Provider keys
Set and rotate server-held provider credentials at runtime, and give a provider several named keys for routing to pin.
Routing
The default routing config, per-request overrides, model aliases, retries, upstream timeouts, and conditional lanes on verified identity.
OpenAI
Route OpenAI requests through the Anyray gateway, which speaks the OpenAI REST API and holds the key.
Anthropic
Run Claude through Anyray with your team's own Claude subscriptions, a shared Anthropic API key, or both. Setup for Claude Code and the Anthropic SDKs.
Vertex AI
Route Vertex AI requests (Gemini, Claude, Llama) through the Anyray gateway, which signs Vertex calls with a server-held service account.
Azure
Route Azure OpenAI requests through the Anyray gateway, which speaks Azure OpenAI and targets your deployment.
Bedrock
Route AWS Bedrock requests through the Anyray gateway, which speaks Bedrock and signs SigV4 itself.
LiteLLM
Put the Anyray gateway in front of your LiteLLM proxy: keep LiteLLM's routing and virtual keys, add optimization and per-developer attribution.
TrueFoundry
Put the Anyray gateway in front of your TrueFoundry AI Gateway: keep TrueFoundry's routing and tokens, add optimization and per-developer attribution.
Tensormesh
Route open-weight models through Tensormesh Serverless Inference, where cached input tokens cost nothing, and let Anyray price the $0 tier correctly.