I conducted a performance comparison between Node.js and Python using the official OpenAI client, and the results were quite revealing. Node.js outperformes Python in this specific use case.

Can you re-run your tests with `node --jitless` flag please? It should be a fair comparison.

to keep the CPU in healthy thresholds I lowered the concurrency to 100.

   real 0m22.188s
   user 0m14.702s
   sys 0m0.449s

It's still a lot faster but less so, could you explain why I would want to run --jitless in production?

I did not ask the comparison with production in mind, but to make things fair between one runtime that executes with JIT by default versus the CPython that does not yet, but we are getting there!