Aliases

Aliases let you use semantic names like smart or fast instead of specific model names. Relai automatically routes to the best model matching that profile.

Available Aliases

Using Aliases

Simply use the alias as your model name:

{
  "model": "smart",
  "messages": [
    {"role": "user", "content": "Explain quantum computing"}
  ]
}

Benefits

Future-Proof Code

When better models become available, your smart requests automatically upgrade:

// Your code doesn't change
const response = await client.chat.completions.create({
  model: "smart",
  messages: [...],
});

// But you get the latest best-in-class model

Cost Optimization

The cheap alias automatically routes to the most cost-effective model that can handle your request:

// Perfect for high-volume, simple tasks
const response = await client.chat.completions.create({
  model: "cheap",
  messages: [{ role: "user", content: "Summarize: ..." }],
});

Latency Optimization

The fast alias prioritizes response time:

// Ideal for real-time applications
const response = await client.chat.completions.create({
  model: "fast",
  messages: [{ role: "user", content: "Quick question..." }],
});

Response Headers

When using aliases, the response includes headers showing the actual model used:

X-Relai-Model: claude-sonnet-4.6
X-Relai-Alias: smart

Custom Aliases (Coming Soon)

Organization-specific aliases will let you define your own routing rules:

# Example future config
aliases:
  internal-chat:
    models: [gpt-5.5-mini, claude-haiku-4]
    strategy: lowest-cost
    max_tokens: 4096