A lot of this depends on the "buzz cycle" of your specific market. My assumption is that you want to build and maintain the buzz. I am not aware of an industry where waiting 7 days until the next version will have negative impact on buzz -- unless you hyped it already and are facing unrealized expectations which you don't seem to have done. Not do the two options seem mutually exclusive to me.
One potential negative option of launching version 3 before version two has realized a life cycle is that you will cut down on the potential buzz that you could be delivering. Build until attention is about to wain-- then then deliver the next. Will attention begin to wain on version 2 in 5 days?
In other words -- do #1` and then have version 3 ready to go at the right time. that might be 7 days when it is ready-- you might embargo it a little bit to fully realize the media and "buzz" cycles of version 2.
Maybe you can embed version 2 with a feature that let's folks pre-buy into version 3 -- acceleration the buzz and testing volume assumptions?