Is there a way to do negative lookahead in vim regex?

In Vim, is there a way to search for lines that match say abc but do not also contain xyz later on the line? So the following lines would match:

The abc is the best
The first three letters are abc

and the following would not match:

The abc is the best but xyz is cheaper
The first three letters are abc and the last are xyz

I know about syntax like the following:

/abc\(xyz\)\@!

but that only avoids matching abcxyz and not if there is anything in between, such as abc-xyz. Using

/abc.*\(xyz\)\@!

also does not work because there are many positions later in the line where xyz is not matched.

(I should note that on the command line I would do something like grep abc <infile | grep -v xyz but I would like to do the above interactively in Vim.)

Your attempt was pretty close; you need to pull the .* that allows an arbitrary distance between the match and the asserted later non-match into the negative look-ahead:

/abc\(.*xyz\)\@!

I guess this works because the non-match is attempted for all possible matches of .*, and only when all branches have been exhausted is the \@! declared as fulfilled.

I always find it weird, that you need to escape brackets in vim, so I try to use the "very magic" mode most of the time, activated with \v:

/\vabc(.*xyz)@!

Although your question is about lookahead, I found it while searching for lookbehind. Thus, I post the solution to lookbehind so that others may find it.

If you search for pattern2 not preceded by pattern1, that is negative lookbehind, search for:

\(pattern1\)\@<!pattern2

This works for me /abc\(.*xyz\)\@!

I think I'd write this as /abc\(\(xyz\)\@!.\)*$

I'm not sure whether this is faster than other suggestions (I think it might be since it tests each position only once) but it makes conceptual sense as:

"abc followed by anything which is not xyz, any number of times until end of line"

I like that better than "abc followed by anything not followed by xyz" as in the other patterns mentioned.