I'll admit to thinking a little more about this subject....
I get my numbers from the Tesla app as my system is 100% Tesla. The production graph in the app plots a curve based on numbers updated at 5 minute intervals. The plotted number is the result of some calculation that comes up with an aggregate number based on the variations within the 5 minute period. So, catching a burst generation number requires looking at the "real time" generation number when the burst occurs, and cannot be observed after it passes. Therefore, any actual highest burst is unknown, only an observed number is available, and then as a memory rather than a record. I therefore have reduced confidence in any burst number I might report.
Looking at the best days, the best burst production I've seen, and the best production curve posted number, a few observations emerge....
1 The best plotted graph numbers I have never appear on the best production days. They appear on days with a production graph chopped up by smaller fast moving clouds, and appear as peaks imediately following a low point resulting from temporary cloud blockage. My highest yet is 13.4 kw.
2 The best burst production numbers that I've observed do seem to occur on some of the best days, but do not last long enough to obviously influence the plotted aggregate numbers.
3 So of course, the best production days are on graphs that have not been "stretched" vertically to accomodate the higher plotted numbers.
Along with the above possibly meaningless observations, I'll share that yesterday (March 24) I had the sixth best production day that I've had since going into operation on June 10 of last year, and the best day since July 5 at 57.3 kwh. My production is a little constrained both early and late in the day from living in the woods.