ai-gateway

Route AI model requests across providers with unified API, failover, and observability.

Updated May 31, 2026

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

Ready to run

npx skills add https://github.com/ChristineTham/lp --skill ai-gateway-christinetham

💬 Prompt for ChatGPT / Claude Chat▼

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

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

AI developers need a single API surface to route requests across multiple providers, manage failover, and keep costs under control.

Core Features & Use Cases

Unified routing and failover across OpenAI, Anthropic, Google, and more
Per-user rate limiting and cost tagging for multi-tenant apps
Observability and simple model-versioning through providerOptions.gateway

Quick Start

Ask the AI gateway to route a model call by providing a provider/model string and a prompt to enable automatic failover and cost-aware routing.

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 requests across multiple providers like OpenAI and Anthropic?▼

Route AI model requests across multiple providers using a unified API by passing a provider/model string and prompt to enable automatic failover and cost-aware routing. It handles provider order configuration to manage fallbacks across OpenAI, Anthropic, Google, and Azure.

How does an AI gateway handle provider failover and cost tracking?▼

An AI gateway handles provider failover by routing requests to alternative models when a primary provider fails, and tracks costs using tag-based cost attribution. This provides observability and per-user rate limiting for multi-tenant applications.

What authentication do I need to set up per-user rate limiting for AI models?▼

You need OIDC-based authentication by default to set up per-user rate limiting for AI models. This authentication mechanism enables per-user controls and tag-based cost attribution to manage budgeting across multiple providers.

Can I use tag-based cost attribution for multi-tenant AI applications?▼

Yes, you can use tag-based cost attribution for multi-tenant AI applications to monitor and manage budgeting. It allows you to track spending per user or tenant across different providers like OpenAI, Anthropic, Google, and Azure.

What is the best way to unify AI model APIs for production applications?▼

The best way to unify AI model APIs for production applications is using an AI gateway that provides a single API surface for routing, failover, and observability. It manages authentication, budgeting, and per-user controls across multiple providers.

Does the AI gateway support selective model fallbacks and versioning?▼

Yes, the AI gateway supports selective model fallbacks and simple model-versioning through providerOptions.gateway. This allows you to configure specific provider order and handle failover across different model versions.