Except for Sergiy's first query, the plans are very similar and are costed exactly equal. I'd guess that my APPLY version with two extra operators (Compute scalar and TOP, each costed at 0%) would perform slightly slower than Sergiy's second query. TOP doesn't come for free even if the plan says otherwise. Scott's might also be a little slower with a left semi join NL rather than a left outer join NL.
For fast, accurate and documented assistance in answering your questions, please read this article.
Understanding and using APPLY, (I) and (II) Paul White
Hidden RBAR: Triangular Joins / The "Numbers" or "Tally" Table: What it is and how it replaces a loop Jeff Moden