Connecting Claude Clients with Azure API Management and Claude Models in Microsoft Foundry

2026-06-04 · Source: Microsoft Foundry Blog articles · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure, Software Development & Engineering · Depth: Intermediate, extended

Summary

A production-ready pattern for connecting Claude clients (Claude Code, Desktop, Cowork) with Anthropic's Claude models in Microsoft Foundry is detailed, utilizing Azure API Management (APIM) as an LLM gateway. This setup, published on June 04, 2026, addresses key organizational challenges like API key sprawl, uncontrolled token usage, and lack of visibility into costs and usage. The architecture involves Claude models in Microsoft Foundry (billed via Azure), APIM authenticating developers with Entra ID, enforcing per-user rate limits and token quotas via GenAI gateway policies, and emitting per-user usage metrics to Azure Monitor. Foundry can reside in a separate Azure subscription (Subscription B) from APIM (Subscription A), with APIM authenticating to Foundry using either a Foundry API key or a managed identity. Setup time is approximately 2 hours for full implementation or 30 minutes for a basic pilot.

Key takeaway

For AI Engineers or MLOps teams deploying Claude Code, this pattern offers a robust solution to manage access and costs. You can centralize billing, enforce per-developer rate limits and token quotas, and gain granular usage visibility without exposing Anthropic API keys. Consider implementing the managed identity option for enhanced security and simplified key rotation in production environments. Start with a Developer SKU APIM instance for a quick pilot.

Key insights

An Azure API Management gateway provides secure, metered access to Anthropic Claude models in Microsoft Foundry.

Principles

• Decouple developer and backend authentication.
• Enforce token and request limits per user.
• Track usage with detailed metrics.

Method

Configure Azure API Management with Entra ID for developer authentication, GenAI policies for rate limiting and token quotas, and either a Foundry API key or managed identity for backend authentication to Claude models in Microsoft Foundry.

In practice

• Deploy Claude Sonnet 4.6, Haiku 4.5, Opus 4.6 in Foundry.
• Use "llm-token-limit" and "rate-limit-by-key" policies.
• Set "ANTHROPIC_BASE_URL" to APIM endpoint.

Topics

• Azure API Management
• Microsoft Foundry
• Anthropic Claude
• LLM Gateway
• Entra ID
• Token Quotas
• Cost Management

Best for: AI Engineer, MLOps Engineer, AI Architect

Related on AIssential

• Giving Developers Claude Code with Azure API Management and Claude Models in Microsoft Foundry — Centralize "Claude Code" access via Azure APIM to manage authentication, usage, and cost for developers.
• Anthropic Launches Claude Platform on AWS — Enterprises prioritize operational ecosystem integration over model choice for AI adoption.
• Alishahryar1 / free-claude-code — A proxy enables free use of Claude Code by routing Anthropic API calls to alternative LLM providers.
• Claude Code Testing New Usage Limits | Opus 4.7 on Claude Pro for AI App | 🔴 Live — Anthropic's increased Claude limits and SpaceX partnership aim for massive compute expansion, enhancing AI development.
• Claude Code, A Technical Guide for the Business Executive — Claude Code is transitioning from a developer tool to a secure, full-stack agentic OS, leveraging whole-project context.
• Wtf Anthropic — Anthropic's compute miscalculation and opaque policies are alienating users, while OpenAI aggressively captures market share.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Microsoft Foundry Blog articles.