1kW peak INPUT power equals 500 Watts average input power. The FCC defined amateur power levels as "input" power back then. When those rules were written in 1934, Joe Ham could afford a meter for DC plate current, and he knew what the final's high voltage should be. Plate current times HV equals input power. In 1934 a RF wattmeter was an exotic lab instrument that cost better part of a working man's annual salary. Took almost 50 years for RF wattmeter technology to get cheap and widespread enough for the FCC to redefine amateur power levels as actual RF power.
RF output is typically about 2/3 of "input", the DC power you draw from the high voltage supply. When sideband took over the FCC declared that the ratio of peak power to average was two-to-one. Hence an amplifier with 1000 Watts average input power would show a peak input power twice the 1000 Watt average. This is why the Drake and Heath 3-500Z-by-two amplifiers were offered as "2kW". A peak (PEP) to average ratio of two to one was the standard assumption back then. Even so, I have seen exactly one linear in my life with a peak-reading plate-current meter. That feature just never did catch on.
But back the the four 811As. If you got that two-thirds of 500 Watts average power, the PEP won't be any more than twice that much, or around 600 Watts PEP.
The plate of the 811A is exquisitely sensitive to overload and permanent damage. Either get used to running at half throttle or watch those tube plates like a hawk.
The 811A was leading-edge tech in the 1930s. So was a flat-head V8.
73