I think it's relevant to add "perspective" to the definition.
Any end-point of communication can be said to also be an entry-point.
Let me elaborate.
Most forms of communication that we utilize today can be captured by a graph consisting of 2 vertices and 1 vertex which is incident to those 2 vertices.
I.e. applications A
and B
communicate and we model this as an undirected graph G:
Or any of the following directed graphs:
Example
Let's take A -> B
as an example. Here we communicate from A
to B
.
Suppose B
has defined a a location L
it controls and defined to be open to communication.
Now A
sends a communication-payload to location L
.
Sender's perspective
From the perspective of A
the recipient B
is an "end-point". In the sense that there is no further control of the communication-payload once sent to B
.
Recipient's perspective
From the perspective of B
however, the location L
is an entry-point. In the sense this is where B
has an opening, a place where it allows entry.
Communication-channel's perspective
The final perspective is that of the communication-channel, i.e. the edge the connects A
and B
. Here, both A and B can be considered to be end-points of communication or rather, delivery-points. Since a communication-payload is delivered through the communication-channel to either A
or B
.
However, you could say that, as with the above statements, the communication-channel is also an entry-point, as it is where a communication-payload enters to be delivered.
So which is it?
In the case of Representational State Transfer (REST), I'd wager the usage of an "endpoint" is an inheritance of RFC 9293 specification which describes the TCP protocol. Since TCP underlies HTTP, and HTTP underlies REST I'd wager this is where the term came from.
Note however that the RFC 9112 specification for HTTP only uses the term "endpoint" once and it does so in the context of a "server endpoint". In other words, this is an endpoint from the perspective of a client communicating to a server.
As for REST, it has no "official" specification. The term "REST" was coined by Roy Thomas Fielding in his PhD. dissertation and his dissertation does not mention entry-points nor endpoints.
Another point worth making is that the entrypoints mentioned in the other answers are actually the "base"-url/path of the API and the OpenAPI specification specifically mentions these as part of the Server Object definition.
TL:DR;
All of this would imply that the "endpoint" and "entry-point" terminology in REST comes down to common usage/phrasing and not clearly defined terms as proposed in the other answers.