Agent Router (formerly Envoy AI Gateway)

Agent Router is an open source AI gateway for teams connecting agents to model providers and MCP tools. It gives applications a shared entry point while platform teams manage access and routing.

• Model routing: Use an OpenAI-compatible API to reach Anthropic, AWS Bedrock, Vertex AI, Azure OpenAI or self-hosted vLLM, with model aliases that hide provider-specific names.
• Traffic controls: Set token limits by team, application or model, and use provider fallback when a request can't reach its intended backend.
• Central credentials: Agents authenticate to the router, which holds provider API keys and AWS, GCP and Azure credentials outside application code.
• MCP tool access: Combine tools from multiple MCP servers into a shared catalog and filter the tools available to each caller.
• Inference routing and monitoring: Kubernetes InferencePool support selects endpoints for self-managed inference, while OpenTelemetry GenAI reporting tracks model choice, token usage, time to first token and fallback events.

The Apache 2.0 project runs on a laptop, a dedicated gateway or Kubernetes. Self-hosting controls where the gateway runs; requests still reach the model providers and MCP servers you select. Tetrate Agent Router Service is a separate hosted offering that runs outside your infrastructure.