The API was obviously meant to be consumed by the ChatGPT website.
But people have already built cool products on top of ChatGPT - that allow it to be used in a variety of scenarios - which naturally get it more publicity and generate more data.
What is OpenAI’s position on using the unofficial API?
Is it okay to use it for now while we wait for an official release of the API?
I imagine the abuse possibility isn’t super high since you ultimately have to do a passthrough to an OpenAI user account whose access token is then used - and that is rate limited. Although it’s obviously still not an optimal situation.
I’d personally take the position that using it is fine while ChatGPT is in Research Preview.
If you have used it in a product, you are taking a bit of a business risk in the long term
There are indications that it will be billed. So its fine while it free, but when they start billing, you will need to have a use case that can service the ongoing running costs
Without this, most of those sites using the unofficial ChatGPT API will probably no longer exist
I guess we can enjoy it while it lasts, and use it to research ideas until then
Also, I think ChatGPT tracks usage by IP address. That would explain all the issues people have with CloudIP and VPN’s that share an IP address. If a ChatGPT app takes off (viral), its hosting IP address will probably get black listed anyway.
If we consider scenarios like the ChatGPT extension for VS Code, or extensions for email, etc. Those requests usually come from your own IP and using your credentials.
The unofficial API… Is really just people using the existing API that the ChatGPT website uses, but outside of the site.
They do have some anti-abuse stuff like captchas, which the API also handles.
As for payment… Since most of the ChatGPT apps I’ve seen so far are themselves free, when it becomes paid, you’d just need to add a payment method on your ChatGPT account.