It's not immediately clear to me why having an LLM output code in JS or Python should inherently consume more tokens than C or assembly. I'd expect token usage to correlate with the complexity of the algorithm being implemented, not with what specific language syntax it's being implemented in.

But either way, token consumption is only a relevant concern if you still expect to be using SaaS LLMs by the time you're ready to have them target lower-level code for you.