ai-gateway

Name: ai-gateway
Availability: InStock
Author: Vercel

Route AI model calls through a unified gateway with OIDC authentication.

246|42|Updated Mar 4, 2026

Search Tags:#authentication #observability #model routing #cost tracking #ai gateway #provider failover #per-user rate limiting

Ready to run

npx skills add https://github.com/vercel/vercel-plugin --skill ai-gateway-vercel

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: ai-gateway
Source: https://github.com/vercel/vercel-plugin/tree/main/skills/ai-gateway
Command: npx skills add https://github.com/vercel/vercel-plugin --skill ai-gateway-vercel

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Centralizes expert guidance for configuring and using Vercel AI Gateway to unify model routing, manage provider failover, enable cost tracking, and ensure observability across providers.

Core Features & Use Cases

Unified multi-provider routing and automatic failover to minimize downtime
Cost attribution and budgeting with per-user tagging and observability
Auditable per-request governance and actionable logs for compliance and monitoring
Easy integration with AI SDKs and OIDC-based authentication for secure deployments
Use Case: Orchestrate OpenAI, Anthropic, and Gemini models behind a single endpoint with consistent policy

Quick Start

Install AI Gateway, connect your project, and route a model using gateway('provider/model').

Frequently Asked Questions about ai-gateway

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I route AI model calls through a unified gateway for multiple providers?▼

Route AI model calls by configuring a unified gateway using gateway('provider/model') to simplify multi-provider usage. This centralizes model routing, provider failover, cost tracking, and per-user governance across your AI workflows.

What is provider failover and how does it minimize downtime for multi-provider AI workflows?▼

Provider failover is an automatic mechanism that switches AI model calls to backup providers when the primary fails. This unified routing minimizes downtime and ensures consistent availability across multi-provider AI deployments.

How do I set up per-user rate limiting and cost tracking for AI models?▼

Set up per-user rate limiting and cost tracking by applying per-user tagging and budgeting rules within the gateway. This enforces per-request governance and provides auditable logs for compliance and cost monitoring.

Can I orchestrate OpenAI, Anthropic, and Gemini models behind a single endpoint?▼

Yes, you can orchestrate OpenAI, Anthropic, and Gemini models behind a single endpoint. The gateway applies per-provider models, routing rules, and OIDC-based authentication to secure multi-provider deployments.

Does AI gateway integration work with AI SDKs and OIDC authentication?▼

Yes, AI gateway integration works directly with AI SDKs and OIDC-based authentication. This combination enables secure deployments while providing actionable logs and auditable per-request governance for compliance.

What are the limitations of using a unified gateway for multi-provider AI routing?▼

A limitation of using a unified gateway is that routing rules and per-provider models must be explicitly configured for each provider. Users must manage OIDC authentication setup and tagging policies to ensure accurate budgeting and compliance logging.