From the source
Vercel is adding Virtual Models to AI Gateway, letting teams point apps and agents at a single model ID that can be reconfigured without code changes or redeploys.
- The feature abstracts the model string and its settings, so a task like coding-agent can be created with a CLI command and then called as vmc/coding-agent wherever that task runs.
- Vercel says 125 new models from more than 10 labs arrived on AI Gateway in the last 90 days, about 10 a week, and that switching between them previously meant editing a line of code and redeploying.
- Editing a virtual model, for example from anthropic/claude-opus-5.5 to openai/gpt-6.1-sol, propagates to every app and agent calling that ID within a few minutes.
- Settings such as fallbacks, Zero Data Retention and tags stay with the virtual model when the underlying model changes, and they override the same settings in a request, so callers cannot change provider order or failover.
- Vercel advises pointing a second virtual model at a new model and running evals against it before switching one that serves production traffic.
- Virtual Models work with every model type on AI Gateway, including image, video, speech, embedding and realtime models, and with the AI SDK and the - and -compatible APIs.
