Relay simplifies your AI infrastructure, with one single connection.

    const response = await fetch("https://api.respan.ai/v1/chat/completions", {
  method: "POST",
  headers: {
    "Content-Type": "application/json",
    "Authorization": "Bearer rspn_live_xxxxxxxxx"
  },
  body: JSON.stringify({
    model: "openai/gpt-4o""anthropic/claude-3-5-sonnet"
    messages: [{
      role: "user",
      content: "Summarize this customer support conversation."
    }]
  })
})

    Use one API for every model

    Point your existing code at Respan instead of each provider. Swap gpt-4 for claude-sonnet by changing one string. No rewrite.

    gpt-5.5
    Down
    Rerouting request sent...
    claude-sonnet-4
    deepseek-chat

    Stay online when providers fail

    If a model goes down or rate-limits, Respan automatically retries and routes traffic to the next available provider. Your users never see the failure.

    Auto tracing....
    SuccessPOST/chat1.23s
    SuccessPOST/chat0.91s
    RetryPOST/chat2.40s

    Balance load across providers

    Spread traffic across models and regions automatically. Get better latency and cost without managing the routing yourself.