Skip to main content
Applicable to: Users whose applications send the x-tfy-routing-config header directly in request headers, or rely on prompt/agent flows that inject this header automatically.

​
What Is Changing

We are retiring support for the x-tfy-routing-config request header in AI Gateway. This header is not supported on and after v0.133.

​
Why This Change

Routing logic should be configured declaratively via Virtual Models rather than passed as request headers.

​
What You Need to Do

  1. Move routing logic (targets, weights/priorities, retries, fallbacks) into a Virtual Model.
  2. Update clients to call the Virtual Model name/path instead of sending x-tfy-routing-config.
  3. Validate behavior in staging before upgrading to v0.133.

​
Migration Guidance

Refer to the Virtual Models documentation for detailed setup instructions.
If you need help migrating, reach out and we’ll be happy to assist.