I haven’t got enough data yet but my intention is to delve deeper along the lines of answering your question:
There are many considerations in running a variable-speed compressor/fan split system but some basic physics probably answers your question best …
Assuming constant occupation and constant heat load and the desire for a constant temperature, it stands to reason that the LOWEST POWER that can keep up with the heat load is ultimately more efficient than a cycling system. All thermostatically controlled systems will cycle around the target temperature and the highest efficiency will come from extending those cycles.
In reality though you typically need to balance occupation, load and thermal mass with the maximum capacity settings of your system. Cooling only the air is something that can happen quickly but if you have a lot of thermal mass and sun-exposed windows warming it or other heat load, it will necessitate either a higher power of cooling or waiting longer (in potential discomfort) for the air to cool the thermal masses.
An interesting aspect of air cooling (the usual manifestation of cooling is, after all, Air Conditioning) is that human thermoregulation for homeostasis is often best achieved starting with the head, hands and feet. Sticking your feet in a bucket of iced water will quickly allow you, or even make you want to, increase the temperature on your thermostat!
I have a reasonably well insulated space with good non-exposed thermal mass (blackout blinds at peak sun!) so running at low power is how I achieve a stable temperature.
One last consideration is efficiency of the entire system. e.g. If you have solar and are losing efficiency by sending power to the grid then during those periods, often of peak solar and so peak heat, it may well be more “efficient” to crank your AC system and do some over-cooling of thermal masses. Generate ice at noon!