Register an app to get a Kong consumer, API key, request-transformer plugin, and a LiteLLM virtual key mapped onto the request.
No apps yet
Create an app, or import Kong consumers that already map to a LiteLLM key.
Live gateway inventory
Kong consumers and LiteLLM teams this portal can see
Kong
Consumer
Key
Bearer
PII
LiteLLM
Team
Keys
PII
Inference Registry
Models served by LiteLLM — self-hosted, OpenAI, Anthropic, SiliconFlow, or custom. Config-backed models stay read-only.
Model
Provider
litellm model
Used by
litellm config.yaml
Preview from the live model list
model_list: []
Langfuse Traces
Requests observed by Langfuse for this gateway. Open a row to see the timeline.
…
Trace
App
Model
Latency
Tokens
Cost
Status
Time
Connection Settings
This portal calls Kong, LiteLLM, and Langfuse directly. Endpoints come from the portal environment — one process, no extra service to deploy.
How the pieces fit together
App Registry creates a Kong consumer and key-auth credential, then attaches a request-transformer so the gateway swaps the client apikey for a LiteLLM virtual key.
Inference Registry registers model backends into LiteLLM. Clients only ever receive the Kong apikey.
Traces reads Langfuse so you can follow a call: which model, latency, tokens, and whether it failed.
Register a new app
Creates the Kong consumer, API key, request-transformer, and LiteLLM team/key.
App
Kong Gateway
LiteLLM
Clients receive only the Kong apikey. The LiteLLM virtual key stays on the platform and is injected by Kong.