This is totally wrong way to measure signal generator amplitude flatness and accuracy from low frequencies up to 500MHz. There is all wrong. Totally. Even if you connect external 50ohm terminator to oscilloscope input it is wrong. This oscilloscope have 50ohm internal selectable impedance but even if there is oscilloscope internal 50ohm input many times they are not at all well enough for accurate amplitude flatness measurements. Because many times they are not enough accurate 50ohm impedance over whole frequency band. Oscilloscope specs say only -3dB @ 50ohm without more detailed data. Also this cable looks like some audio quality coaxial.
About external feed thru terminator (what do not with this oscilloscope exept if want go over oscilloscope specified max level with its internal 50ohm) connected to oscilloscpe 1Mohm example 16pF input (what in real world have all three: resistance, capasitive reactance and also some inductive reactance. Result is far away from 50ohm if look 500MHz (problems start in practice when we go roughly saying over 100MHz and more and more out of order what higher freq we go. If it is 50ohm terminator with coaxial T - this is most crap way. Specially if terminator is cheap "Ethernet network grade crap terminator". Without perfect impedance match you work with standing waves what amplify your signal with some frequency anmd attenuate with some frequency, dpending these waves phase. Also these reflected return to generator output may affect its ALC depending how it is done some generators.
Fun side note about this oscilloscope. Sad that this level of scope have this kind of trigger jitter. Visible around video time position 12.15. Do they have at all fine interpolation for accurate positioning. Also there was linear interpolation. Sinc interpolation is better for sinewave like signals.
For good level flatness measurements you need different tools. 500MHz is not anymore "audio".
Rough measurements you can do using enough good and trusted SA, also it can show something about sinewave quality. If you claim from oscilloscope screen that sinewave quality is very very good it is bit loosy to say so. Simply, it is difficult to see it from oscilloscope screen, exept if we talk about very poor sinewave what have very high levels of harmonics or other impurities. Example if there is second harmonic.
Think we simply combine 500MHz 0dBm sinewave and then 1000MHz -30dBm sinewave and this combined wave you connect to oscilloscope. What you see. Looks like quite pure 500MHz sinewave. How about if this, 500MHz sine, 1000MHz (500MHz 2. harmonic) is -20dBm and then 3. harmonic (1500MHz) is example -40dBm. Still you can see it is quite nice looking sine shape in oscilloscope screen. But in reality, it is quite poor.
Just for info so that random reader do not need calculate: With 500MHz, 16pF reactance is 19.9 ohms!
(16pF is because in this example it is nominal 1M input capacitance with this named oscilloscope.)
Summarized briefly:
This video do not tell anything about SDG6000 amplitude accuracy or flatness or sinewave quality.
It can be worse or better or what ever. This test is faulty for this purpose.
But, other ways still very nice to see this LeCroy model and this Siglent model live in video.