The three detail levels

The same document can be heard three ways. All that changes is how much gets told and whether it may step outside the slide. You pick it from the top bar, at any time, and it is remembered for the next time you open that document.

The three levels, and what each one promises.
LevelWhat it tellsWhat it guarantees
SummaryThe core ideas: what you would remember if you could only remember one thing. It aims for a third of what the full level would run to.It adds nothing that is not on the slide. It does not promise to cover everything: that is what the other two are for.
FullEverything the slide says, without skipping a concept, a figure, a condition or an exception. If there are eight items, all eight show up.Total coverage, verified in code. And it does not invent: if the slide does not say it, it does not get said.
ExtendedEverything in full and then develops it: says it in other words, gives a concrete example, tells you why it matters in practice and flags the typical mistake.Total coverage, and it is the only level that brings in context from outside the slide. When it does, it shows in how it is said.

Which one to use

You can change level halfway through a document and nothing is lost: what was already generated at the other level stays on disk, so going back to it is instant.

What else the level changes

The level also governs what you are asked: the question at the end is about what was explained, not about what the slide says and the summary left out. Asking you about something nobody told you measures nothing.

What does not change are the marks. Whatever you marked as not understood is stored per slide block and not per segment, so it stays marked when you change level: the same paragraph that is one segment in summary is three in full.