MIDI is a completely different world compared to audio.
Imagine a piece containing e.g. an organ player pressing down one key and holding it for 5 minutes.
The MIDI file would look trivial, just one note on with a length of 5 minutes and a note off. This can be coded in a dozen bytes.
Then you do an audio export of that piece and the audio engine has to generate audio for the duration of 5 min, i.e. generate roughly 40000 samples each second, so that ends up with dozens of MEGA bytes.
You can’t seriously compare these two worlds.