Class NaruCustomProvider

java.lang.Object
net.thevpc.naru.api.model.AbstractNaruModelProvider
net.thevpc.naru.ext.models.custom.NaruCustomProvider
All Implemented Interfaces:
NaruModelProvider, net.thevpc.nuts.spi.NComponent

public class NaruCustomProvider extends AbstractNaruModelProvider
Generic OpenAI-compatible provider driven entirely by configuration. Lets users point NARU at any OpenAI-compatible endpoint (LM Studio, vLLM, llama.cpp server, LiteLLM proxy, ...) with zero code.

Configuration (per named endpoint):

 custom.endpoints=<name1>,<name2>,...
 custom.endpoints.<name>.url=https://my-server/v1
 custom.endpoints.<name>.apiKey=sk-...
 custom.endpoints.<name>.models=model-a,model-b
 custom.endpoints.<name>.chatPath=chat/completions   (optional)
 custom.endpoints.<name>.contextLength=32768         (optional)
 

Models are addressed as custom/&lt;endpoint&gt;/&lt;model&gt;.