ModelAlias

Stop hardcodingLLM model names.

Create stable aliases for model identifiers. Update them once from your dashboard and every app follows immediately.

No redeployment required.

Works with any provider or SDK

terminal

# Resolve an alias from any environment

> curl https://alias.vibekit.dev/v1/resolve/coding \

-H "Authorization: Bearer ma_live_••••"

> claude-opus-4-6

Screenshot of a developer post describing the frustration of changing AI model names across many sites when providers update or deprecate models.

THE PROBLEM

Model names change. Your code should not.

Providers rename, deprecate, and release models constantly. ModelAlias gives your app one stable name in code, then lets you update the real model from a dashboard — no deploy required.

Update once, every app stays current
Works with any provider: OpenAI, Anthropic, xAI, and more
No proxy or SDK lock-in — just a plain-text name lookup

How it works

Three steps. No SDK. No magic.

Use stable names like coding, cheap, and support across your apps, then change the real model behind them anytime.

01

Create your aliases

In your dashboard, map a short name to a real model identifier. One alias can point to any provider.

coding→claude-opus-4-6
02

Call the resolve endpoint

Your code makes a single GET request with your API key. ModelAlias returns the model name as plain text.

cheap→gpt-4.1-mini
03

Switch models without redeploying

Update the alias in your dashboard. Every app using that alias picks up the new value immediately — zero code changes.

support→gemini-2-flash
AliasResolved modelcodingRESOLVES TOclaude-opus-4-6

Pricing

One simple plan for stable model aliases.

DEVELOPER
$3/ month
  • 5 aliases
  • 1 API key
  • Reasonable rate limits
  • Plain-text resolve endpoint
  • Dashboard access