In studies like this one mapping out distributions around the world or the analysis described in this article (suggesting a contribution from a non-Denisovan but Denisovan-like unidentified hominin to Melanesian populations, various single digit percentages are given of an average as high as 6%.
One complication in talking about this is what we mean by percent. Just through shared DNA we'd have over 99%, if we knew which DNA was shared. There are surely other complications, like how do you count repetitions toward the percentage, etc. But I mean to ask what percentage of DNA that is DNA descended from introgressions adds up to, if you put together the different and overlapping pieces various people have.
I realize it may also be hard to pin down exactly what one would mean by a Denisovan genome -- as I said one of the linked articles suggests there may at least one Denisovan-like hominin species represented in the introgressions.