Ryan API
中文Sign inRegisterOpen console
CapabilitiesHow it worksModelsDocs中文Sign inRegisterOpen console
Keep control of your model connections

Ryan API

Route DeepSeek, Qwen, GLM and Kimi through one clear, observable and recoverable path to your application.

Open consoleSee how it works
DeepSeek
Qwen
GLM
Kimi
Four official providers
Ryan APIRouting and health control
Ready
Your application
One interface, four model providers
Capabilities

A small, focused gateway core.

Only the connection, selection, fallback and visibility needed for model traffic, without a heavyweight business suite.

01

One protocol

Connect four model providers with a familiar Chat Completions request.

02

Controlled routing

Use priority, weight, health cooldown and stable session selection.

03

Keys stay yours

Upstream keys are encrypted by your deployment and kept out of prompts and logs.

How it works

One request, three explicit decisions.

01

Match

Filter enabled and available channels by model.

02

Select

Apply priority, then weight or stable session routing.

03

Recover

Cool down failed channels and move recoverable calls to the next connection.

Compatible endpoint

Change less in the application you already have.

Call different upstream models through one endpoint while the console handles configuration, testing and visibility.

Configure a channel
chat-completions.sh
curl /v1/chat/completions \
  -H 'Authorization: Bearer $RYAN_API_KEY' \
  -d '{"model":"deepseek-chat","messages":[...]}'