Vercel AI Gateway now supports routing rules. Routing rules are firewall-style rules that control which models your team can use, applied at the gateway level instead of in your application code.
When a model goes down or gets retired, you usually have to ship a code change to move off it. With routing rules, you push one rule and every request reroutes instantly. There are two types:
Type | What it does | Use it to | | Serves a request for one model using another | Keep traffic flowing when a model is unavailable, migrate off a retired model, standardize on one model, or route an expensive model to a cheaper one | | Blocks requests for a model | Keep your team off models you haven’t approved |
Rules apply to every request made with your team's AI Gateway credentials. You manage them with the Vercel CLI.
Create a rewrite with a source and a destination model. The Gateway swaps in the destination transparently, so your application keeps requesting the source model:
Create a deny rule to block a model. Requests for it return a 403
:
Rules only change which model serves a request. Everything else you’ve configured still applies to the destination model, including:
Request-level: [BYOK], model fallbacks, sorting, the only
filter, and provider options. Team-level: Zero Data Retention and the provider allowlist.
Routing rules are in beta. For more information, read the routing rules docs.