I would say that is "cloud hosted", which is obviously very expensive compared to running on hardware you own (assuming you own a computer and a GPU). That was the comparison I was interested in, the fact that renting a computer is more expensive than the OpenAI API is not a surprising result.
The very first definition from the link you provide is:
> Self-hosting is the practice of running and maintaining a website or service using a private web server, instead of using a service outside of someone's own control.
Hosting anything on Amazon is not "using a private web server" and is the very definition of using "a service outside of someone's own control".
The fact that the rest of the article talks about "enabled users to run their own servers on remote hardware or virtual machines" is just wrong. It's not "their own servers", and we don't have "more control over their data, privacy" when it's literally in the possession of others.
> The practice of self-hosting web services became more feasible with the development of cloud computing and virtualization technologies, which enabled users to run their own servers on remote hardware or virtual machines. The first public cloud service, Amazon Web Services (AWS), was launched in 2006, offering Simple Storage Service (S3) and Elastic Compute Cloud (EC2) as its initial products.[3]
I hate when terms get diluted like this. "self hosted", to me, means you own the physical machine. This reminds of how "air-gapped server" now means a route configuration vs an actual gap of air, no physical connection, between two networks. It really confuses things.
No. I googled "self hosting", read the first few definitions, and they agree with the article, not you. E.g., wikipedia -- https://en.wikipedia.org/wiki/Self-hosting_(web_services)