Could someone briefly sum up the differences between F# and OCaml? Are there any reasons to switch from OCaml to F# in a Unix environment?

OCaml can compile to native. That makes it faster than F# running in a VM. OCaml can compile to bytecode as well. Also, if you ever need interfacing with C/CPP, OCaml is better suited (faster) to the task than, say, Ruby or JVM based languages.

Apart from that, multithreading in OCaml is a bit like on its on Python, IIRC, with GIL taking the sting out of it. But I am sure the researchers in France and elsewhere would bring true multithreading / multicore support to OCaml sooner rather than later.

More on what OCaml gets right: https://realworldocaml.org/ (which seems to be down right now).

There is a multicore runtime in the works, with an ETA of "when it's done". That said, you have a couple of full-featured monadic concurrency libraries (Lwt and Async).

It also has functors (they let you parametrize a module over other modules), which F# doesn't, and you can do AST rewriting at compile-time via PPX (don't know if F# has anything like that). On the other hand, F# has access to the .NET ecosystem (which is admittedly larger than the OCaml one, though it suffers from enterprisitis), and has some nice goodies like type providers.