Comment by tom_mellior

> Interpreters usually execute by walking the abstract syntax tree at runtime, bytecode VMs usually execute actual bytecode similar to machine code.

This isn't the right way to separate these concepts. A VM that executes bytecodes by interpretation is also an interpreter (Python, Ruby, PHP are well-known examples). Many VMs don't interpret, or they interpret mostly during startup, but actually try to compile hot parts of the code and execute that machine code rather than interpreting (Java and JavaScript VMs are well-known examples).

The VM part more properly refers to things like data representation and memory management. Interpreters of an abstract syntax tree, interpreters of bytecodes, and just in time compilers all need to run inside a system that takes care of loading code, allocating memory when requested, and often doing garbage collection. These services are what make a VM. The exact execution strategy is not what determines VM-ness.