Since 5.9 we've seen a number of different compilers/interpreters/JITs that produce different results. When optimizing for binary size you will generally

​

• build static binaries
• turn off as many Qt features as you can
• prefer smaller compilation artifacts over run time efficiency (ie, you might not use qmlcachegen but rather compile at run time)

​

Let's take an example that works well with all the different technologies and compare how small or large the binary gets if we turn the different knobs.