If you rank by total murders Chicago comes in an easy first. Comparing a very large city like Chicago to comparatively smaller cities blurs the numbers when you are talking per capita.
For example St Louis is a small inner city surrounded by numerous suburbs cities that have much lower crime rates. Those low crime suburbs don't get counted into the per capita rate. Chicago is large enough of a city that both the high crime and low crime areas all get counted together.
But ranking by total murders is definitely wrong, isn't it?
You can have criticisms of the per-capita rate, and you can argue that the smaller cities should have their suburbs included. But the alternative of just saying "let's look at the raw numbers independent of population and see which one is better" would be even more absurd.
Just as an example here, there are plenty of reasons to be distrustful of the per-capita comparisons of Covid infection across states, or to argue that they don't capture all of the nuance of what's going on. But nobody would ever seriously argue that those problems mean you should stop paying attention to population size when comparing Covid rates between states. Similarly, it just seems really silly to me to argue that missing suburbs means we should stop controlling for total population in crime data.