Pilot jets and passages not factory-fresh clean from "dying when throttle opened" symptoms.
Soaking is not enough, chemicals needs to be used with scrub-brushes on all hidden secret passages from end-to-end. Also bend copper wire 10mm from end at 90-degree angle. Use to poke out bleed-holes in carb-venturi in reverse direction back towards float-bowls.
Also clean out choke-jet and passages. Jet's not removable and needs tiny wire to poke through. Had case where jet was so clocked, a micro drill-bit needed to be used to clear.
These tough hard dried-petrol plastic deposits can't be chemically removed. Tough metal-based mechanical methods needs to be used.
Also need to actually measure objective concrete quantifiable data on petcock flow-rate.
If petcock's not flowing enough, doesn't matter how clean carbs are, bike will die when petrol-usage rate is higher than petcock can flow.
Problem with this test is you may be artificially applying more vacuum than under actual operating conditions. Same with sucking on vacuum line. I had case like this where petcock test flows just fine. Then I'd get 30-seconds/quarter way through lap at track and it'd die! Hauled it back to pits, repeat petcock test and it's fine. Start it up and it'd run just fine for 2-5 minutes to warm up. Get on track and 30-seconds later, it'd die again!!!
Had to connect vacuum-gauge to petcock hose with rubber-T and monitor actual vacuum applied at petcock port. AHAHH!!!! Petcock was getting only 2-3 in.Hg vacuum! WTF?@#$$!? Turned out I had crimped and clogged vacuum-line between carb & petcock. It would dribble enough through at idle and low-loads to fill up flow-bowl. But as soon as I went WOT, engine would consume petrol at faster-rate than dribble from petcock could supply. Replaced that vacuum-hose between petcock and made sure it wasn't crimped fixed that problem for good. NEVER, EVER would've found it without measuring with gauge to gather actual in-the-field hard data.