ai-gateway

Name: ai-gateway
Availability: InStock
Author: Cobalt

Route AI requests across providers with failover and cost tracking.

7|Updated Mar 15, 2026

Search Tags:#llm #observability #failover #model routing #cost tracking #ai gateway

Ready to run

npx skills add https://github.com/Cobalt-Money/Cobalt --skill ai-gateway-cobalt-money

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: ai-gateway
Source: https://github.com/Cobalt-Money/Cobalt/tree/main/.agents/skills/ai-gateway
Command: npx skills add https://github.com/Cobalt-Money/Cobalt --skill ai-gateway-cobalt-money

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires ai, @ai-sdk/gateway.

What problem does it solve?

This Skill solves the complexity of managing multiple AI providers by providing a single, unified API endpoint that handles model routing, provider failover, and granular cost tracking.

Core Features & Use Cases

Intelligent Routing & Failover: Automatically switches to backup providers or models if the primary service experiences downtime or rate limits.
Cost Attribution & Budgeting: Tracks token usage per user and feature using custom tags, allowing for precise budget alerts and spend management.
Use Case: A multi-tenant SaaS application can use this to route requests across OpenAI and Anthropic, ensuring high availability while enforcing per-user rate limits and monitoring costs by feature.

Quick Start

Configure the AI Gateway to route requests through openai/gpt-5.4 with a fallback to anthropic/claude-sonnet-4.6 for production reliability.

Frequently Asked Questions about ai-gateway

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

FAQPage Schema

How do I handle AI model failover across multiple providers like OpenAI and Anthropic?▼

AI model failover automatically routes requests to backup providers like Anthropic when primary services like OpenAI experience downtime. This ensures high availability for production AI deployments by abstracting provider-specific SDKs through a unified endpoint.

What is the best way to track LLM API costs per user in a multi-tenant application?▼

Tracking LLM API costs per user requires custom tags applied to token usage data. This enables precise budget attribution and spend management across features, allowing multi-tenant SaaS applications to monitor consumption and enforce per-user rate limits.

Does the AI gateway require OIDC-based authentication for secure token management?▼

Yes, OIDC-based authentication is required for secure, zero-maintenance token management. This integration ensures that API requests routed through the Vercel AI SDK remain protected while accessing multiple model providers.

Can I use this with the Vercel AI SDK to enforce per-user rate limits?▼

Yes, integrating with the Vercel AI SDK allows you to enforce per-user rate limits effectively. The gateway abstracts the underlying provider SDKs, managing routing and throttling to ensure production-grade reliability for your application.

Why do I need an AI gateway instead of calling provider SDKs directly?▼

You need an AI gateway to avoid managing provider-specific SDKs directly. It provides a unified interface for model routing, automatic failover during outages, and granular cost tracking, which are essential for maintaining production-grade AI deployments.

How do I configure intelligent routing for production AI reliability?▼

Configure intelligent routing by setting a primary model like openai/gpt-5.4 with a fallback to anthropic/claude-sonnet-4.6. This setup ensures the gateway automatically switches providers during downtime or rate limits, maintaining production reliability.