Vercel Launches Virtual Models for Centralized AI Management

What happened

Vercel has introduced Virtual Models on its AI Gateway platform, enabling developers to manage model configurations and updates from a central location. This change allows developers to adjust model settings and switch between different models without altering their application code or redeploying their services.

Previously, developers had to make a one-line change to switch models, which required redeploying their applications. With Virtual Models, the same task can now be accomplished by simply assigning a virtual model ID to an agent task, with the model string used as the identifier.

Why it matters

This update simplifies the process of managing multiple AI models, reducing the complexity and time needed to update or switch models. Developers can now focus more on their core applications rather than the underlying model management, which enhances productivity and system flexibility.

Key details

Since the launch of AI Gateway, 125 new models from over 10 labs have been added in the past 90 days, averaging about 10 new models per week. This demonstrates the growing ecosystem of available models that developers can integrate into their workflows.

The use of a single virtual model ID allows developers to change models and settings behind the ID in AI Gateway, with updates propagating automatically to all connected apps and agents. This centralized approach streamlines model management and ensures consistency across applications.