Quickly comparing datasheets of 3004 vs. 2004 :

4x memory 40/80M vs. 10/20M (interleaved/noninterleaved)

2x sample rate (2.5/5G vs. 1.25/2.5G)

14pf vs. 9pf input capacitance (1M)

500uV vs. 1mV/div

runt & rise/fall time triggers

"Up to 64Kwfms/sec" vs. "Up to 50K" ( a little disappointing)

8 vs. 4 active measurements

Some sort of sequence triggering ( edge A/B) - not sure exactly what this is

Additional maths : maximum, minimum, square, square root, absolute value, positive wave, negative wave, reciprocal,

inverse, log10, ln, derivation, integration

~~(but no filter options like Keysight have - lowpass is handy for visualising PWM signals)~~ Mentioned in manual but not datasheet.

Spectrogram (option)

More decodes - similar set to Keysight -ARINC, 1553, I2S etc. No USB or ethernet.

4 vs. 2 decodes

Uart decode 6Mbaud vs. 3Mbaud max

Wavegen 10V vs 5V max output, ~~AM/FM/ASK/FSK, Sweep~~ RTB2000 has this but not on datasheet

Fast segmented mode for up to 700Kwfms/sec. Overlay/avarage/enve lope segments

"spectrum analysis and spectrogram" option - not obvious how the former differs from FFT

Power analysis option

So as expected, as the RTB2004 was clearly aimed at the MSOX2000, the 3004 is aimed at the MSOX3000, with very similar ( but not quite all) functions & options.

I'm supposed to be getting one but don't know when.