What is the purpose of core versions of Vulkan?

A total of 4 core versions were released, promoting extensions to commands that must be implemented and sometimes bringing new features, as in version 1.1, which brought support for protected operations (supported on Android). But I don't understand the need to release these core versions when the extensions already exist and can be used to drive innovation without major changes to the API and code by the developers.

What are they hoping to achieve with these core versions?

> the extensions already exist

But they don't necessarily exist. While many features of any particular version of Vulkan do come from extensions, they don't all do. Vulkan 1.3 added buffer device addresses, for example. Also, when promoted to core, extensions are often modified for various reasons.

But the biggest thing is something you'll notice if you look though some of those differences. Many of the things they "change" is that things that were optional in the extension become mandatory for a particular core version.

Yeah, sure, dynamic state exists as an extension. But in 1.3, it's not even something you have to ask about. If an implementation supports 1.3, then it has dynamic state. And not in the "well technically it's implemented but I don't support any of the optional parts" because there are no optional parts.

And that matters. In 1.3, there is a baseline of functionality that just exists; you don't have to ask about it or check beyond the version check. By coding against 1.3, you're coding against a higher baseline of functionality than 1.2 + a bunch of extensions.