← Back to context

Comment by dangoodmanUT

7 days ago

Yes everyone should just write cpp to call local LLMs obviously

Yes, but llama.cpp already comes with a ready-made OpenAI-compatible inference server.

  • I think people are getting hung up on the "llama.cpp" name and thinking they need to write C++ code to use it.

    llama.cpp isn't (just) a C++ library/codebase -- it's a CLI application, server application (llama-server), etc.