llm-gateway

Manifest LLM Gateway is an open-source LLM gateway designed for AI agents and applications. It consolidates API keys, subscriptions, and local models into a single OpenAI-compatible endpoint, allowing each query to be routed to the appropriate model without provider lock-in. Key capabilities include custom routing across API keys, subscriptions, local models, and custom providers; full body logging for success and error messages; cost tracking with notifications and limits; and automatic fallback to different models when queries fail, including self-healing for bad requests. The project supports 300+ models through 35 built-in provider connections plus any custom OpenAI or Anthropic-compatible endpoint. Users can bring their own API keys, reuse one of 18 subscription flows, or run models locally. Providers include OpenAI, Anthropic, Google, AWS Bedrock, DeepSeek, Mistral, Ollama, LM Studio, llama.cpp, OpenRouter, Hugging Face, and many others. Sending "model": "auto" lets Manifest select the model. Deployment options include a cloud version at app.manifest.build, a self-hosted Docker image, and one-click deployment templates for Railway, Render, DigitalOcean, AWS, GCP, Fly.io, Coolify, Easypanel, Heroku, Koyeb, and Apple Containers. The project is MIT licensed and provides documentation, an n8n community node, and a Discord community.