Interesting comparing the answers here that are for a "looks-like" vs a "works-like" prototype, ref:
https://predictabledesigns.com/the-essential-guide-to-protot...Having no experience in gamedev, I could imagine that as in other fields these milestones are a bit orthogonal. You need a works-like prototype for internal use to confirm the basic questions like "is it fun" before committing to polish, and you need that vertical slice looks-like prototype to do demos, make a sizzle reel for Kickstarter, convince investors, etc.
The excellent NoClip doc on Horizon Zero Dawn talks about them actually making a handful of prototypes that explored various aspects of the game's mechanics (combat, traversal, etc) and then having to figure out how to integrate those concepts into a single entity. There are a few clips at https://youtu.be/h9tLcD1r-6w?t=1186 that show a very early demo of the Thunderjaw concept, with the destroyable parts just as big colored boxes.
On the other side of things, it's wild when you look at a game like Hollow Knight: Silksong, for which there was a reveal trailer over two years ago that showed seemingly dozens of fully finished environments, enemy encounters, etc— but in terms of the overall stage of the project, it must have been in alpha or pre-alpha at that point (unless it's encountered some significant unforeseen development barriers): https://www.youtube.com/watch?v=pFAknD_9U7c