This list contains programming languages that are obviously in use and some which are proposed for usage in machine learning. A fundamental issue with programming languages remains the same. There are n programming languages available of which none is considered perfect so someone invents a new one that fixes a couple things. We end up with n=n+1 programming languages. Then someone else invents another programming language and again n+1…. .


  • C++
    • underlying language of many python packages and certainly the most used language for any deployment case that requires low latency
  • CUDA
    • most ignored language as basically everything that runs on GPUs is somewhat compiled cuda coda (sometimes via some abstractions layers)
  • Cython
    • abstraction language to create C extensions for Python easily (many python packages - e.g. cupy use it to achieve at least some decent performance)


  • dex
    • research language for typed, functional array processing



  • Python
    • most likely the defactor language for machine learning research, teaching and deployment


  • R
    • in decline, seems to get replaced by python a lot; still more relevant for more classical statistics applications


  • Swift
    • used for deployment of machine learning solutions in the Apple universe


  • Zig
    • claims to be a simple language that is claimed to be useful for machine learning