bifrost

Bifrost is an open-source AI gateway written in Go that routes requests across 23+ providers, including OpenAI, Anthropic, AWS Bedrock, Google Vertex, Azure, Groq, and Ollama, through a single OpenAI-compatible HTTP API. It connects to Claude via its Anthropic provider integration and includes a built-in MCP gateway that enables AI models to invoke external tools such as filesystem access, web search, and databases. Deployment requires no initial configuration and takes under a minute using either npx or Docker, with a web UI at localhost:8080 for visual setup, real-time monitoring, and analytics. Core features include automatic provider failover, load balancing across multiple API keys, semantic caching, virtual key budgeting with hierarchical cost controls, Prometheus metrics, and distributed tracing. The project claims under 100 microseconds of added overhead at 5,000 requests per second. Enterprise tiers add adaptive load balancing, cluster mode, guardrails, and OIDC-based user provisioning. The primary audience is platform and backend engineering teams managing multi-provider LLM infrastructure at production scale.

Lo que la gente pregunta sobre bifrost