documentation llama.cpp

llama.cpp

base URL
http://localhost:8080/v1
key
not needed

A bare llama.cpp build serves an OpenAI-compatible API on your own machine.

Base URL

http://localhost:8080/v1

No API key is needed.

Adding it

Open Settings → Providers, add llama.cpp from the preset list, then use Fetch models. If the build exposes no /models route, choose Type a model name instead — hand-typed names are remembered per provider.

all documentation