maximhq/bifrost
High-performance AI gateway written in Go that unifies access to multiple LLM providers through a single OpenAI-compatible API with built-in load balancing and failover.

Bifrost is an enterprise-grade AI gateway that provides unified API access to over 23 LLM providers including OpenAI, Anthropic, AWS Bedrock, and Google Vertex. It acts as a reverse proxy with adaptive load balancing, automatic failover, semantic caching, and guardrails. The gateway supports cluster mode for horizontal scaling, offers an MCP gateway and client for model context protocol integration, and includes a web UI for configuration and monitoring. It aims to achieve sub-100 microsecond overhead while handling 5,000 requests per second.