Skip to main content
Use this checklist before you move from local testing to production traffic.

Request and model safety

  1. Pin exact production model identifiers.
  2. Validate the exact request shape you send in production.
  3. Re-test capability-specific behavior when switching providers or models.

Stateful behavior

  1. Decide how you will store conversation state.
  2. Verify tool loops, retries, and replay behavior if your app uses tools.
  3. Test streaming behavior if your UI or workers depend on it.

Error handling

  1. Handle synchronous API errors such as auth failures, validation errors, and insufficient credits.
  2. Handle mid-stream failures if you use streaming.
  3. Log request IDs, status codes, and failure modes in your own monitoring.

Access and cost control

  1. Separate development and production API keys.
  2. Confirm paid-model access and account balance before launch.
  3. Monitor costs and usage through account activity and usage data.