Yes, I am thinking more of high impedance amplifiers, like this small design: https://www.eevblog.com/forum/projects/measuring-switch-mode-supply-noise-hp461a-replacement-ada4895/msg4652395/#msg4652395
But where do you see the problem with 50 ohms? High Z input can be terminated with 50 ohms (input noise changes of course, but that's not relevant here). Output needs to be capable of driving a 50 ohms load, which might be more of a problem. I had to include cable drivers in my "HP461A" LNA design.
Our comment was related to measuring a high impedance, here using a high impedance measurement instrument such as a DSO is generally better than a low Z device like a VNA. WRT to driving a low Z, most AWG generators and signal sources are fully capable of such.
We have "enhanced" our SDG2000X and SDG6000X AWGs to full capability as mentioned. Both are superb performers, producing clean, accurate waveforms from 16 bit DACs, the UI is a little quirky tho, but one quickly gets used to such. Either can be employed with our SDS2000X+ for Bode functionality to 120MHz.
The Bode implementation is quite good and allows user to control the input signal level across the sweep, and dynamically controls the measurement (scale factor on Ch 2) to achieve a wide dynamic range.
rfloop had demonstrated quite remarkable DR for this core 8 bit DSO in Bode function, can't remember where this is but search for Bode Plot should revel such.
Here's an example of using the SDS2000X+ and SDG6000X, note we are using in scope 50 ohms for Channel 2 and simply interfaced via LAN.
Edit: Since we had this setup, decided to measure a Bourns 3590s 10K multi-turn precision pot set for 20dBV attenuation, note the complex frequency dependent behavior reveling the inductive nature of the resistive windings and the deep notch around 70MHz indicating a resonance within.
Anyway, hope this helps.
Best,