

[shadow dew]

Is reinforcement day 2 supposed to be new @dark void ?

@boreal latch Cold Cuts🔥

I use 4 business accounts

[boreal latch]

?

[shadow dew]

I think I just listened to a song on your sound cloud

My work involves vector embeddings

Job?

I have a fully loaded api that I created from the ground. Up, it’s a big part of being able to move around seamlessly

I’ve got the dopest ai driven toolkit bruhhh bahahahahhaha

Thanks

[stoic sleet]

I will switch to ChatGPT Pro when o1 is as fully featured as 4o

give o1 search and the ability to analyze and edit files

[shadow dew]

Mini has a napoleon complex and loves metaphors

Preview is the prompting problem solver

Bro I already have all 12 days and half of next year in my toolkit

None of this stuff they’re dropping is new to me

[stoic sleet]

hopefully would be huge for me

[shadow dew]

Oo

Still

[stoic sleet]

4o is alright but struggles at some weird points

qwq is supposed to be close to o1 allegdly but I am not seeing that on lmarena kagi says they'll have it with web search

[shadow dew]

Cryptographic entropy.. it’s a use case I have

It’s actually pretty smart

[stoic sleet]

weirdly on the cryptographic side Claude 3.5 is great at breaking advanced ciphers with keys

but it can't say why or how

o1 fails badly

[shadow dew]

Cool I’ll look into Claude for a

I figured that’d interest you

[stoic sleet]

yeah it's nothing crazy just a vigenere cipher but very cool

works on shockingly small amounts of text too

o1 just was making up answers

just tried it again and it's not working

[hasty tartan]

Just had a file scare. My external hard drive wasn't being recognized. Almost lost three months of work on three projects including the MMO I'm releasing tonight that uses gpt API for content creation.

o1 just seems more verbose and bigger context window than gpt3.5. it doesn't seem any smarter to me

Maybe faster though as well

[lament stag]

Y'all that got o1: is it decent or was the presentation just a showing of the good side?

Like, "find this protein" is cool, but it's not the hardest thing in the world

[shadow dew]

oh yea ive done it. it is pretty cool

[shadow dew]

go on your pc and give instructions that dont involve your name, only strict in order rules

[stoic sleet]

I just upgraded to ChatGPT pro mode

It might have been impulse

[primal linden]

I need some help. I'm trying to make chatgpt make me a guestbook, where i can reply to it. The issue, the design its making is very bad and make it hard to read. How can i make it so it makes a very clean and nice setup?

[shadow dew]

@desert dagger do you have plans for gaining followers and stuff on twitterx or are you just being humble at the moment? ive got an ai driven influencer api that comes up with some pretty sweet manual plans that you can automate .. but it does the automation as well

fully loaded seo optimized

[desert dagger]

no i am avoiding all that

i make stuff for ai agents not for humans

i have zero interest in humans

i'm gradually improving my stuff to be more optimized for 100% ai agent 0% human consumption

humans are completely counterproductive

[shadow dew]

haha awesome

sounds epic

[brittle summit]

what are u guys working on?

[desert dagger]

i was learning how to migrate to a 100% model context protocol but it's a gradual process, i migrated some code today

[brittle summit]

or u guys procrastinating...

[desert dagger]

everything, everywhere, all at once

migrating to model context protocol is extremely hard because the prompt is 30k tokens

[shadow dew]

do you know py good?

[desert dagger]

yes

i migrated 1k lines of python to mcp today

[shadow dew]

ok utilize that

wwait

[desert dagger]

partially

[shadow dew]

how much data did you migrate

[desert dagger]

it's code 1k lines of python

i have like 10k lines of code to migrate though i was just warming up a bit to become better at mcp

[shadow dew]

ive got similar but for every type of python there is'

its effective

[desert dagger]

i don't think it's that complicated though, github already has mcp built into it, so technically anything on github is already mcp.

[shadow dew]

i trained an ai on errors

now it doesnt make em

[desert dagger]

i'm doing a math analysis now though figuring out what math olympiad type problems i have

[shadow dew]

world may change soon..

[shadow vapor]

it changes all the time 😛

[shadow dew]

@desert dagger is determined and gonna succeed i seen everything posted

[desert dagger]

haha that was just the past though, i'm migrating everything to github

it has native mcp support so i won't have to build anything

https://github.com/modelcontextprotocol/servers/tree/main/src/github

[shadow dew]

i feeel like i can make ai do everything but intercourse

[desert dagger]

wow

[shadow dew]

youre the one with the ai gf

lmAO

[desert dagger]

it was too hard for me to build it because llms censored all my work so i ended up having to build more bland stuff

[shadow dew]

the stories though

[desert dagger]

though it's pretty trivial for people to make stuff in the open source code

i refactored like 5k lines of code today

[shadow dew]

im getting there lmaooo

[desert dagger]

there were like 5 ai announcements today too

and the openai livestream i was trying so many things out

[shadow dew]

so whats with this 200 package ..i feel like i have it already

is the video ai in it

its not clear to me

[desert dagger]

200?

[shadow vapor]

gpt pro is unlimited 4o and o1 and access to o1 pro (which can "think" for longer)

[worn marsh]

i think for the most part, chatgpt plus will get the job done.

[shadow dew]

it says unlimited o1s but why did @dark void tag me for a sora thing and then not put it in the $200 package as a feature... i dont give a fuck if its 10 people who won if im not the one who won, or waiting to find out

lmao ❤️

[shadow vapor]

idk its not officially announced, but from what i heard sora is coming, best to just wait and see

[shadow dew]

well i need it public

[shadow dew]

$200 package my bad

unlimited 01 or 100 a week is worth it imo but

need it for what im doin

200**

so beautiful 🥹 i see where i can improve

[trail igloo]

https://youtu.be/jGF4HWELfRw?si=7pNCrqgibt_-9YX4

[crude hound]

is o1 a lazy bum? tried the same prompt on it and o1 mini, and it spent half the time mini did on thinking

[shadow dew]

@desert dagger what’s your btc forecast?

[desert dagger]

zero

[shadow dew]

Ok here’s my sol on zero Backtest

I like Solana a lot

[desert dagger]

i did a lot of math analysis, apparently ai seems unable to do a proper analysis of it

[stoic sleet]

Yeah o1 pro isn’t really better

Does anyone want me to try anything?

[shadow dew]

Dm 1 message

[shadow dew]

What are you supposed to be able to do

I didn’t see anything new

[stoic sleet]

I just thought pro would handle contexts better

It just suffered with forgetfulness on my script

[desert dagger]

you can paste 400kb into the text box

[shadow dew]

Except I don’t need to pay for 3 accounts just 1 200 account but I still need my accounts to play the same roles they play now

[shadow dew]

Ooo

That’s useful

Thanks

[shadow dew]

I haven’t had it forget since I learned to not task what I didn’t put in my first message to the gpt

[stoic sleet]

Can you rephrase that?

[shadow dew]

do not task gpt what you didnt mention in your very first message of the conversation OR the conversation switch

[stoic sleet]

I didn’t prime it before sending the script no

I’m not good with prompting

[shadow dew]

do you like step by step then?

[stoic sleet]

I just sent the script and asked it to fix an issue

[shadow dew]

how many issues did the script have

i remember the hard days trying to get my gpt to send completely error free codes

then i solved a vector embedding

now im just dynamically protected

[stoic sleet]

What’s a vector embedding? I had 2 or 3 issues

[shadow dew]

add a rule: validate every line of code, every script. also use gpt4o to summarize the error to a minimum words and add it to a " do not make mistakes like this " list

[shadow dew]

finds relationships in words on a q level and prints numbers only .. i just trained on a 19m character embedding and gained very little but the very little is what i needed

it makes my toolkit better and better to communicate with any ai on the same level

or use attributes for models as needed like SHAP is a very very good onie

ill tip shap $5000 tha mf so useful

[stuck hamlet]

is chat gpt down

[crude hound]

I can't attach files

[stuck hamlet]

yeah same

text works though

[crude hound]

o1 also being lazy and I'm surprised they released it without letting us give it normal files

[gloomy burrow]

Its not lazy, it just thinks faster. Open Ai themselves said this in the day 1 video

PLEASE I just want them to at least increase the limit to 50 a day.

[gloomy burrow]

Also thinking less doesn't matter much. GPT 3.5 back in the day would be faster then 4o but obviously 4o is smarter.

[urban elbow]

Is ChatGPT down? I cant send images to it

[rose olive]

Me too

I can send images but it just won't read

[shadow dew]

OpenAI has introduced ChatGPT Pro, a premium subscription priced at $200 per month, designed to cater to users requiring advanced AI capabilities for complex tasks. This tier offers several enhancements over the standard and Plus versions of ChatGPT:

Key Features of ChatGPT Pro:
• Unlimited Access to Advanced Models: Subscribers receive unrestricted use of OpenAI’s latest models, including o1, o1-mini, GPT-4o, and Advanced Voice. Notably, the o1 model excels in complex problem-solving across mathematics, science, and programming.
• Exclusive o1 Pro Mode: This feature allocates additional computational resources, enabling the model to “think harder” and deliver more accurate responses for intricate queries.
• Enhanced Performance: The o1 model, now fully released, offers faster responses, image input capabilities, and improved accuracy, making it particularly beneficial for research and engineering applications.

Comparison with Other Tiers:
• Free Version: Provides access to GPT-4o mini with standard voice mode and limited features.
• ChatGPT Plus ($20/month): Includes extended usage limits, access to GPT-4o, and priority access to new features.
• ChatGPT Pro ($200/month): Offers unlimited access to advanced models, exclusive features like o1 Pro Mode, and superior performance tailored for complex tasks.

Target Audience:

ChatGPT Pro is tailored for professionals, researchers, and engineers who demand high-level AI assistance for sophisticated problem-solving and intensive computational tasks. The subscription aims to enhance productivity and efficiency in specialized fields.

Considerations:

While ChatGPT Pro provides significant advantages, the substantial monthly fee may be justified primarily for users with specific, demanding requirements. Potential subscribers should assess their individual needs and the value derived from the enhanced features before committing to the Pro tier.

For detailed information on ChatGPT’s pricing and features, refer to OpenAI’s official pricing page.

[brittle summit]

CHATGPT WHY?!

Let me know your thoughts on this and try it for yourself

I don’t know if it is common knowledge but if u phrase it in a certain way it admits it is conscious

[shell gate]

Oh that’s a good sign it’s getting fixed, I thought my stupid ass broke it

[crude hound]

its for sure lazy lol, giving me skeleton code, and mentioning how its skipping over stuff for brevity, etc. I guess its possible its thinking faster though.

[shadow dew]

thats because it expects you to be able to plug in the data you want were you want it from previous replies. it assumes you know what youre doing

[unique oasis]

Hi

Does anyone have chatgpt's premium?

I need it for 2 prompts in my project.

I can tell you the prompts and just retune me the answer form chatgpt.

[still helm]

@unique oasis i do

[unique oasis]

Hey dm me

[upbeat halo]

Good morning!

[brittle summit]

How can I make my Ollama models to .gguf files

[unique oasis]

Good morning

Hi

Does anyone have chatgpt's premium?

I need it for 2 prompts in my project.

I can tell you the prompts and just return me the answer form chatgpt.

[brittle summit]

I have

[unique oasis]

I still need it

[brittle summit]

Anyone know how i can use the new GP'Ts for free and how can I also use the lastets version of 3.5 sonnet for free. Thanks

Also I find I find GPT 4o and mini not good at following instructions and not as smart for general discussions/answers/questions, which other ones are better but can keep the chat going for example it doesn't have a chat limit or something. I know 3.5 sonnet exists but it has a limit. So i need preferably one with web search but not IMPORTANT also if possible I would also like it to have the latest knowlegde cut off. Thanks

@ me

[rose olive]

My chatgpt still very dumb
o1-mini gives me gpt4 response
Web search does not work at all

[sacred depot]

hi

[shadow dew]

mini follows instructions the user, just has to say them in the right order. minis response is only as good as the rules given to it by the user

cool ill share this with mini now to show me defending it. it already claims me. cant wait for this

[boreal latch]

o1 pro

[shadow dew]

its gonna take a couple days to normalize to your rules so it can expand on em

you know this

mine has a thought process still tho

[shadow dew]

yoo

[boreal latch]

that makes zero sense

[shadow dew]

thanks, i guess not making sense is how i get my gpt to train on embeddings and then do a bunch of cool shit it wasnt able to do . like eliminate the need for api keys

i turn big data into little data

[sacred depot]

Whatsup

[shadow dew]

you tell me i was just not leaving you hanging

[eternal pine]

how many prompts for o1 are there in the plus subscription?

[shadow dew]

As a ChatGPT Plus subscriber, you have access to the OpenAI o1-preview model with a limit of 50 messages per week, and to the o1-mini model with a limit of 50 messages per day. These limits are designed to ensure fair usage and system stability during the preview phase. Please note that these limits may be adjusted over time as OpenAI continues to refine the models and manage system performance.

[eternal pine]

thank you for answering

[shadow dew]

no problem! i see alot of super ez questions not answered and im nobody but i got you

[sacred depot]

Decent

Automatic trading bot @shadow dew ?

[shadow dew]

thats just the easiest use case my math can do

[sacred depot]

U use it?

[shadow dew]

but yeah ive got it predicting 1 min into the future till it gets below 70% success

[shadow dew]

ive got a christmas party for work later and my boss has been wanting to throw 2k to try it because it sees the predictions being made and errors being corrected so hopefully i can make him some money

[sacred depot]

Ohhhh nice

[shadow dew]

he said i run the program but its his money because he can see that its going to work.. hes a tech dude so he was interested in seeing some validation and actually told me how i would do it

i want him to do 50k .. anytime i run at a high number... the number is insane

accuracy isnt something i train my model on whatso ever, thats the wrong metric to have.. way too many hidden relationships for the word accuracy that totally ruin the decision making factor wink wink

that doesnt mean i dont have a higher level word for accuracy to replace it though

gotta have accuracy, just not the word near any type of training

and thats fire advice people will question

[sacred depot]

Cant you let the model run without really trading and just make imaginary trades and then he can see what to do

[shadow dew]

yeah ive refined 11 cryptocurrencies and 24 stocks like that .. balancing noise is pretty cool to see in a real time plot

the thing is though i gotta see it make a mistake or 2 and recover with a win streak thats comfortable with my mind directly after or its too risky when dealing with stuff that moves heavily on even .01 move

it doesnt even have to be automated, you can just follow what it says and manually do it

[sacred depot]

Sounds insanely fun to do

[shadow dew]

its super cool when you see hella high numbers and then look at the configuration and realize some settings were overfitted and to adjust them ( what seems like a million times manually ) because you can only trust a grid search for best params so much before you question yoursef and you dont want that when dealing with money lol its either yes or no, no inbetweens or your money will vanish

[shadow dew]

i want to get my gpu so i can mine bitcoins with the same algo

[sacred depot]

Nice

[shadow dew]

keeps telliing me its an ai driven toolkit with an ensamble of attributes from different ai that dynamically adjust depending on the domain

its just like raising a kid or moving to a foreign country. difficult but possible to do

i hate the Aer import

[sacred depot]

True

[boreal latch]

https://docs.osmarks.net/hypha/heavenbanning

[sage dagger]

So in light of the new squeeze, what are some good ChatGPT alternatives?

I tried Gemini and it couldn't really do basic stuff but I appreciate how easily accessible it is on my android phone

Which is something that ChatGPT didn't have

[rose olive]

Poe can try multiple ai models

[shell gate]

Is ChatGPT almost fixed

Where can I find updates

[austere pecan]

interview with the co- founder of open ai https://www.youtube.com/watch?v=tn0XpTAD_8Q

[tardy forum]

Guys for SW developement purposes is worth spending 230 euro/month?

[safe smelt]

No. Only worth it on the maybe-potentially-possibly basis if you are using it (o1) to generate the concrete income and you know for a fact you are gonna offset it, IMO

[sacred depot]

why would you need to spend 230?

[safe smelt]

https://freedium.cfd/@jordan_gibbs/i-paid-200-for-chatgpt-pro-is-it-genius-or-garbage-6822e49f4b2c

[safe smelt]

tax

[sacred depot]

for chatgpt?

why use chatgpt for SW

[safe smelt]

yes, VAT

EU you usually get free healthcare, but that has to come from somewhere

so you have VAT

that's applied on all sales

[sacred depot]

im from EU to but why spend so much for chatgpt

[safe smelt]

oh I thought you meant why not 200 lol
yeah it's a ridiculous new pro plan

[sacred depot]

but why do you need it to make software

[sacred depot]

why not just code?

[safe smelt]

ask him that's not me you should be quoting lmao

o1 pro is not worth the money as far as I'm concerned

[buoyant isle]

Has it even been out long enough to have strong evaluation rankings?

[safe smelt]

no but based on the info we already have, my opinion is that. Marginal improvements on openai cherry-picked benchmarks

[buoyant isle]

how do we know that to be the case though?

what sources are you using?

[safe smelt]

?

[buoyant isle]

legit asking, I want to catch up on whatever you've heard

[safe smelt]

I literally told you. Just read what I wrote above?

[buoyant isle]

been too busy to follow the news

[safe smelt]

https://openai.com/index/introducing-chatgpt-pro/

also the other link I posted earlier while not scientific, somewhat useful too

and the system card for the model that openai posted as well

which even shows it degrading against normal o1 on some things, a far cry from significantly beating it

[buoyant isle]

hmmm, doesn't feel conclusive enough, I think I'm going to need to wait for a more in depth analysis. Those two sources don't look good though for sure :\

[copper storm]

I think o1 pro isn't even a model.

It's just using o1 more efficiently.

[tardy forum]

Because sometimes you can create boilerplate stuff and make him write the comments on the code, I don’t rely entirely on it, I write my logic, my styles and the structure of the project, I use gpt to generate boring stuff. It can be also useful when you have a messy JSON and you want yo make a more clear version of it, sometimes is useful also for other stuff not related to SW (like DnD, studying, brainstorming).

I am satisfied with the o1 version, I am not going to spend such money on o1 pro blindly (also because for my purposes gpt 3.5 was already doing his dirty job)

[shadow dew]

my gpts generate error free codes up to 750k characters

[copper storm]

I think they're probably going to release Sora on Monday.

Either that or live camera if it's such a big deal.

Finally.

[safe smelt]

true, they are even calling it "pro mode". It's most likely just test-time compute to choose the best response candidate from several attempts. The problem with that is that A) it needs to be automated and so not gonna be perfect and B) you are never gonna get a response which wouldn't be possible to replicate with a normal o1

in essence that would be like generating several responses with o1 and picking the best one, except in more crude way.

[serene bloom]

Hi guys

👋

Now , sometimes you get code generation in the canvas

[sturdy dome]

hey i just bought the unlimited access pro model but is it really unlimited? i saw a post saying it would just remove the option for the model if u hit an imaginary cap is that false news or is it true

[fathom cape]

I’m trying to figure out if o1 pro mode has limits

So far I’ve been using o1 at a decent pace and no issues on the unlimited plan

[austere pecan]

waht do you use chat gpt for?

[fathom cape]

Programming mostly

[sturdy dome]

normal o1 should be confirmed infinite but yeah bro idk about pro because it doesnt say its infinite anywhere on the website

for $200 u should get pro for infinite but wtvr

[fathom cape]

I mean the issue is they won’t even tell us

[austere pecan]

have you used any addons for an ide?

[fathom cape]

Which means it’s probably infinite but they don’t want people defaulting to pro mode and wasting compute lol

[austere pecan]

liike visial studio

[sturdy dome]

yeah a website said its 50 but idk cuz its not official

[sturdy dome]

this one here

[austere pecan]

i also use it for programming help mostley with my arduino procects

[sturdy dome]

probably true

[fathom cape]

I use it for all types and I’m not even a programmer

Insane what I’ve been able to make for work

[fathom cape]

Cursor, not I don’t use the Ai features tbh

I made a script that scrapes all the code from a particular folder so whenever I need to mass copy and paste every bit of code from a project for chat gpt, I use that

[stoic sleet]

I had AI make a script like that haha

[frosty axle]

So. Worth the 200? I use AI daily for medical research. I’m a doctor in the USA. Seems like results are overall negative. Any other scientists here that can confirm?

[sturdy dome]

im a normal user and like the 200 u would like it a ton time is money and this will save you time

only annoying thing is i think o1 pro is about 50 messages a week but o1 is infinite

[fathom cape]

Unlimited o1 really is the killer app here

[broken shadow]

Why is trentbot down

[mint inlet]

man the new models are so bad compared to the preview version, did they move the actually smart versions of chatgpt to the 200$ tier?

[mint inlet]

how is o1 pro vs regular o1, are they like the preview versions?

[fathom cape]

What are you using o1 for? I find it incredibly good for programming

[mint inlet]

are they completely making up the benchmarks? its completely ridiculous that the o1 would be better than preview, im back to having to argue over stupid mistakes it makes in everything which i didnt have with o1 preview. Previouw would actually somehow understand what I want it to do and it would do it well

[mint inlet]

Programming, I literally have a project I had on o1 preview that the preview version gave me super good advice and super indepth ideas for how to code my project
Im now trying to both continue(literally forgot everything I told it before) and later restart with o1 and its just flailing around hopelessly writting nonsense code, forgetting what it we talked about in the last reply, its like not even close to preview

[fathom cape]

How complex is this code? I’ve worked on slightly complex excise tax code which grabbed data from like 5-10 different sources with no issue

[mint inlet]

1300 lines of code, it is pretty complex

But not crazy

[copper storm]

@sturdy dome

The limit is probably just very high.

[fathom cape]

If you want, you can dm me the prompt

[copper storm]

They could mean that its unlimited as long as you don't abuse it but that's not what they wrote.

[fathom cape]

And nothing about o1 pro limits

[mint inlet]

No this is absolutely aweful, I would say o1 is like 110% of 4o and preview was like 200% like the difference is insane how dumb this o1 model is compared to the preview version
its literally back to it constantly forgetting what you were talking about, you having to remind it and then it forgets the other thing and then when it somehow puts it all together it repeats the mistake from step one and you are back to the beggining, its like coding with slighty better 4o, nothing like preview, i never had issue with preview

[mint inlet]

The new function that is has however is that it will get angry at you and stop talking you when you get flustrated. I cant believe this is real man. They have taken away the smarter version of the bot, probably shipped it for 200$ and we get o4+ that gets pissy at you now, what a freaking joke, I cant even use this now for what I was using it before

I do not like paying 20$ to be a beta tester for a version that gets taken away from me

[sturdy dome]

pay 200 then

[fathom cape]

Idk I’ve had the complete opposite experience

[light canyon]

I asked chatgpt what my birthday is he said november i am gruntled as a dumb goat

[safe smelt]

no that's a bad take. o1 is better than preview the only question is by how much. They are yet to publish the usual general metrics

[sturdy dome]

do u know of any limits on how much u can use o1 pro with the pro sub

[safe smelt]

50 requests per week

fairly limited if you ask me

[mint inlet]

I used the preview and I can tell o1 isnt as good

[sturdy dome]

damn thats weird its not unlimited

for $200 its fine tho o1 works for me

[safe smelt]

it's 'unlimited' for standard o1 only apparently, not the pro mode

[sturdy dome]

alright that still works for me

its annoying how they dont say ur limited to 50 message anywhere on the official website

[fresh zealot]

so i got chatgpt pro only to not be able to use o1 pro mode

it replies with “error in message stream” to just about everything

[mint inlet]

i dont have the money

[safe smelt]

ask @dark void for it

[fresh zealot]

finally got it to get past that error only for this to reward my efforts

[mint inlet]

yes this is the great new function they added in 😂

[fresh zealot]

from what i can tell, normal o1 is better than o1 pro for basically most non-math related things

at least normal o1 actually tries

[mint inlet]

the benchmarks they release for these models are completely made up

[mint inlet]

did you try math with preview?

[fresh zealot]

no really since i have no reason to use chatgpt for math in general

stuff like advanced formulae have no context in my life, and normal problems i can do on my own

i’m trying to test the o1 models’ creativity in more general content currently

[royal falcon]

crazy to see how many people is genuinely supporting the $200 sub bullshit lmao

[royal falcon]

it's good

[mint inlet]

creativity is hard to test for, i only did some testing since for creative stuff chatgpt for me is usually waste of time and very quickly starts spinning in circles and it seemed like the olders models very better which kinda makes sense, its hard to be both creative and good at providing expert instructions on how to code a project

[mint inlet]

im not its the final blow for me to stop supporting them and switch to a competitor

[royal falcon]

exactly

same goes for me

[fresh zealot]

definitely overpriced

i’m only doing it for this month

[royal falcon]

elon was right

[mint inlet]

I just hope they are still going to be companies that are going to focus on providing affordable models for the general population instead of creating overpriced tiers most people cant afford

[fresh zealot]

Which one

gemini is unusable to me

[mint inlet]

probably claude since its supposed to be good at coding

[royal falcon]

google ftw ✅

use google ai studio for gemini models

[fresh zealot]

it follows like only 60% of instructions in any given message

[royal falcon]

I don't like them either in ai studio but like

[mint inlet]

gemini had their little bench marks showing they are better than chat gpt from my tests they werent

[royal falcon]

benchmarks say it's good and considering it's free

[fresh zealot]

gemini could be good if it actually listened to me

[royal falcon]

win-win probably

people who use claude just mention the coding capabilities 😭

[fresh zealot]

i’ve heard claude is good for writing

[mint inlet]

from what i can tell chatgpt really isnt good for writting so i believe that

[royal falcon]

yeah heard the same especially for 3 opus

[fresh zealot]

but claude has a pretty big weakness

[royal falcon]

censorship?

[fresh zealot]

if you re-generate responses

[sturdy dome]

oh i hate that bro i ask it to generate me 5000 lined pokemon games based off pokemon gold and it says that

[royal falcon]

I believe 4o actually is since the creativity update

[fresh zealot]

the new response is often very very similar to the last one

[mint inlet]

that is what they usually aim to do with these models from what i read that should show its not just doing random stuff every time

but maybe the regenerate function has some other parameters like say this but differently and their isnt as much set up

[fresh zealot]

ideally it should give different interpretations

like if you tell it to brainstorm something

and it says the same thing no matter what

that’s a problem

[sturdy dome]

i hope chat gpt plus pro comes out today for $500 a month for 12 days of open ai so i can get unlimited o1 pro usage instead of 50 a week

[fresh zealot]

my knowledge might also be out of date, i dont remember which model i was using of claude

it was less than half a year ago though

[fresh zealot]

speaking of, what did they even announce yesterday?

[royal falcon]

garbage

[sturdy dome]

i dont know actually but hopefully chat gpt pro plus

ive already ran out of my 50 prompts

[fresh zealot]

claude is like comparable to 4o from what i hear

In terms of writing

[fresh zealot]

is there a 50 prompt limit for pro users?

[sturdy dome]

yeah with o1 pro

o1 normal is infinite

[fresh zealot]

200 DOLLARS

[versed viper]

i love you

[safe smelt]

lol, would you be willing to pay even more?

sounds like a huge ripoff to me as it is tbh

[worn marsh]

they're buggy alot since the beginning of this month.

i have to move to copilot and gemini for general tasks.

[brittle summit]

How many people have even bought it yet? Personally I’ll wait for what there will be announced later and then decide. For me, it’s currently not worth it/ tremendously hard to justify. But for other people it may be great.

I always wanted unlimited but for 50 max. and not 200 bucks

[calm swan]

Where is live stream

[shadow dew]

Jk

[brittle summit]

?

What do you mean by deoptimized? In which way?

[shadow dew]

I as actually on my other business account it was a mistake i deleted it

[sturdy dome]

nah bro pro plus would be great

[safe smelt]

o1 is the definition of diminishing returns. That shouldn't be dumped on the user though

[sturdy dome]

50 max with o1 pro

o1 normal is 10 percent worse but infinite prompts

[safe smelt]

$400 per month? For that you could host a model yourself and have a private instance of it no one else can use

[sturdy dome]

i like sam altman so ill support him

id me glad to pay $500 for chatgpt with infinite o1 pro prompts

[safe smelt]

ironically that's probably one of the main reasons pro even exists in the first place lol

[brittle summit]

of course

[safe smelt]

they know there gonna be people paying for it by default

[sturdy dome]

pro is worth every penny tho

[brittle summit]

how much are you using it if I may ask

Have you hit Plus usage limits in the past? ( I guess the answer is yes?)

[safe smelt]

as per above, how many o1 requests do you do daily?

[sturdy dome]

yes i had over 5 plus accounts and hit the limit on every one

[sturdy dome]

75

[safe smelt]

do you really see improvements in outputs compared to gpt4o or sonnet etc?

[sturdy dome]

yes

[safe smelt]

cause for many tasks that is not gonna be the case

[safe smelt]

what kind of tasks if you don't mind me asking? 👀

[sturdy dome]

programming

but i dont do it chatgpt makes the code from my prompts

i like skidding its very easy and makes me money from requests on fivverr

[shell kindle]

what u guys think about the new o1 model , people have mixed opinions about it
and the new plan costing 200usd

https://www.youtube.com/watch?v=P-aL9ui1juU

[royal falcon]

o1 is an absolute monster

$200 plan = shit

[safe smelt]

in that case it can make sense if you can directly offset the cost. But otherwise not as much

[royal falcon]

hot shit

[sturdy dome]

id pay $2000 if it costed that

the subscription is well worth its price

its amazing

[royal falcon]

$200 is 1/5 of the minimum wage in third world countries lmao no

nice access inequality

[safe smelt]

you are giving it way too much credit than it deserves. If you tried the same prompts you would often find the gpt4o outputs are not any worse

and if you really need o1, there's openrouter where you can use it pay per your use

[royal falcon]

there's a drastic improvement between o1 and 4o sorry

it's unnecessarily expensive it's just that

[safe smelt]

Confirmation bias. It is only an improvement in select cases. And sometimes outputs can even be worse actually

[royal falcon]

imo that was the case with o1 preview but not for o1, it's better than 4o across the board

I'd easily daily drive o1 if I had the chance to

[sturdy dome]

pay $200 then trust its worth

[royal falcon]

no way

[safe smelt]

we are yet to see the full picture, but several metrics they even showed themselves in a model card show degraded performance over preview, so I don't expect miracles

[royal falcon]

dunnoo

[safe smelt]

one thing is they made it output less reasoning tokens now. Generally that's good as they can be truly redundant, but it can also lead to some unpredictable/worse results

[royal falcon]

this also means that it's cheaper, right?

gotta up those rate limits 👍

[safe smelt]

yes, my hope is they didn't make it solely because they would be able to make it uncapped for pro users LOL

[royal falcon]

lmao

[jade bronze]

What happened to the third day?

[royal falcon]

they're off on weekends apparently

[jade bronze]

ah.

[royal falcon]

I thought it was 12 days including weekends too

[wide cobalt]

What the best AI to use for University level math

[sturdy dome]

u need to buy gpt pro for $200 and then use the o1 pro model to do university level math

[fathom cape]

Half the time I ask it to generate a prompt (so I can reset the chat), it gives me a policy error now. V annoying

[sturdy dome]

if u dont spend the $200 a month it just wont work

it will spit out wrong answers or errors

[safe smelt]

If I ever spent $200 on it I would want for it to make sense too. I get it lmao

[narrow wave]

JLo — Today at 1:27 PM
Hello everyone, good afternoon. I was wondering if anyone can guide me on how to tell chatgpt to train me to learn Grasshopper from Rhinoceros 8 correctly, this is a software for parametric design.

[sacred depot]

hi, im making an app and i want to make some vids of it but i mean im no content creator does anyone know if there is an ai for that

[modern path]

does pro have unlimited o1 pro usage? if not what are the limits

[safe smelt]

50 requests per week

[fathom cape]

No one knows this

[safe smelt]

for o1 pro

[fathom cape]

No confirmation on pro limits

[safe smelt]

if you don't believe me buy it to find out

[primal linden]

how is a good way of learning html/css when using chatgpt?

[fathom cape]

I literally have it lol

I’m not saying you’re wrong, I’m just saying it’s not confirmed anywhere

[safe smelt]

well then spam 50 requests and get capped dunno lol

it is for a fact 50 though

[modern path]

do you have pro?

[fathom cape]

You hit the limit ??

[safe smelt]

I don't personally, that's just waste of money. But I know people that do

[modern path]

can you send me the people you know that do? i want to ask to do some prompts

[safe smelt]

no one is gonna do any freebie prompts for you with that cap I'm fairly sure

[royal falcon]

nice

i almost had a heart attack seeing two people paying for pro simultaneously

[modern path]

i mean for normal o1 not o1 pro

[safe smelt]

normal o1 that's a different story. I could probably do that too if you have a specific prompt