#GPT 4 API waitlist
46 messages · Page 1 of 1 (latest)
I think you can pay 20usd /month and you will have GPT4... still we're living in a money driven world 😄
Or you are talking about the API endpoint so you can make calls to the AI ?
the API endpoint haha
@severe ice @graceful phoenix
Hello👋, the time is not specificly set, you can keep an eye on the openAI website, regrading to the GPT-4 API. You can try reaching to OpenAI customer supoort or community forums, to see if they can provide you with more specific information about the waitlist. Hope it helped👍
Did you tell me how to identify GPT-4? what are the main indications of GPT-4?
I got the invitation but still cant use it lol
Thing is it really isn't a "list" because as usual some people are getting access ahead of others despite signing up later. Who knows what criteria they use but I too have been waiting since day 1 without any luck so far
If the OpenAi authorized you for GPT4 API, why can't use it? Is that possible they don't allow any integration for application development yet?
Is there any transparency on this waitlist? I swear I've been waiting way longer than other people that have already gotten access. Was just talking with a coworker who said they were on the list for just like 2 weeks and then got access!
Access is granted based on quota, so it may take a while for some people to get their access to GPT-4
but rest assured you'll be emailed when it's ready for you and your waitlist request was submitted
It may take a day or two to show up on your account fully
The point is some of us, including myself , having been waiting many weeks if not longer, while others singing up recently only waited a week or even less. so clearly the list isn't FIFO?
It seems possible if our use case is estimated to require more resources, that we are being bucketed into a more expensive part of the waitlist
For example, we can already see Auto-GPT getting rate limited on the chatgpt-3.5-turbo endpoint. Agents using gpt4 must be having similar demands on the API (such as always using the 25 per 3 hours, unlike a human).
The list is not a FIFO, no
Access is granted based on capacity, it wouldn't be a simple first come
The API doesn't have a 25 per 3 hours limit
It is rate limited by requests per minute and tokens per minute
Been waiting for fricking 11+ days
Millions of us are waiting. You've found the support group
took me 2 days 
Access is granted based on capacity, it wouldn't be a simple first come
This still isn't clear to me. Can someone who apparently is more intelligent than me understand what is being said here? Clearly they are judging some applicants differently than others and I was simply asking for some transparency into that measures they are using. But "capacity" means nothing. How is one applicant different than another as far as capacity goes.
Saying you are granting access based on capacity, in MY mind, simply means you are looking at how busy the service is and if there is free capacity you add more. This perfectly implies a first in first out system as I stated, which it clearly isn't.
So what about applicants allows their "capacity" to get granted before another? The word used is not making anything clear at all to me
Based on capacity mens based on how much computation power they have.
we're becoming GPT junkies. Gimme more NOW!
Right I get that but how does that disqualify simply giving access based on when someone signed up?
My point I'm trying to make/understand is that first come first served is not incompatible with access being given based on capacity.
Yea dude, I hear you . Been waiting since the announcement too. Seems like it's full random to me.
Does submitting an updated use case reset your position on the waitlist? We have a new group proposing something for the gpt api for the org and determining if we think it should be a new submittal.
Been wondering the same actually
same unfortunetley
Okay, first I'm not from OpenAI so I cannot say for sure, but I'm "from the field" and I can give some thoughts on how they are probably doing it.
They had thousands of people joining the API waitlist given the nature of the hype, so it's completely fair to assume that a considerable number of people in this first wave will never use the API or will not build any product at all (which is the purpose of the API to start with), and that's fine, you're paying for it after all, but it's probably something OpenAI don't want to see the API being used for.
Given that, the best way to avoid the first wave of people that will not make the best use of the API while not prioritizing one group over another is to just do it randomly and maybe filter out manually the selected ones as well, in this step they may also filter out the use cases that they don't want to see GPT-4 being used for (or a use case that may compete with them or investors). However, that is a thing that I don't think they're doing objectively, I think they're just randomly selecting and sorting out the absurd cases like we normally do.
They may also do it in waves, they're optimizing the model and scaling infrastructure, once they have "more capacity", they can select 100/1000 or whatever the number of people they want and sort out some of them to give out access (they don't need to release to those 100+ at once, but over the time if they want to).
And to be clear, I agree that who signed up first should get access first, but that's not what we normally do in this matter to ensure that we are choosing the right individuals.
I'm really interested in building some nice products but still didn't get any access to the API even tho I signed early on, the best I can do is wait and work in the other parts.
They say somewhere they’re prioritising people who give quality feedback on the something something. I had a look and it wasn’t for me, so I’ll just wait. I’m sure my own personal project idea doesn’t really inspire them to bump me to the top of the list either. Hopefully by the time I need it, it’ll come through…
Yeah that's about the evaluation models but that is going to be a very, very small percentage of people
I have applied for the API multiple times, and I am not getting access. anyone that got access have any tips? or can someone share the application that got them access?
well, for me it took exactly two days to get onto it, and the only thing i can guess, is that either a human or even a ai is reading what you guys are writing there.
it is clearly stated on the site where you apply, that they primarily want people to have access, that have great ideas and drive the whole thing further with their contributions.
what i did, was showing clearly that i understand the capabilities of this technologie and also showing that i already got ideas and good use cases for it, (to be honest i made a little list of my thoughts on it, as i am currently working on a game that utilizes gpt 4 not only for speech of characters and story, but also for the technical parts, for validating data or calling the right functions without implementing the logic, and i also pointed out other projects of mine, where the upcoming computer-vision feature could be used for disabled people (blinds for example) or as automated software support and auto.fixing...)
from what i heard from a few friends it pretty much seems like they act in the way they state it and give priority to those having big ideas and dont only want to circumvent the rate limit or want to make a little applications for themselves and family. could be wrong with that assumption though^^
Well I guess maybe I should rewrite my proposal to make it sound more interesting
This is a very good way of putting it!!!!
You need to show a good use case for the API, they aren't giving access just for fun
to be honest i think this is the wrong approach.
i am personally pretty sure that openai is using their own software to as a first instance of filter, before the proposals go to any human being (i mean they would be pretty dumb if they had these tools which are that great at sentiment and context analysis and wouldn't use them, right?). therefore you should not at any point really believe that making the proposal only SOUND more interesting is going anywhere, as still the conceptions and ideas are the primary factor, and not your rhethorical skills that may let real people believe an idea may be great.
take your time, and rather look how you can make the idea itself more interesting as a whole concept and try it again then. anything else will just be a major waste of your energies (and lets be honest it is also a waste of computational and human recources^^)
thanks a lot, but i think it is the most logical one^^ i mean if openai would not clearly communicate this i would ask the same questions and would wonder, but even though i integrated ai in most parts of my life now and still extend that, i am still able to read whole texts before i send a proposal^^;)
I approached it as a student, and they gave me access to the Chat version within 4 hours
I've been waiting for the API access since Day 1. I pay for GPT4 access, but the API key is realy what I want to play around with. Anyone figure anything out to speed it up or increase the likelihood?
GPT Plus doesn't have anything to do with the API, there is nothing else that can increase the likelihood apart from a good application on the waitlist form
Well I’m not just talking about swapping out my words for more interesting synonyms. I can explain more why it’s interesting. I’m making a fully autonomous physical bot. I’m hoping it won’t get bored while I’m out at work and start calling to my neighbors to let it out. That’s interesting enough, right?