We are listening, but we don’t have enough capacity to onboard more developers into the GPT-4 beta at the moment. The best way to try it out is via ChatGPT Plus!
The idea above is that these test accounts would have limited quota and wouldn’t increase capacity requirements.
You could even increase the price per token, it wouldn’t be an issue due to the restricted quota.
Also, as mentioned, OpenAI would be able to see how people are going to use the API if they got access.
Copy and pasting REST requests in to ChatGPT+ is error prone. It’s also hard to create realistic and compelling demos.
Also, fwiw, I’ve seen some code scattered around the web to hook up to a headless browser/chrome extensions. Most def not going to do this as I’m sure it’s not what OpenAI wants, but a test account would be a way of dealing with that foolish nonsense.
@logankilpatrick the gpt4 throttling on my chat plus subscription is starting to become rather arduous. Any idea when capacity will increase? If it’s months from now, I might as well cancel my subscription and come back later. As a typescript coder, gpt-3.5-turbo is just too clunky. gpt-3.5-turbo makes too many errors, hallucinates, cannot dig it’s way out like gpt-4, gets stuck in loops.
for gpt-4, is demand the same 24hrs per day 7 days a weak? or are there periods of low demand when it would make sense to provide incentives (ie increased requests/hr) for usage during off peak times?
We are trying to make the experience for developers better, who have been asking for this feature. The scammers and people violating the TOS will be dealt with. Thank you for flagging the above thread, please ping me anything else like that you see.
Thanks for the clarification, it really helps a lot for those of us who desperately need access but are unwilling to contravene the wishes of OpenAI.
The restrictions below mention API keys specifically, but not proxying requests to the API.
Is reselling quota considered against the TOS if it’s through a proxy service?
(c) Restrictions . You may not (i) use the Services in a way that infringes, misappropriates or violates any person’s rights; (ii) reverse assemble, reverse compile, decompile, translate or otherwise attempt to discover the source code or underlying components of models, algorithms, and systems of the Services (except to the extent such restrictions are contrary to applicable law); (iii) use output from the Services to develop models that compete with OpenAI; (iv) except as permitted through the API, use any automated or programmatic method to extract data or output from the Services, including scraping, web harvesting, or web data extraction; (v) represent that output from the Services was human-generated when it is not or otherwise violate our Usage Policies; (vii) buy, sell, or transfer API keys without our prior consent; or (viii), send us any personal information of children under 13 or the applicable age of digital consent. You will comply with any rate limits and other requirements in our documentation. You may use Services only in geographies currently supported by OpenAI.
I sincerely apologize for any potential violation of OpenAI’s rules regarding API usage. Here I hope to provide some sincere clarification on my actions. The post mentioned by @qrdlhere may give the impression that I was selling API keys, however, it in fact should not be the case. I did not actually sell or transfer any API key:
(vii) buy, sell, or transfer API keys without our prior consent; —OpenAI term of use, 14 Mar 2023
I used a middleware to re-post user requests to GPT-4 and control their usage using the moderation endpoint. It is not designed to give out API keys; it is originally just part of our own product that collects and processes end-user data stream. As so many forum members expressed interest in trying AutoGPT, I thought it would be beneficial and legitimate to propel the community forward by efficiently sharing our scarce unused quota via this middleware. This middleware operates the same way as other product built on GPT-4: I did not give out my API key, and no one can abuse their GPT-4 access. Most importantly, this proxy operates on a non-profit basis, allocating a very limited quota only for sharing purposes.
I have always strived to adhere to OpenAI’s rules and have no intention of violating them. In light of this situation, @logankilpatrick, I kindly request re-consider the restriction placed on my account. We are a small startup team that has poured countless hours and significant resources into building our product on the GPT-4 API. Losing access now would not only be a devastating blow to our passion but would also render our tireless efforts in vain. We would greatly appreciate any assistance you could provide. Really sorry for my oversight.
Once again, I sincerely apologize for any inconvenience this may have caused, and I appreciate your understanding & your time.
Also, does anyone else find it interesting and slightly cool that there is already an “AI black market”. @logankilpatrick Could be an interesting test case in how you monitor and control misuse and access. I’m 99% sure there will be export controls on AI in the next year or two. Luckily you can deal with this issue early before it becomes a legal requirement.
Microsoft is pretty liberal with their access though. I’ve been on their waitlists for a couple of things like dataset and services before. I usually get access within less than a week. I think it is more of a “cool kids club” effect for marketing purposes and a way to filter out people who are just excited but don’t actually intend on using it.
In the case of OpenAI they are pretty researcher heavy and I think they will have a rough time bug fixing and researching everything. So if people want to develop business critical tools, do it on Azure, if you want to help research different applications that your business does not rely on, use OpenAI API.
They are a research company not a service provider, that’s the whole point of working with Microsoft.
Sure, you need to first join the MS OpenAI service pool, which takes just 1-2 days. Then you need to apply for GPT4 api access via form. If your use case is approved, they will shortlist you to ‘GPT4’ queue.
In my case, I stuck here for 5-6 days, and got their sales staff’s email to schedule a short phone meeting. The meeting is scheduled 3 days later.
I don’t mean the queue takes shorter time - but best thing is that I know clearly where I am and know whom to speak with. MS got sufficient business guys dealing with such things. If you got problems, just email them. You will get replied.
As GPT products continue to expand, I can imagine how challenging it must be for OpenAI to manage everyday business/debugging/pre-sale/after-sale supports. I am also AI researcher and really don’t hope top researchers to be dogged down by business concerns.