Sine Gateway / AI model gateway
A unified model API with routing you control.
Manage provider connections, discover models and publish routes for your application. Bring your own provider keys or use available platform routes, with call records and organization-level controls in one Console.
What you can build
Manage your model connections
Keep organization provider credentials in Gateway, validate connections and discover models. Inspect model capabilities before choosing a route for your application.
Choose how requests reach models
Create a single-target route, configure fallback targets or distribute calls across targets. Your application uses the route’s model ID through the unified API.
Inspect calls and control spending
Review persistent call records and recorded usage. Organization owners and admins manage provider connections and the separate budgets required for platform-managed calls.
Sine Gateway
Get started in the Console
Built for your workflow
- A shared model entry point for your applications
- BYOK access with centrally managed provider connections
- Model fallback and traffic distribution through named routes
Open Gateway in your organization
Review available platform models or add a BYOK provider connection. Creating a project or running a separate Gateway service is not required.
Configure and verify a route
Select models and a routing strategy. Check the provider’s capabilities and required parameters, then test the chosen route in the Console when you are ready to make a call.
Integrate the model endpoint
Create a Cloud API key with Gateway permissions and read the model list. Use a returned route ID in your application’s requests and inspect the resulting call records.
Read-only API example
Discover the routes your key can use.
Read the model list from the Gateway API and use its model IDs when making calls. Listing models does not invoke a provider model.
export SINE_CLOUD_URL="https://sinecloud.dev"
curl "$SINE_CLOUD_URL/api/products/sine-gateway/v1/models" \
-H "Authorization: Bearer $SINE_CLOUD_API_KEY"The Cloud API key needs Gateway modelRead permission. Invocations separately require modelInvoke access, a configured route and any applicable budget and wallet checks.
Set SINE_CLOUD_API_KEY securely in your environment with the required product permissions. Keep it on your server; never expose it in browser code.
Access and billing
Workspace access
Gateway routes are scoped to your organization and current permissions. Model capabilities and supported operations depend on the configured provider and route; inspect the model details before preparing a request.
How usage is charged
BYOK provider usage is billed by your provider; the Cloud BYOK invocation service fee is currently zero. Platform-managed calls use published prices, organization budgets and the Cloud wallet. A successful HTTP response alone does not establish final usage or settlement.
View pricing