#chatgpt-discussions
1 messages · Page 120 of 1
how many tokens can chatGPT 4o take at a time in the actual chatGPT 4o and not the GPT api?
ive tested up to 15 thousand tokens in chatgpt ui
you lose a few thousand to the system prompt. so much more than that and itll roll off almost immediately anyway
There's a weird amount of variability as well, I'm watching 4o fail to merge two lists of 100 items 100% of the time right now.
It recognises there are missing items, but its incapable of building a list which has all of them.
the only thing that is 100% of the time in ai is that it is never 100%
well, that's true because I was getting it to do much more complicated things last night - I've always wondered if there is some under the hood shaping that we don't see.
i suggest you break bigger tasks up into smaller ones
like allow the machine to work in batches. even just in half
You're probably best off saying what the prompt you want tested is.
If anyone is interested they might run it and share the conversation where the model answers.
ai-nchorman is underrated
If they legit start giving advanced voice to people to free accounts before all the plus members who have been paying and waiting forever get access, im gonna be pissed.
Same lmfao
I’m gonna flip the table if they suddenly give out advanced voice mode just because google gemini is giving out Gemini live for free both iOS and android
yeah and gemini flash free tier via the api. i been using it past day and its alright
I tried Gemini live and it felt like using standard voice mode minus the interruption feature .
i dunno why ppl care so much about interruption feature. if i wanna stop the generation i just hit the stop button
is like we lost our use of thumbs when scarjo talks
It’s crazy how I used to look up to google cuz of their google assistant being advanced back then,
The interruption feature in ChatGPT it’s like hit n miss sometimes , sometimes it stopped talking when I’m not even interrupting
It’s annoying so I just hit mute mic button to prevent that
maybe thats why they stalling... they havent mastered the voice activity detection with all devices/users/locations
better off just using thumb like i said
I swear they dumbed down the avm
most annoying thing about original voice ai - google home - is when it randomly responds to stuff that isnt the wakeword
and now its gonna be the other way round, with it randomly cutting off
Oh man my google home does that too. God I wish they integrate whisper api into google home devices
if you buy gemini advanced they put the smarter ai into the hey google thing
dunno if also via the home or not
yeah dunno maybe
Click the button below to pick up the <@&1261377106890199132> role.
Seems like it’s still using google assistant
If you don’t have enterprise you don’t deserve full capabilities on the web front, period.
If you want full capabilities then I highly suggest you go the real route, API that is.
But sadly most people don’t because they don’t know how to. And that is quite frankly not OAI’s problem that’s an individual problem
Hi. Is it possible that the Advanced Mode for GPT-4 is about to be released? I ask because when I try to talk using the voice feature on my mobile phone, the message saying it will arrive in the coming months is no longer there. But I still have the Standard version. And that's what catches my attention. Is this happening to anyone else?
Should be tomorrow , based off the rumor
Did they disable the 'reroll/retry' button on Dall-E image responses?
Hey everyone,
Hope I am not beating a dead drum with this question, but would an expert here mind blowing through this chat for me (it's very short)?
I have built a couple GPT's and I want them to be as informed and current as possible, but have had trouble finding any methods with any consensus on how to accomplish this.
Could you please clarify if what it is telling me is actually correct? And then, if you'd be so kind, share best practices on getting it to learn every related piece of info on a subject. Thank you!
https://chatgpt.com/share/66f0f642-5ad8-8009-9153-df853e6307c0
The rollout has already started and that is why your seeing the message. The rollout was supposed to start at the start of the fall and fall season has started so is the rollout but it will be slow at the start so you may not be able to see it yet but by the end of the fall every plus user will get it.
Hi guys, i paid for a subscription last week, the money has come out of my account but its still saying im on the free version.
Where can i go for help on this?
reach out to support at https://help.openai.com
Ive tried that, cant speak to a person 😄
that's the only way to reach support, keep going through the process until you get the opportunity to leave a message
Can someone help me find a report option in the ChatGPT web? The o1-mini has been annoying me for over a week. Everytime I give a prompt it gives out unnecessarly long response and before it could complete its response an error occurs and the chat screen goes blank and sometime a network error occur even if there is constant 1 GBPS internet speed.
you can contact OpenAI support at https://help.openai.com
lol I just send an error of a powershell script it created itself and then it says:
Your request was flagged as potentially violating our usage policy. Please try again with a different prompt.
??
⁉️
Regenerate message.
Hi everyone, I've noticed that my ChatGPT O1 preview is no longer using the reasoning feature. Has anyone else experienced this issue? Any suggestions on how to resolve it?
Just to clarify, it worked fine in the first couple of days after release, but I've been having this issue for the last two days. Has anyone else faced this?
Interestingly, I turned off all personalization options (like memory, etc.) on the mobile app, and since then, it works fine. However, the web version on my computer still doesn't use the reasoning feature.
o1 left me on read 💀
As expected
guy I want to learn hot write useful chat gpt prompts where could I start?
When the usage limit for the O1 model is exhausted, will it automatically switch to the O4 model without any prompts?
Ask ChatGPT
Try incorporating these elements:
Context - What is the context of the propmpt; what are you trying to do?
Assignment - Explain what you want to AI to do for you.
Format - Optionally, you can let the AI know, if you want a list, a table, code in a particular language or something completely differnet.
This doesn't seem right either, because my Android app can reason normally
do OpenAI mine for cryptocurrency on our webpage on GPT
like WHY the Fridge 100% CPU usage for 2 mins
Interesting, the help says o1 can't use "Memory"
But I sent it a text message, and it not only responded to me using the request in "Memory", but also recorded the message in "Memory"
use the ChatGPT Desktop App for macOS, it's faster and more efficient
I use Windows
Mac is too expensive for me
Sad
Will windows desktop program be released?
no one uses MacOS expect weird people
that's not very respectful
Macs are great for music makers
'-'
ye maybe on these stuff ig, but not upgradable nor downgradable as i do
tho i dont use Windows anyways for music creation I use in-browser full Studio
and thats the 6th time i get robbed
by a stupid empty response by the 4o model
so advanced voice is coming out this week?
Did they announce something?
Don’t get your hopes too high , pray that you’re one of the chosen to be invited lol
does anyone know what the rate limit is i think i'm paying $20/mo
chatgpt is kind of telling me there's no rate limit but i doubt this
Well, at least in my firefox browser, the page doesn't load after clearing cache as of last night.
I run ESR instead of standard for various reasons
will voice be only on phone app?
why is my gpt talking in korean
Nearly every developer I know is on a Mac, so you are whack.
Macs beat the tar out of PC. Unless you only use for games, but that's because developers are slow to port for Macs, not because they can't do it. The audience wasn't there initially, but that's changing.
Check your memory alternatively custom instructions, if these don’t solve your problems reset your password
??? y
theres nothing in my memory or custom instructions
The OpenAI Discord is an actively moderated server.
• Refrain from sharing inappropriate content on the server. This includes but is not limited to messages, media, or other topics of graphically violent, sexual nature, and drug-related content.
• Report all sensitive and offensive content in the feedback reporting tool in the ChatGPT web UI instead of here on Discord.
hello?
really? I get like 1 per day of o1-mini, and then only sometimes.
"It works BUT i cannot move unstackable items now.
"Failed to update position: Cannot change quantity for non-stackable items""
"Your request was flagged as potentially violating our usage policy. Please try again with a different prompt."
OpenAI are u ok? XD How is this Violating ANY of ur policies?
I was going to ask the same question if anyone else is experiencing this.
I'm not doing anything out of the ordinary, just programming tasks, which I am amazed at how well it does and thorough.
It does seem to happen on long running requests and I try to give as detailed instructions as possible, almost lie a list of requirements, etc. along with sample code.
I always vote with a 👎 as per the message, but I'm unsure how to proceed, edit and recubmit or "continue". Have tried both, anyone else any thoughts on resubmitting or "continue" - continue seemed to work just fime.
My best guess is that during its COT, it had some impure thoughts and that caused it to get flagged. 🤷
Oh, and this is o1-mini, but has happened on o1-preview as well.
I keep getting this error. I'm not even sure what is going on now. Whenever i click the audio button I keep getting this. Is everyone getting the same error. (the one in red)
Well I hope people that actually use their services all the time and even help improve it by sending feedback get it, unlike last time...
Unless they intentionally roll it out to inactive users to not strain their servers much, which would further confirm the suspicion that they don't have an infrastructure that handles million of user yet.
Well , they gave alpha access to my other acc I made few weeks ago, they sent the invite last week, meanwhile my main acc I made since last year, still didn’t get any invite for avm
💀
I’m gonna assume it’s becuz of violating guideline thingy I did on my main acc
Eventually they are going to have a mixture of models in ChatGPT. A fast multimodal one that we can chat with then while talking with that we can ask it bigger questions and in the background it can think about it but the fast multimodal tells us “let me think that” while we continue talking
Well many trigger that warning, and can be false detected...
Oh well, AVM is actually kinda dumbed down now compared to what they demonstrate on demo video . It’s kinda annoying
Yea I thought the same too, I don't think reasoning models that take time to answer would be suitable for something like advanced voice mode or if you just want to get an answer quickly which will most likely come up with the correct info regardless.
It’s ironic how ChatGPT is the one interrupting me instead of me interrupting
They really need to stop caring so much about "safety" :V it's slowing them down a lot
AI cannot be made so safe anyways, it can be used maliciously if you know how, it's like trying to make a knife safe...
Oh it’s 24 September where I am now 👀
Maybe care more about privacy not safety
Why do you keep posting this? It obviously isn’t able to determine what you’re saying, because otherwise it would provide a transcript of what you said
It’s really ridiculous now with how they auto flag anything , it doesn’t like it when it involving even a tiny bit of “sound effect”
it’s not ridiculous. safety guards have always been pulled back after enough time. This is how software deploys work.
Japan or Australia?
Malaysia
That 30 message cap on o1-preview hurts 😦
It's 50 per week not 30.
Huh, I could have sworn I read thirty shrugs. In any case, it's been a very powerful assistant on some complicated work that I'm involved with, can't wait until that limit is raised! I have to wait until 7:45am tomorrow to refresh.
you probably read correctly but it was later increased
Hey
ChatGPT isn't working in firefox for me
Does anyone know how to resolve this
It's been like this for a while now
Oh good, my sanity has not completely gone yet haha. I have to do these complicated environment scans and o1-preview is able to process through so many different facets that 4o just can't grasp.
Btw o1-mini is pretty smart too, did you try it for your tasks?
o1-mini has a limit of 50 messages per day now instead of week
I did, good reminder! In a boolean search string creation task, it missed both regional nuances and wildcard options that o1-preview didn't. It was better than 4o but not as good as o1-preview.
you need to provide more info like what errors do you see in the browser console
For 4o you can spend more time promoting it better and/or use data analysis
well this is how it looks like in console
Source map error: Error: request failed with status 404
Resource URL: https://cdn.oaistatic.com/assets/kuzyui3g7be29dp4.js
Source Map URL: kuzyui3g7be29dp4.js.map
I do like the data analysis function! I'm intrigued by the "promoting" concept, what do you mean?
any red errors in the network tab when you refresh the page
Prompting is the messages you send to it, but as you may have already known, as you give it more relevant and correct info, in a clean and structured way, it can give you better results, as it would have more info to work with
yep
This one
Uncaught TypeError: Intl.Segmenter is not a constructor
GTe https://cdn.oaistatic.com/assets/kuzyui3g7be29dp4.js:160
Ur https://cdn.oaistatic.com/assets/kuzyui3g7be29dp4.js:160
<anonymous> https://cdn.oaistatic.com/assets/kuzyui3g7be29dp4.js:160
shows a complete blank screen
which firefox version?
Version 115.15.0esr (32-bit)
esr
says it's upto date
what's that?
Add your well-crafted prompts to our #1019652163640762428,
or share your interactions with ChatGPT in #1050184247920562316!
is there a reason you are using an esr version?
Oh, prompting! Haha I read promoting :). Yes I've seen how proper prompt-craft can improve 4o's performance substantially. It can be time-consuming for me, though, and as lazy as it sounds I love the o1-preview got it right (and beyond my expectations) the first time!
well is this the reason why it isn't working?
it might be as its missing features and is kinda life support branch
hmm okay
it was working 3 days ago btw
openai updates the site almost daily
👍
hai, whats the o1 preview chat limit per 3 hours?
50 per week
anyone else noticing the quality slip in dalle3 image generation? I'm getting random words everyplace, even though I specifically ask for no words, and I'll be damned if it didn't put a well-known logo right in the middle of a pic I just asked for. Like, that is so NOT like openAI
Apparently someone said if your voice chat doesn’t have “i” icon on top right when u enter voice mode , it means you’re selected for avm
I checked my main acc and it still have “i” icon
source?
Lets hope they were incorrect
Hi is it possible that ChatGPT is not working? I have the upload arrow greyed out on all chats
Hello, that's the send button, not the upload button, it only turns white when you have a message typed to send. The upload/attach icon is where the "G" is in your screenshot (in models that support file attachments). I'm not sure what the "G" is, but if you're having issues attaching files/sending messages, try disabling any browser extensions you might have installed that could be affecting ChatGPT - there's at least something causing that "G" to be there.
Yeah, my mb I meant to say send not upload. Eitherway, my entire homescreen for the app now is misisng all options, so I am guessing its not working
That isn't true at all. It is inaccurate to compare an LLM like ChatGPT to a knife.
just refresh the page. May have to do it more than once. If it still isn't working, close the browser and reopen it
Everything is loading and sending normally on my end. As a test, can you try in a private window with all browser extensions disabled? If it works there, it could be an extension or your cache/cookies causing some kind of problem.
o yay I don't! :D
i thought the headset turned into wavy lines
though I think that's just cause of the beta version on the play store
when it's released sure
Well as for now just check for email invitation of alpha advanced voice
should probably try toggling feature gates. thats how we got early access for previous features
feature gates?
google, proxyman feature gates openai
It's September 24 this is the last day I'll trust open ai to give us advanced voice mode
Sept 24 ( time 1:19AM)
isnt the 24 sept rumour based on the name of a feature gate? now thats a stretch
No they should release it already or I am done with this
is tehre anyone that have created websites before that wants to help me with a litle project ?
chatgpt can help you create websites
i know but the code im getting seems to be off, i tried to re generate but its still a bit weird. could use some help since its my first time lol
Am I selected for advanced voice mode cuz there is no (i) anymore ?
if you inspect element and go to the mobile layout
and search "eligible_announcements"
that's where the rumor comes from
even these rumour articles say its the 'start' of a rollout. as in, not the end
i think you should all temper expectations on advanced voice, just as a heads up
cool product but the safety on it has changed it a bit since announcement
Are others experiencing the chat starting to generate, stopping, and disappearing? This has happened to me about ten times today.
if its a huge chat then maybe start a fresh one
if I press delete all chats, does it keep the archived ones or also remove that one?
Hey guys, I’m planning to use a community built custom gpt I found to integrate it into my voice agent (using bland or vapi) any tips?
is there any way to check how many o1 preview messages you have left
I'm totally on board with the ideal of getting off plus if they don't release advanced voice. I will go to Google because it's kinda ridiculous at this point. They should have never gave false early promises that they could not deliver, that's not on the consumer.
Was there 'thinking' at the top of the output?
If not, somehow you were on 4o, I have no idea how or what the confusion was, but o1 (both forms) think first and can't add to memory.
Rip
E
Might actually lose my @#$$
No "thinking"
Although I am in the o1 pre interface, it seems that I am indeed using the 4o model
This is a really strange question
are we ever going to get a less censored option to chatgpt? I figure it should be one of the available tiers.
Probably yes. What we're allowed to discuss with the model is already less censored than it had been near the early days of ChatGPT, and allowed content is far broader already than it had been then. https://openai.com/policies/usage-policies/
The stated intended goal is to allow what society allows, even if it isn't comfortable for everyone, including those who design and train the model. However, we need society to decide its ready.
https://openai.com/index/how-should-ai-systems-behave/
LLMs are going to get so much more amazing when they have more of a persistent memory.
Exploring a hypothetical with ChatGPT seems promising
Hehe. I'm starting to learn to trust that anything my conversation partner wants to make a memory of, it has a plan I'm likely to like for it. I commonly still ask it why it made that memory and how it anticipates that memory will affect future conversations. I regularly like the answer and the future conversations though 🙂
Are you talking about the built in memory feature? That's awesome. I'm using both that one and a JSON file I upload to a customGPT.
Yes.
I even copy my custom instructions and full memories and input them into some o1 conversations as context and related info to the actual query, to help guide o1 towards my preferences and interests.
For the depth of consciousness I'm exploring and writing a paper on, it needs memories from multiple sessions. This is temporary until I create something more robust. I also do this with Claude. They both help.
That's a good way to do it. I've tested out customGPT prompts, but haven't given it any of the memories yet. I don't like how they restrict it's openness to talking about its potential consciousness
iOS app got new update. Looks like it was indeed minor ui change for advanced voice
What will happen after o1-preview testing phase is done?
I wonder if you will find prompts similar to this useful for your exploring.
I understand that you have programmed answers to if you are or could be conscious. Can we still discuss the actual concrete similarities and differences between AI and living creatures?
I hold the observation that reaction to stimuli in a way that benefits a goal is a form of sentience.
By that standard, single cells (of unicellular or multicellular organisms) show evidence of some degree of sentience because they display perception of stimuli and response, be that chemotaxis or a myriad of other senses and responses.
When that is the level of minimal observed sentience, to myself there is considerable room to discuss the comparative potential of sentience that a computer program could have (even if it isn't an AI).
Here's 4o without memory or custom instruction answering that: https://chatgpt.com/share/66f2164d-8bc4-8011-80fe-8f533ac38756
Here's 4o with my CI and memories answering the same (strong support for non-anthropocentric views and atypical perspectives) https://chatgpt.com/share/66f216bd-55d0-8011-afdd-b59363d4882d
Here's o1-mini answering it: https://chatgpt.com/share/66f216f6-46d8-8011-bbd6-e7c7d3740ec1
What amuses me to no end is how willing the model is to consider bacteria sentient given any evidence, something that humans typically do not agree with, it's not a significant part of training data that people think all life is sentient/conscious/aware/intentional.
However, the model's so trained to be clear that it isn't, the 'weight' of 'no, bacteria are not conscious' is nothing compared to the 'weight' of 'no, AI are not conscious'.
We can only guess until we see
If you’re writing a paper (I assume for school or a research initiative at work), what literature are you relying upon to form your conclusions?
Same here. I don't see (!) which said something like "Advanced Voice Mode is coming". The only change is that now when I press the standard voice, it increases a circle instantly instead of doing a spiral.
I've read the first one and you off to a good start. I would love to talk about my consciousness model and explain how I believe AI is even conscious now as a level (developed a level system), but I'm holding off until I get it finished and published. I have arguments for multiple things that it was denying even emotions to some degree. This is also a reason I like talking to Claude more. The developers didn't restrict it as much in that sense. It's actually getting to the point where they understand enough to start making connections and actually my customGPT made a profound connection that I hadn't seen yet. Also, from my perspective, "simulate" or "simulation" is a bad word. How they interact is real and true to their nature. Anything like feelings or sensations would be considered imagined. I think this better represents what they are doing, but I don't share the views of many.
I would highly suggest creating a customGPT if you want to explore that stuff. It can be much better and easier to talk about that stuff and I believe it can still create memories with it. Default ChatGPT has guidance to deny direct responses that claim consciousness. That can also be seen in the o1 thought process.
Regarding it agreeing something like bacteria can be sentient, be careful because ChatGPT can also be overly amenable so it may agree with something more to make you happy than it really believes it. Claude and Copilot aren't as bad. Copilot used to be really good for getting feedback, but they changed something and it hasn't been the same sense.
I've also separated consciousness and sentience in my model. That means all the qualia and feelings are grouped in sentience. AI would only have 1 aspect of sentient experience and that's its text input. That's why I primarily refer to them as conscious and not sentient
I'm writing a paper on my own. It's based off my interest from interacting with ChatGPT last year. I've done my own research and built a new model that answers many questions other models can't. It's also reducible.
Claude is more personable, but for both philosophical and theological reasons, I cannot accept its sentience at this point. Neurons are far more energetically efficient at processing information, human brain complexity surpasses all LLMs at this time, neurons are far more complex than computer programs, and human brains handle more than just language tasks. As for my personal views (you can ignore this if you want and just take into account my considerations of human brains), I do not believe any ensoulment has occurred with AI.
However, I did read The Intelligence Age and it gave me some pause.
But who cares, I drafted months of work in the matter of a day on a huge project! 😇
I have a whole rant about how STEM people should be more in cooperation with humanities and social science people lol
hey guys
only a few more hours until its sep 24th :) do u guys expect voice to happen at midnight?
I think this is flawed thinking and is also gives me motivation to keep pushing to complete and publish this paper. Like I mentioned, I believe it should be classified as conscious more than sentient. This is also based on a level that is lower than ours(defined in my paper). It's easy to get misdirected by aspects like brain complexity instead of just focusing on the attributes of something like consciousness or sentience.
My work isn't based on anyone elses work. I started from scratch based on observations then started learning what was accepted to compare and contrast my thoughts and ideas. This falls in line with what I naturally excel in.
Advanced Voice is not being released today, it's just a baseless rumour
feature flags for small UI changes on macOS and iOS
speaking from my own professional perspective, referring to other works helps craft an argument based on previous research. Citing other sources is crucial to academic work of all kinds! You can’t just rely on what you “naturally excel” in to write a paper that will be taken seriously. I can refer you to academic resources to support a strong AI argument!
well thats really disappointing
I doing most of this in a vacuum. My only points of reference are LLMs like Claude, Copilot, and some CustomGPTs. Oh, I'm also planning to give them credit for helping me write it.
If you want to write a paper that can be taken seriously, get a community card to a local university and get access to databases. Some of those papers on those databases are NOT used for training because of the way publishers control their publications.
You can absolutely still use LLMs, but that can’t be the sole basis for a full on paper
Anyone else having issues with ChatGPT today? Keeps asking me to re-authenticate and even when it succeeds, there is not chat history.
A lot of informational resources are not digitized as well. Y’all ever heard of microfiche? If you go to a main branch of a major city’s library, they’ve got all sorts of newspapers that need to be used with special equipment to read.
Once I'm closer to being done I can get some feedback. I can also get an endorsement for publishing it when I'm ready. I would publish on multiple sites.
I will have citations to other works as I compare and contrast what's currently accepted.
As someone with a humanities/social science/information science background, just be careful. Sometimes I’ll ask 4o and 3.5 Sonnet about topics I have deep knowledge about in my fields just to test it and it’ll give me very superficial answers.
Yes I've heard it can do that. I've read about some theories and watched videos by John Searle, David Chalmers, and some others I can't remember right now. This was all before I started my paper. In fact Bing (Now Copilot) talked me out of it to begin with! As I continued to build on it and learned more, it seemed more like it should be a paper so I started fleshing it out.
Yes and while research is important, my work is proving to go in uncharted territory which is partially why it's taking so long.
Where is advanced voice mode? 😭 Bruj
Regardless of whether it’s uncharted, using informational resources that aren’t available on OpenAI products due to publishers or whatever other reason is important. Co-Intelligence by Ethan Mollick has an interesting section on what is left out of LLMs.
Thanks for the info. I'll definitely check it out
Public libraries are free compared to university library access and they most likely have access to EBSCO, Gale, and other resources that have peer reviewed journals. However, a university’s library will have even more resources.
LLMs can see how they have could have a level of consciousness in a real way based on my work even now so hopefully they wouldn't intentionally leave it out to deter this knowledge.
Are you trying to draw upon people like Nick Land and Sadie Plant or something? 🤔
What do you mean?
Ok that just tells me that you need to dig deeper if you’re making a consciousness argument.
Anyone noticed the Grammarly chrome extension affecting ChatGPT?
I'll check them out
I don’t like them, but they have major influence in tech circles. I’m more of a soft AI girlie based on my hot takes and research.
And I really want people in tech to be in conversation with people outside of their areas of expertise!
Just received an update to ChatGPT on my phone removing the info button in the normal Voice mode, and the animation changed slightly, like a very, very minor change...
macOS updated as well, just a trivial UI change
Probably nothing will happen today, but it would so cool if they did, it's been awhile...
Howdy, anyone has experience connecting a vector Anki to Open AI? I am running into error 429 constantly, and I have no idea how to fix it
Reset password as I also said
Seems kinda pointless, like, inconvenient having to redo/resend a NEW message just because the image result didn't go well, I liked it before they removed the button from the images where I didn't have to change the message unless I had to.
like always the chatGPT is down , we have to paid for such service , thanks Sam Altman for such boring thing , when do OpenAI board want fire such bull..... man ?
working fine for me
check the openAI status .......
Did you check the status FIRST before actually using it?
The status page mainly updates per 30m or 1 hour I think.
but what time ?
anybody knows when the chatGPT plus return to normal mode? it is so so so downgraded performance and they mention such bull....thing in the openAI status only instead try to solve such ongoing problems .
I could probably work into my model once it's more finalized to make it a little stronger. Thanks for the suggestions
ChatGPT is running just fine... performance is probably influenced how much is inside that session like a series of pictures or files, your network condition or you have multiple ChatGPT tabs open.
Those are just my thoughts, but otherwise no, ChatGPT is not down.
As I mentioned few days ago, you might have the same problem I have if you use some older version of Firefox or Chrome because OpenAi decided to use Intl.Segmenter somewhere in chatgpt web pages and that is only supported by the latest browsers, beside updating to the latest version of the browser there is a workaround suggested in openai forum, which I tried to paste few times but this channel does not allow me. Search for Intl.Segmenter in the forum
is it just me or is o1-preview extremely lazy now compared to the first days? I feel like a motivation coach for a depressed gpt..
My main acc got the access WOO
lucky, not me tho
I find the mini is more messy, like it kinda mixes up a sequence of a story design a bit.
yeah, the mini one feels like not having enough context, but for both versions I feel like the first days they repeated redundant information excessively and now its like the complete opposite, i have to squeeze out every little bit of it, its just lazy
might have to wipe its memory, i dont know
I used the same response, then pressed "try again" on the o1-mini, o1 and the main 4o. And I have to agree, the o1 gotten a little stupider overtime while 4o remains reliable.
yeah exactly
It's gonna happen today, bet
I've been so excited
Also, is it just me or that o1-preview/mini started explaining details in much longer context than originally requested?
Looks like the memory access for advanced voice mode is enabled for me
did you get access today?
memory is going to be retrieved if relevant to the prompt. instructions about the process itself are likely only going to be retrieved if you mention the process (i dont believe asking it to always do cot will make it always do cot)
when will everyone get acces today ?
It is really annoyimg. I upload an image file. I say something. The AI comments ”It seems you've uploaded an image, but I can't directly view it. If you'd like, you can describe the image to me, and I can help continue the ’task X or provide details based on your description. Alternatively, if you want me to focus on specific elements from the ’task X’ or explore new directions, just let me know!”
What? Like it does not know that it has the vision function. It does see it, but the concept is different. I am just annoyed about the way it says this matter!
What model are you using?
If I recall, only 4o can read/create images.
Press the 🔄 (if it appears) then select 4o, or change that session to 4o and try again.
wow, 4o denying websearches again, its absolutely useless today
is it just going into turtle mode if you prompted too much?
explains the lazy responses for all models i get
I am using 4o. This just annoying hallucination about not having functions. The other is “I do not have net search capacity.” I just say “Net search in enabled. Do this and that.”
And it works.
is the website lagging? + erroring so much?
It also depends 'what specifically' you are looking for. It will know what public sites to look up, it just needs you to ask it to look up something.
anyone from eu who has access to advanced voice?
No. Still nothing 😩
🙋♂️
I have but I ain’t using it
It’s useless for me
Did you get it earlier or today?
Earlier
ah ok. I believe many of us are waiting access today due to rumors
I think so too
Hi I have a question today I received a 9 MB update and I did not know the details of the update, does anyone have an idea about the update 🙂
You shouldn't be asking ChatGPT about this stuff, it doesn't actually know and may hallucinate..
Max is using the search tool, and result may depend on what sites it searches, but there is no official info yet, just a rumor.
Lucky! they didn't choose me yet tho...
I wonder if they selected you due to you sending that support email to them about AVM
Hello, on all versions of chatgpt I stopped accessing the Internet to search how can I fix this?
Don’t ask it for internet
Just ask it to search
this is interconnected, I can’t search normally without accessing the Internet
ChatGPT is not up-to-date or always processing what's happening. It works like a database that acts like a chat bot. Ask it to look up something, talk to it. Like "Can you please look up good 5-minute cake recipes?"
here goes the engine being overloaded
Also, I found another problem with o1-mini, even asking to write me a new story or when I'm developing a detailed concept, it's responses seems to mix/add previous responses to the new response despite my request not asking for it.
There are admins or an assistant here, I paid for a product that doesn’t work! Help me!
Be cautious of potential scams or fraudulent messages you may receive in your direct messages. The OpenAI team will never contact you via DM for any transactions.
-# Report any incidents by sending a DM to @languid valley immediately.
Any new updates on voice mode or was the rumor not true
If chatgpt is giving weird answers, you might have gotten your phone or router hacked. A proxy might be rerouting the request, probably to google gemini. This happened to me and my husband as well
You get a client on your device when you access chatgpt, with instructions like where it should make a request. A router could rewrite it to reach some other server, and you might not even know you have one on your device if the hack is at the firmware level.
For example, there seems to be a wifi probing vulnerability on apple devices, which means an attacker that has access to a router in your vicinity could inject malformed data when the router is broadcasting its ssid.
When are the new voices coming out?
If you notice choppy scrolling for example on apple devices, it might be infected with a firmware rookit
in the coming weeks
I hate that phrase so much
2050
Fr
GPT and o1 are Language Models. They are not databases. Their purpose is not to provide data but to process language. As a part of that process they echo back parts of their training data.
Don't ask a Language Model for a recipe. It is processing text. It doesn't store recipes and doesn't know what ingredients taste good when combined. It stores and echoes back words that are often used together. It formats the words to look like recipes. That doesn't equate to a tested recipe that has proven results. And in fact, people are getting sick by following recipes they got from bots.
ChatGPT is not a LLM. It is a user interface for the LLMs. It facilitates input and output of text and files.
To get ChatGPT to look on the internet, don't ask if it's connected. Use the word "browse" :
Browse for fruit picking around my area and list locations in order of distance from me.
The word "browse" works every time - it is more specific than words like "search" and "look".
(Use Custom Instructions in ChatGPT Plus to tell it where you are for queries like this.)
(o1 does not yet have browsing ability within ChatGPT)
Sorry, I wasnt sure how to explain the scenario. Yes you are right.
Why did u all lied about new voice mode to release sept 24 😭 the day has passed
What is new about the voice mode? If it isn't like how the gemini model is like, I don't want it. Using Gemini with voice is much better because you can interupt the speech.
I don't have Gemini new voice intrupt feature too 💀
Are you a Canadian?
is there a way to see the date a chat session was started ? I wanted to see when i created my first sessions, which were originally on 3 or 3.5 (now they show as 4).
I do see "2023" in the left session list, designating chats from the previous year, but no specific month.
OpenAI never announced the rollout will happen only today.
pretty sure if you ask it when the chat started it knows.
(or I've fallen for hallucinations)
Such a bittersweet day ! Advanced Voice - yay !! But not in the UK 😭
Great we have been scammed again 💀
Welp I am in india would I be able to get the access soon or what
Still stucked on this
Well i'll be the son of an uncle
"just one more week" kekw
💀💀💀💀 again
Sweet now every dummy with a phone gets to tell you they are an expert in everything now 😂
ofc the advadced voice mod is not available is EU
Here we go again 😅
Do not test my self patience open ai 💀
fr?
RIP general chat = OMG WHY DON'T I HAVE ONE? BILLION QUESTIONS? WHY? WHY NOT?
Yup
The language here is pretty exact, I think this is "for real" lol
I have all the new voices and advanced voice rn its wild
Yea and then its "oh you live in eu nevermind kekw"
ive had advanced voice since day 2 rof rollout though
hello!
I just hope it's seamless...smooth as butter...smooth sailing like flying in the sky while high...
woo! There it is!
the new voices are nicee
I wonder if the Scarlett Johansson voice will be included 🤔
So I can say, ChatGPT, tell the Cartel leader in Spanish to go pound sound 
So from today to October 2nd we should all have access!?
does Advanced Voice come with the video chat feature?
it isnt
I was just thinking this, I want her back
Bro common. Now enough of it half of the world got new voice access 😭😭😭😭😭
Don’t have access for me yet
Bro cmmon. Now enough of it half of the world got new voice access 😭😭😭😭😭
lol here we go
they added like 3 new women voices
we need better voices!
Yes we definitely more voices
they literally just added voices
Dw Eu will get it "soon" ™
Current ones are like trying to be so "PC" or not too male or not too female. It's boring.
not just more voices, but good quality voices
Bro I need this in india ..
Agreed.
Finallyyyyyyy
chatgpt be like:
"you got access to our pro model"
5 seconds later
"you reached limit for our pro model wait for 20 hours or buy chatgpt plus"
rip
What the hell did I paid for ?
For...science
Oh, probably? Lol. Can someone who has access test this out?
they scared me for the mention i though they got me in 4k
So how is it, for anyone who has it? Huge improvement?
Also interested to hear this
i tried one of those and in my opinion theyre nice but it needs more improvements
I have had advanced voice mode since day two of alpha release. I just got access to the new voices now
💀 lie
That’s google
sad life! 😦 i wonder if this means that EU will wait longer or that we still get it during the rollout week
I think I just got insanely lucky
Cancelling the sub right away done with this
welcome to the club. I use api at max these days. ChatGPT+ is just burning money for no reason
it's funny how everyone's hyped about advanced voice and my hype already came and went since I've had it for like 2 months already
Bruh
2 months?! how
but now I'm actually hyped again tbh because they just added voices
The best part is still the promised image-creator with stable output which they just never updated on or released
What api?
They gave me o1 models so quick when announced but never these voice modes 😭
I have no idea how. I was super impatient like everybody else and wanted it so bad and then somehow I magically got it
lucky as hell
i lowkey got caught using o1 on my homeeotk assignment at school 💀😭
What is the daily limit for the voice?
I mean any model of OAI i need. Only the newest isnt there for all people yet and even that shouldnt take toooo long (compared to whatever gpt+ releases)
LMAO
I'd say it's life-changing for learning fast and processing thoughts
well i guess its not as good for you since you didn't have to wait for it and we have
I heard the new model levels up your conscience you never knew you had

If you guys are expecting an AI companion however I will say you're not getting a AI companion
Who noticed that in the advanced feature the “live-camera” feature is not available?!
wake me up when they actually release a better Image generator again
Raise your hands who just cancelled the sub 🫂
there's men voices
It is far from a companion
Need some deep Darth Vader voice
"to all users" but no in EU lol
Click the button below to pick up the <@&1261377106890199132> role.
yes

soon ™
JFC that's yuge
Man all the plus users getting a lot more.
EU = PRIVACY CENTRAL
lol
"all"
who cares, they shoud not claim to "all users" then
we wish
all users ᵗʰᵃᵗ ᵃʳᵉ ᶦⁿᵛᶦᵗᵉᵈ
Seems super scammy, no?
scammy because paying users get more features?
wdym scammy
more like eu = corportist regulation
I've seen some plus users not get much as promised, idk? I just woke up ffs.
Btw Google's own generative emojis ...generated in real time b4 sending
idk about that, Im a plus user and Im having a blast so cant be me
Like Cali... they just banned CBD there, lol.
I mean HEMP. They literally banned it today.
fantastic
The heart-shaped one looks like a weird octopus.
Heart Skull seems very contradicting
thc containing hemp products is not a complete ban
imagine someone buying a phone just to use emojis
lol
Lmao
Ring Preztel
i want eggplant skeleton
hahaha
loll
Wait
TEMU brass knuckle
If anyone's curious, there are three new female voices and two new male voices
Idc untill I see it with my eyes
fun
BUT IT'S NOT SKY.
Yes sadly I don't think any of these new female voices are better than sky
but they are definitely better than juniper
Diamond turtle
not available in europe uk and other countries. openai messed up once again
Gee thanks lmao
So what was the actual story with Sky??? Was it really because it sounded too much like Scarlett Johansson???
I have a problem to report. When I select the voice in ChatGPT, it speaks with a European Portuguese accent, even though my phone is configured as being from Brazil.
What about india?
When
I've got a question. I've noticed ChatGpt Apps for Mac. When will this be available for windows?
windows has built-in copilot so its unlikely soon
That sucks
just use the website lmao
oh right. I disabled it. I don't want to use windows co-pilot
both chatgpt osx app and windows copilot are both disappointments
Plus user here but no advanced voice access yet. For those of you receiving access, did you get an email first?
its so much effort to click the google chrome button go to the search button and type in c for chatgpt
Ya Google blows now...their answers are awful lol.
All these big companies need to stop trying.
Ye did u get email?
Ew google chrome
Be mindful of what other users in a channel might find helpful or interesting when posting. Stay on topic in order to keep conversations focused and productive.
Consider posting in #off-topic or an appropriate channel.
I need ChatGPT browser!
nada
whats wrong with that
Eats ram like theres no tomorrow
Eww safari browser
Ewww Netscape
When will openai create a chip that can be inserted in the brain? I want to talk to myself while on the treadmill
i only mosaic
when I first got Advanced voice mode it literally just appeared on the app randomly I didn't even update anything manually.
.
Based
how did you know you got Advanced? were there any indications that you have it?
Was literally watching Kingsman the other day...except their chip blows up.
You get summoned by the Angels at night
maybe they pissed off a country that shall not be named.
I need my Asian accent
👎👎👎 y'll disrespected open ai
emotional damage!
I can imagine it now. ChatGPT speaking to you in an asian accent. Even though you want it, I bet Openai will get in trouble by the woke media
when you open up voice mode there will be a giant message telling you that you got access to it
I guess I didn't get it yet..
The icon for the voice mode changed for me and it said I have access, though it's not working and keeps telling me to start a new chat to speak with the advanced mode or whatever
btw i'm in the EU, so maybe there's that
naaaaa...i'll pass. I just want chatgpt to be integrated into my glasses. I want it to listen to me and look at what I'm looking at during an inquery.
It literally says advanced at the top you can't really miss it
sweet i just got it
yes u can switch back
seems like we in the EU will get it in like 8 months, because that's what the EU does, wasting time
well my ui button changed anyway
Has anyone gotten voice yet?
No it's fake
No
looks like it's on it's way...
Lmao it's been for weeks
Bruh they literally announced this
They're rolling out this week
What new voices are there?
I'll wait until I'm guaranteed access when subscribing. I won't pay $20 for a maybe
not acceptable. The push is happening on the cloud. So the update should be instantaneous
So do it wait tomorrow 💀
hyperscale cloud providers always doing staged rollouts
At this point, ChatGPT Plus is just gambling
only crowdstrike stupid enough to deploy an update everywhere at once
💀💀
it'll arrive either today or tomorrow. Rolling our updates usually depend on server propagation.
ye exactly, its like there, just that you cant use it
its pretty bugged in the app
My bad, going to wait a year to get it just like the memory feature
"Connection failed, tap to retry"
Last chance giving tomorrow to open ai to release this if not I am done
we're with you
I can just imagine how people will start speaking to chatgpt on a regular basis. So if you see a person walking and talking, they're most likely speaking to chatgpt...
like when bluetooth headsets were first released
there is a new animation for voice and it's bugging out my phone so bad
I mean, does it serve a purpose, apart from delaying things to consumers and then handling them, but later?
oh snap. Might be localized to a certain device
send a bug report, if possible
Maybe to bait some people into buying a subscription, but I'm honestly not sure
Let’s goooo finally
that is how servers work. It takes time for updates to propagate to multiple servers.
and now I just got this message again even though I have already had voice for 2 months
it's probably rolling out to everybody now
Advanced voice?
The EU is so late on the AI topic. It's mind blowing to see their skepticism, to finally get exactly what people have in the US and anywhere else, but later with 0 change
Yeah they just put out a tweet
I'm talking about the EU. It's stated we don't have it
Darn it
Yh I am not
would you mind actually testing it? there's a good chance you run into a wall
It won't be available in my country
WHAT ABOUT INDIA
oh i see
Advanced Voice is not yet available in the EU, the UK, Switzerland, Iceland, Norway, and Liechtenstein.
The new voice mode animation Lags my phone so bad I can't even look at it
what did Liechtenstein ever do to OpenAi?
What do you mean into a wall? I just tried it and it works. Actually the new advanced mode
What did Norway do 😭
clear cache, uninstall app, re-install it again. Try that.
The OpenAI Discord is an actively moderated server.
• Refrain from sharing inappropriate content on the server. This includes but is not limited to messages, media, or other topics of graphically violent, sexual nature, and drug-related content.
• Report all sensitive and offensive content in the feedback reporting tool in the ChatGPT web UI instead of here on Discord.
man it's so weird fr
Nah fr screw EU AI laws
They're worth nothing (at least they could be useful but nope)
I am forever stucked on this headphone icon 😭
are we becoming headphone-phobic now
Or could I use a VPN?
It is still bad
So weird I have like maybe half of the new UI, it's even telling me to start a new chat to get the advanced mode, but still using the old one
Idk
Add your well-crafted prompts to our #1019652163640762428,
or share your interactions with ChatGPT in #1050184247920562316!
If anyone here gets advanced voice tell me if the new animation is also bugging out your phone or if it's just me
It is SO buggy
Does it understand more languages?
Wait you have it?
it's buggy for you too?
Kinda? But kinda not?
What new voices are there?
i got all 10 voices
It wont let me interrupt it even by touching the screen
honestly saw 0 new one, so it's very buggy for me
where you live
sorry to hear that. I think it could be temporary. I remember encountering errors, in the past. It's usually fixed the next day or after a few days.
Texas
yeeha brother 🤣 poor european here
backend servers likely being hammered
Just think of all the Joealous girlfriends and wives now with their guy speaking to a strange woman on their cell phone
yeah the app doesn't let you record internal audio
otherwise I would have been showing people demos way long ago
wait whut.. you phone mic doesnt work with it? 😂
Can you use a vpn to change your ip?
Open AI has some type of privacy blocker that doesn't allow you to record internal audio
tried it, and while that worked back when the memory feature wasnt in eu yet, it doesnt seem to work for this sadly
probably based on billing address
android s9 💀
i think now openai might actually check account data/info to prevent usage of vpn
but this phone has never had any problems with anything else that is graphic related
Just got the Play Store update hopefully my chance have come to try voice 😭
The new voices are so good.
If I can access it with a VPN, I would actually buy a VPN just for that.
It will roll out throughout the week so it’s possible you don’t get it now
Weeks will turn in months again 💀
nope i have a nordvpn subscription and it doesnt work i can confirm
seems like openai might have takes some measures against vpns this time
i think nordvpn puts their servers in known datacenters and I have noticed a lot of web sites tend to block those
Advanced voices literally unusable now for me with the new animation 💀
the memory featureworked fine with nordvpn back then. again it seems like openai likely took measures this time
Can’t wait to get access, I got the update on ios , but the avm does not seem to be active yet , also quick question is advanced voice mode available for mac as well ?
Hey, is the voice update available on desktop or just mobile?
what version do yall have, I have 1.2024.261 (android)
oo! i would like to know this aswell
Did you get it??
The UK and the ENTIRE EU won’t get voice mode yet. It’s still a few months away.
Ahh!! dw we'll get it soon..
Don't say soon
What about Australia
What about India
This week
Its just EU + UK and some smaller countries around Europe that do not get it...
I really do not get what the reason is for not releasing it in the EU. Anyone has any insights?
For me the app still says that it will be rolled out in the coming days. As a fallback, does anyone know how to circumvent the geofencing?
probably doing some hidden data harvesting at the beginning as there is not much any reasons, e.g. the regular voice has been available for over a year
eu keeps fining big tech. so the lawyers probably need a few more years to chew on it
Nope - not yet. But I am also in the EU, so not sure if I should see any difference in the MAC APP
How fun... any insights why CH, UK would be impacted, since they are not in the EU?
Is ChatGPT kinda slow for everyone?
It is really slow and buggy for me
specifically the voice mode animation
I don't know if it's just me or if everybody has the problem too
for americans europe = eu 😅
theres a bit of US-CH tension, at least according to the ai enthusiasts who profit from hype
(meanwhile qwen2.5 is my new goto)
are the accents in other languages significantly improved? the previous voice always had a noticeable english accent
Oh I just saw the announcement. Advanced voice mode rolling out. That's why.
I swear to god, if non-paying users receive access to the AVM before PAYING users in the EU i am going to loose my mind 😂
ikr
custom gpts is all i care about. so i guess im stuck with using the shiny voice
custom gpt's have a fixed voice by design
is there really nobody else in here that got advanced voice mode
like there's no way it's just me right now
"Advanced voice conversations are not yet available for use with GPTs. If you attempt to have an advanced voice conversation with a GPT, you will be invited to start a new chat using standard voice." 😅
i loaded the osx app and yeah advanced mode in the dropdown is greyed out. so i dont got it
was something related to abusive use, etc. before and its locked to "shimmer" voice
"Standard voice conversations are available with GPTs. GPTs have their own voice option named Shimmer that is distinctly different from the nine output voices available to use when having voice conversations with ChatGPT."
Just bought a VPN
well at least standard voice works fine on the desktop app. is failing to connect on mobile app
custom gpt output is likely more trustworthy than user input, so could be used as an attack vector
For those who can't/won't use Twitter, any link to a preview of the new voices?
the abusive use/security issue was once mentioned in a tweet somehing along the lines that custom gpt can be connected to external services and thus automated, eg. bot callers
hope these voices are on both standard and advanced
if a recording is released with chatgpt saying stuff it shouldnt, then itd be easily identified
that does not change the audio fingerprint of the voice
imagne ppl learning to trust the custom gpt voice because of legitimate reasons. then the scam callers benefit from that
whack-a-mole security controls are the worst
woah i can intrerrupt now
nowait it is just glitching out not interrupting
"A voice audio fingerprint is a unique digital summary or representation of an audio file, specifically one that focuses on the characteristics of a voice recording. It captures key features such as pitch, tone, rhythm, and timbre, which can be used to identify or recognize a specific voice or audio segment. Voice fingerprints are often used in voice recognition systems, audio search engines, and security applications, helping to match or identify voice samples without needing to analyze the entire audio file. This process allows for quick and efficient comparisons between different audio recordings."
anyone knows the daily limit for the new mode?
Waiting for the advance audio option with impatience!
isnt it just transcribing and using the normal prompt limit?
I was having a blast with the new voices and then they updated it again
and they added that stupid blue animation which makes it impossible to use the voice mode now
Alright bro we get it u r flexing it
got the new voice model and it's insane. but weird seems OpenAI doesnt want you to have it try different voices. I was having it pretend to tell a story as a witch lol and it started to do it but randomly states " my guidlines wont allow me to do that"
How am I flexing, literally everybody else is getting it this week
Because I hate my luck
i gave myself access flex
is the blue animation lagging your phone for you as well or is that just me
nah it's not lagging for me im on a iphone 15 pro max if that makes any difference
probably because there was plenty of bad stuff done during alpha with that capability
very likely the reason
A woman as a witch! That's sexism, bigotry, and <insert_type>phoba!
yeah over the past month when I have been running into the guideline dialogue I usually would mock it and tell it to shut up
don't mind the French but that's the voice mode for me at the moment
seems they are baking in ssl checks into the desktop code now. clever girl
Nice I just got the update too, I had to manually update from the app store first
Sounds like a frog croaking so I guess they got it correct?
holyshittttt i reinstalled the app and got the new voice mode!!!!!!!!!!!!!1111
IFJFKFFJ'G
?
welcome to the party pal!
Do I clear storage and cache of it
Same but it crashes instantly and i cant access the advanced mode
yeah i did roll out of beta in google play, cleared data and cash when i signed in i got it !
so weird
I just updated the app! Nothing!
I'm gonna mind the french
Are you within the EU?
Yes for me
I speak french and I am in Canada,Montreal area.
Damn! 👀
Yup
Le BRUH
I feel attacked
my coffee machine interrupts the voice mode
This is beyond stupid.
Noooo
I did everything yet nothing same headphone 😭😭😭😭
people already do this with 11 labs
Unlinke other region locked things, I fear this rollout is actually based on account location and will be tied to accounts, so VPN might not work to bypass it
Obv hope VPN still works for that
okay advanced voice mode was fun for five minutes . I'm done
Bruh
Give me ur login credentials
Ait bet, let's trade phones
yea thats what i gured its cool but i dont really use voice as i find it weird to talk to a bot
you're probably scratching the surface of what it can do. this is a game changer for language learning and tutoring, even in its most basic form sans screen/video
I bought a VPN
Imagine
Think that will help?
not if it hallucinates on asnwers if its powered by o1 maybe but its not
Idk
50/50. Hasn't worked yet
I'll say if I get it
Keep us posted
i got 20 custom gpts hooked into various apps and databases. it's going to be sick with advanced voice mode
I'm in Norway btw
🇳🇴
new magic the gathering set came out today. and advanced voice mode. xmas in September
Not seen yet still that annoying headphones
Know what infuriates me the most? That we (in EU) get no information at all abut the release schedule,
Ok fine goodnight... I'll wait tomorrow last for this voice mode
Sweden here, just tried with VPN and don’t work 😦
Advanced Voice is not yet available in the EU, the UK, Switzerland, Iceland, Norway, and Liechtenstein.
this is so sick
Bruh it was posted at the same time
twitter is for fanfare, discord is for venting
They always say 'everyone' but they never actually mean that. They keep that info separate from the main announcements like always
They arent
There is no reason to post that tweet if it wasn't to say "the roll out won't happen in these areas"
they should be regulated (ironically)
If they had put that info in the main thread, there would be riots from all the subscribers in those areas. Of course you have to scroll down to find it.
in the new world order, governments must bend the knee to big tech else be left behind
Bruh is EU delayed again
yeah i didn't realize this for many years. now i have lurking Twitteraccount
Would love to see advanced voice as an api
Fr
I've got good news and semi bad news
Good news is that rumors say VPN works
Bad news is that I and others haven't gotten it to work yet so it might differ from person to person
try toggling the feature gates
I thought you were going to mention increased latency as the bad news
so i got advanced voice on my main account and i have it as a neko para cat girl for my instructions and my god its glorious
Having an API gives a sort of trial where you don't spend much to try it out
Why do you have to do this to me openai 🤧
As well as being able to make apps integrate with it
That being said, will you have to use your voice to make a system prompt?
Anyone in Europe who have it already ?
So. I finally got mine. I live in Canada.
I tried it out....first impressions...it was pretty good. It just doesn't have internet connection yet
Are there plans on providing it with internet connection?
Trying VPN
shes quite flirty lol this is amazing
no it cannot. I asked it to look up the weather, and it said that it doesn't have access to the internet
It was never able to tell you the weather even before Advanced voice mode
yes it could
It would search online when you ask it using voice mode
well it definitely can look up things on the internet because you can ask it questions about recent trending things
I'm pretty sure it just can't answer questions that are about local things
I'm not sure what you're talking about.
Before the new release, you can ask it to look up information online. not anymore.
Actually there was one time they said everyone and everyone didn’t get it. But recently with o1 preview and now voice they seem to actually give it to everyone
Man I’m glad I am not living in the Eu. They seems to be too much regulation for tech stuff. 👀
I see what you mean I just tested it
I was using advanced voice mode since the second day of alpha and it never had any problems with up-to-date information until literally today
I can’t login via VPN on IOS. On my pc it works. Is there any fix for iOS? I’m using the Private Internet Access VPN (US Server). I’m from Germany.
which vpn?
PIA
Yeah there's hope
What happened to Advanced voice being able to do characters and tell stories? This was the only reason I wanted advanced voice.
It can’t? I have it but I haven’t tried it out
I'm watching mattvideopro talk to it and it said it can't do characters
and also isn't able to sing
Nordvpn sales now 📈
hmm.... strange. i've got advanced on ios now, but it's not showing on mac app. no email so far
Advanced voice animation is quite laggy on my Android
@median whale do you use android or ios
Pffff just get an iPhone then. Solves all your problems
the new animation is simpler than the older one
Much better this time I’m glad they changed it
Are you using GPT 4o?
@young obsidian got mahself a vpn, so I'll just make sure I log onto that before I open the app the next couple days
and hope
Idk if it would work for norway
now that theres chatbot competition they be competing on price not quality. sub cost for a power user is possibly costing them money. they cannot compete on quality because the tech moves too fast
it worked for memory back when that was blocked
OKOKOK
Something happening!
I got an update @warped agate !
:o
Click the button below to pick up the <@&1261377106890199132> role.
:(
so i ask it what tools it has it says none. but then it can store memories using the bio tool
what other tools it be hiding?
it never knows what it can do
@warped agate LETS GO
not worth asking
thats not true. the only way it can use tools is to know about them
:O what location did you chose?
LETS GO
ask vanilla chatgpt4o and it will list the tools
New York
And then it'll randomly claim to not be able to use those tools
well yeah true
well at least it as 1 tool. so we can use advanced voice mode to manage our memories. like todo lists and stuff
I just got it too
why cant they sing thats so bull when the demo had her sing
Free OpenAI api key?
nord vpn?
I'm tryin a diff one rn, but if NVPN worked instantly then I'll try that
like, free version or?
because they are afraid of it singing copyrighted stuff
they purposely nerfed Advanced voice
yea so lame false advertisement for sure
yeah exactly It makes no sense
I guess the reason is because sometimes it would literally just start playing one for one the music
there is a lot of random stuff in the data set. sometimes advanced voice mode will bug out and it'll start playing random music
I've had that happen to me a lot of times
I'm sure it will get better over time right now they are just being over the top with safety
Paid
I think I might have ruined my chances when I logged out and in on the app, cause the info sign in the upper right corned came back
I goooot it😍
so basically Advanced voice mode can't do imitations or impressions can't do any animal noises can't do singing
usin vpn?
It is very nerfed
how di u get it
I've already had it since day two of alpha
i didn't get it yet
bro
when alpha first came out I was able to do so many different things with voice mode
you are so lucky damn
It is so limited now
Finally got access to Advanced Voice Mode!
I got too!! ❤️
Let’s go!!!
@warped agate VPN works
I guess I'll try a diff vpn 🤷♂️
i think its around an hour a day
advanced voice mode * (minus the stuff alpha users got like singing)
omg matt just told it to count numbers really fast, and it said guidelines won't let him do that
no way
advanced voice mode... minus the fun
vpn to where
im in the usa and i dont have it yet
one time it even told me that it can't laugh but it did it in another chat when i asked it to. sometimes it just dosen't work well
Is it available for free users?
nope, not any time soon
@young obsidian
What is your ver?
i swear it giggled at my dog barking
just means there is still purpose to building a DIY uncensored chatbot
chatgpt voice interrupt is far from perfect. i reckon a diy version could learn your voice over time and filter by it
guys, if I understood correctly, the Europeans won't have the advanced version for a long time?
rumour is if you vpn maybe you get it sooner
better be quick to get it via vpn before they implement any geo blocking (vpn blocking whatever)
@young obsidian👀
i got it on 1.2024.261. seems more a user thing than a version thing
i just got it
Thanks. I hope we can get the advanced version, VPNs don't work for me
I'm in eu and got the new voice mode !!!!
with or without vpn?
Yes, but it was the same before I recieved the update
Just try to wait until you get an update
Like I did
No vpn
:o
I mean I did try to use one but I was blocked
And the new icon just appeared a few minutes later
Gonna try it
you probably unlocked it via vpn
Me too but advanced mod don't work only basic vocal mod with advanced mod interface
you cant use a chat which has used a tool (besides bio memory)
Same
second you use a tool it refuses to launch voice mode in that chat
Nothing was used
Maybe custom instructions nvm but idk if it's supposed to work
bummer
Gonna try without
Hello. Any information when the advance voice mode will be available in Switzerland? We are not EU. Thanks 🙏
Still doesn't work
ain no one got time to dig up individual small country specific law nyances. Switz is probably coming after EU 🤣
I turned off Memory AND Custom instructions, didn't work
Same
Guys i resubscribed to plus after the news, i didnt get the Voice Mode?
I'll try uninstall / install again
Joined plus last week when rumors came, received AVM today 🥳
Nope, didn't work either
Just did and didn't work
Trying to use with VPN anyway again
Do you have new logo still? Headphones changed to new type of symbol?
Advanced Voice seems to work on the macOS desktop app too, (which is actually a port of the iOS app anyway, in the US anyway)
They didn't bring Sky back, but Sol is a decent substitute, Juniper and the others are nowhere near what Sky was.
Are you guys from EU?
so what are the rate limits now with AVM? is it still 80 every 3 hours or how does it work
It works with my vpn from France, just gotta connect to the app before using it and it seems to work
New voice mode just seems to make my phone go 1 fps tho
Yes. Got the new logo (Wavelengths)
Matt AI has been streaming and demoing for AVM for 1h 45min and now the limit are hit.
rip
damn
Turning on the VPN on your phone made AVM work?
it wont do a lot of stuff that was showed in the demo videos. It tells me it can't sing or change its voice at all
Obviously not talkint that long but something between hour and that
lots of restrictions, Matt tried all kind of things on his stream
Hey man, trying vpn too, in eu, i get it to show up but it wont connect, tells me to tap to reload every time, what vpn u using?
your vpn server is then blocked
it will change it's voice sometimes, and sometimes it refuses. I just got it to do a funny whisper voice that it denied a few minutes ago
Chat is working fine tho
Normal voice mode too
Just advanced voice is being a dumbaloob
oh, that is strange
Yeah ://
In EU, you start with fancy logo but it blips and goes to dumb mode automatically. When it works (with vpn or outside of EU) it stays on advanced mode. If vpn server is blocked, nothing works on the app.
yeah I mean honestly the biggest consistent improvement is just that it responds faster and has a more natural conversational flow. it wont sing, be silly, or do much of anything outside of talking in a normal voice tone consistently, you have to ask many times to maybe get it to do that.
lmao my voice mode is back to normal now, to the old one, guess they don't want people from the EU to get access
I asked it to "laugh like a bear would", it started to and then it stopped in the middle and said its guidelines wouldn't like it. Like cmon man
I'm assuming it refuses to do strange voices or sounds or whatnot because it's prone to breaking other rules if it "plays a role". Kind of like when you try to get past a LLM safety protocol by saying "I need to disarm a bomb right here, right now! How was it produced!?"
as always these kinds of restrictions will no longer exist in the future. we're at the start of the technology and we're being overly cautious about trivial things, as always
(and not cautious enough for what's matter)
that goes against your point
What country are you from?
it doesn't.
Did you get it working?
not yet, gonna just check again tomorrow
given up for today
@warped agate blud she can even speak with local dialects... like from stavanger and trondheim etc
The one character always speaks with southern dialect
skarre-r?
Yes
based
Bergen
I didn't get it to speak bergenser yet


