Relay simplifies your AI infrastructure,
with one single connection.

const res = await fetch(
  "https://api.relay.dev/v1/chat", {
  method: "POST",
  body: JSON.stringify({
    model: "claude-sonnet",
    messages,
  }),
})

Use one API for every model

Point your existing code at Relay instead of each provider. Swap models by changing one string — no rewrite.

Book a call
gpt-5.5 down
claude-sonnet-4 rerouting
deepseek-chat

Stay online when providers fail

If a model goes down or rate-limits, Relay retries and routes traffic to the next provider. Users never see the failure.

Book a call
SuccessPOST /chatgpt-4o1.23s
SuccessPOST /chatclaude-40.91s
RetryPOST /chatgpt-4o2.40s

Balance load across providers

Spread traffic across models and regions automatically for better latency and cost.

Book a call