The question is: What is the gap between microdnf and dnf?

Linux Containers have emerged as a key open source application packaging and delivery technology, combining lightweight application isolation with the flexibility of image-based deployment methods.

Red Hat Enterprise Linux (RHEL) base images are meant to form the foundation for the container images you build. As of April 2019, new Universal Base Image (UBI) versions of RHEL standard, minimal, init, and Red Hat Software Collections images are available that add to those images the ability to be freely redistributed.

RHEL minimal images provide a base for your own container images that is less than half the size of the standard image, while still being able to draw on RHEL software repositories and maintain any compliance requirements your software has.

Building custom images using Containerfile or Dokerfile  sometimes you need to install packages on top of the minimal images of RHEL.  You need to use microdnf to install things not dnf /yum.

Answer: As minimal as stated: no Python and no Python module dependencies. Which are quite many packages to think of it.

I suppose the actual gap will come also from the fact of not using Python:

• There is no Python interface, and thus you can’t invoke microdnf from a Python code using a consistent API. You’ll have to resort to using the subprocess Python module
• Actual dnf can be expanded with many additional commands provided by the dnf-plugins-core and other plugin packages. You may not expect any of those features in microdnf. They will hardly ever make it to microdnf.