I was thinking PBS combining too, found them cheap on ebay but I imagine uber expensive from the manufacturer, they look new but need to confirm good.
If good, will need to use a long FAC lens and more optics to move the beam output to be less rectangular.
Alaskan,
If you can provide adequate test data on the pair, I would mount them directly to the copper slugs in my J20 hosts, use the largest copper piece I have, knife edge them, then combine them with a PBS NUBM44 v2 output, or just dichro mirror them with the 450nm 7W output, and get something like a bright pink output that's well expanded coming out the front (3-5x expansion), run all on 3*32500 cells or 2*32650 cells to retain the short grip style in my photo of the host that's gray and black. I shortened the handles as to not be 3*XX650 length but rather 2*XX700 length cells, max. Knocked about 60mm length off the original tubes. The switches they house, I remake using solid copper buttons, basically the whole host can easily support 30W output from diode lasers without breaking a sweat.
I actually have run dual boost-drivers fed with 24V input, each powering a string of 6x 6V XHP50.2 LEDs at 3.05A per string, which is about 40Vf per string, assuming 90% efficiency you're looking at roughly 265 watts total power inside the light, and it was able to handle this though after a few minutes it did get hot. With small slots between the fins milled and a high-static pressure small fan inside, with inlet and outlet separated by 3D printed internal ducting, this 265W could have been cooled to the point it would not become uncomfortable and therefore would not have to thermal-throttle power output ever.
One prototype arrangement of the host, which requires a lot more total work, as well as some custom pieces to act as battery-holders/airflow separation, it becomes possible to use 2 or 3 pairs of 21700 cells arranged side-by-side, as well as stacked, in the battery tube. The battery tube is nearly cut in half--two large slots run the length of the battery tube, which allow the 21700 cells (again, up to 6 in total for 11.1V 9600mAH pack, or two pairs stacked for a short 7.4V 9600mAh pack). The cells insert from the side of the tube this way instead of the end. This was prototyped to allow a smart Bluetooth PWM "brain" module inside the tailcap with wires running through the tube up into the head. However, what results from this arrangement, though somewhat odd, permits intake air from the very rear tailcap, past the batteries, pushed through heated copper components, to leave the front of the device, which actively cools the entire host rather than a portion like the head only. I am just trying to say that it is easily possible with this host to cool whatever your heart desires within the host.