Skip to article frontmatterSkip to article content
Site not loading correctly?

This may be due to an incorrect BASE_URL configuration. See the MyST Documentation for reference.

5.1. Juno Native API

InferenceApiServer (Javalin) is the single HTTP entry point on the coordinator. It exposes the Juno-native surface below alongside the OpenAI-compatible API. Both surfaces share the same underlying RequestScheduler and GenerationLoop, so behavior and performance are identical regardless of which surface a client uses.

Native inference endpoints

MethodPathHandler
POST/v1/inferencehandleBlockingInference: blocking, returns GenerationResult
POST/v1/inference/streamhandleStreamingInference: SSE, one event per token
GET/v1/modelsOpenAiChatHandler.handleListModels
GET/v1/models/{modelId}OpenAiChatHandler.handleGetModel
DELETE/v1/models/{modelId}handleUnloadModel
GET/v1/cluster/healthhandleClusterHealth: per-node health rollup

Health and console

MethodPathHandler
GET/handleConsole: embedded coordinator web console
GET/health-uihandleHealthDashboard: node health dashboard HTML
POST/health/probehandleHealthProbeProxy: proxies probe to HealthReporter
GET/health-datahandleHealthDataProxy: proxies health JSON from nodes

See also


<- 4.8 Testing Checklist  |  Table of Contents  |  5.2 OpenAI-Compatible API ->