Make your first model call with one configuration
This is the shortest practical path through the console: sign in, prepare a model configuration, save its Access Key, copy the inference Base URL and downstream model ID, then send a minimal validation request.
Sign in to the console
Sign in to the Lumoswitch console with an existing account. On first use, complete account verification, then create or maintain your model and invocation configuration.
Connect or choose candidate models
Connect existing upstream model credentials in the console, or choose model resources available in this deployment. Create a model configuration, select candidates and a scheduling strategy, then define the downstream model ID used by clients.
Create the configuration and save its Access Key
After you submit the model configuration, the console creates an Access Key associated with it. The complete key is shown only once at creation, so copy it immediately to your team's appropriate secret manager.
Copy the Base URL and downstream model ID
Copy the Lumoswitch inference Base URL from the configuration or client integration page and use the downstream model ID defined by that configuration. Do not put an upstream provider credential in a client; clients should use the invocation information associated with this configuration.
Validate with a minimal request
First validate authentication, the model ID, and response format with a minimal request before integrating production traffic. If it fails, check configuration enablement, model availability, the saved key, and account and runtime conditions in that order.
Confirm one request returns successfully
This example assumes the Base URL shown in the console includes the applicable API version path. Replace the environment variables; use the current console and documentation for actual paths, capabilities, and model IDs.
curl "$LUMOSWITCH_BASE_URL/chat/completions" \
-H "Authorization: Bearer $LUMOSWITCH_ACCESS_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "YOUR_DOWNSTREAM_MODEL_ID",
"messages": [{ "role": "user", "content": "Hello" }]
}'Before production
- The complete Access Key is shown only once at creation; save it and keep it out of frontend code and public repositories.
- Clients use the configuration's downstream model ID, not an arbitrary upstream model name.
- Validate a minimal request first, then introduce production traffic in line with the current configuration, model status, and billing conditions.