ClaudeGate
An open-source proxy that lets Claude Code and the Anthropic SDK talk to 24+ other LLM providers, including local models, without changing your setup.
🔗 Visit ClaudeGate
Description
Claude Code and the Anthropic SDK are built to talk to Anthropic's models — but plenty of teams want to point that same tooling at a different provider, a self-hosted model, or a cheaper alternative during an outage. ClaudeGate is an open-source proxy that sits between Claude Code (or any Anthropic SDK client) and 24+ OpenAI-compatible providers — DeepSeek, Groq, OpenRouter, Ollama, and more — so you can swap the backend without changing your workflow.
It handles SSE streaming without crashing on mid-stream disconnects, automatic failover across providers, bidirectional tool/function calling, and a built-in PII/secret sanitizer that strips sensitive data before requests leave your machine. It supports both cloud providers and fully local/private models (Ollama, LM Studio, vLLM), plus enterprise routes like Azure OpenAI and AWS Amazon Q. It's MIT-licensed, free, and actively developed, though still early with 6 GitHub stars.
💬 Our review
The short version: if you use Claude Code but want the flexibility to fall back to DeepSeek, Groq, or a local Ollama model — for cost, latency, or privacy reasons — ClaudeGate gets you there without touching your existing Claude Code setup, and it's free.
There's no direct like-for-like competitor with the same Claude Code-specific focus; general LLM gateways like LiteLLM cover similar multi-provider routing but aren't purpose-built around the Anthropic SDK's streaming quirks. ClaudeGate's PII sanitizer and zero-crash streaming are meaningful differentiators if you're routing sensitive data through third-party providers, but running your own proxy is an extra piece of infrastructure to maintain, and the project's low star count means less community-tested edge cases than a more established gateway.
💰 Pricing
📊 Global score
🤖 AI-enriched data
100% gratuit, licence MIT. Auto-hébergé, aucun coût d'usage propre — vous payez uniquement les providers LLM que vous routez à travers.
Pros
Compatible avec 24+ providers cloud et locaux out-of-the-box
Streaming SSE sans crash même en cas de déconnexion
Sanitizer PII/secrets intégré avant l'envoi des requêtes
Failover automatique multi-provider
Fonctionne avec des modèles totalement locaux/privés (Ollama, LM Studio, vLLM)
Cons
Nécessite d'héberger et maintenir un serveur proxy local
Communauté encore réduite (6 stars GitHub)
Documentation limitée pour les déploiements entreprise avancés
Dépend de la disponibilité des API des providers externes