b798e2679f
Expose the llama.cpp/GGUF command-line binaries through a lightweight OpenAI-compatible transcription endpoint.