Skip to main content
Astrolabe Cloud is a managed AI API operating layer. It gives teams one hosted gateway for routing requests across models with configurable quality, cost, latency, fallback, and policy controls. Use Astrolabe Cloud when you want the application-facing API to stay simple while routing, billing, and model operations stay centralized.

​
What Astrolabe Cloud owns

  • Hosted OpenAI-compatible gateway at https://api.astrolabe.run/v1
  • Workspace API keys minted in the Cloud app
  • Prepaid credit balance and usage ledger
  • Managed routing stack templates
  • Custom routing stacks with policy boundaries and objectives
  • Per-request stack selection through metadata.astrolabe.stack
  • Route traces, response headers, and model-call records

​
Request flow

  1. Your app sends a request with an Astrolabe Cloud API key.
  2. The gateway authenticates the key and checks workspace status.
  3. Astrolabe reserves prepaid balance before provider work starts.
  4. The gateway resolves a routing stack from request, key, workspace, or managed default settings.
  5. The stack chooses an allowed serving model and fallback behavior.
  6. OpenRouter executes the selected model call.
  7. Astrolabe finalizes usage, updates billing, stores traces, and returns routing headers.

​
Default virtual models

Use virtual models when you want Astrolabe to choose the concrete provider model:
  • astrolabe/auto
  • astrolabe/coding
  • astrolabe/research
  • astrolabe/vision
  • astrolabe/strict-json
  • astrolabe/cheap
  • astrolabe/safe
You can also request an allowed concrete model ID directly. Routing stacks still apply policy boundaries before the request leaves the gateway.

Quickstart

Create a workspace key, add credits, and make your first hosted request.

Routing stacks

Learn how managed and custom stacks control model selection.

Customize stacks

Set model boundaries, quality targets, cost modes, fallback, verification, and trace behavior.

Call a stack

Select a specific stack in metadata.astrolabe.stack.

Model roster

Review public virtual models and hosted model pricing roles.

API reference

Check gateway endpoints, auth, headers, and error shapes.