Skip to main content
Astrolabe Cloud exposes an OpenAI-compatible hosted gateway at: 
https://api.astrolabe.run/v1
The gateway supports both Responses and Chat Completions while applying the same Cloud auth, balance, stack, routing, billing, and trace runtime.

​
Endpoints

  • GET /health
  • GET /ready
  • GET /v1/models
  • GET /v1/lanes
  • POST /v1/responses
  • POST /v1/chat/completions

​
Authentication

Send an Astrolabe Cloud workspace API key with one of: 
Authorization: Bearer $ASTROLABE_CLOUD_KEY

x-api-key: $ASTROLABE_CLOUD_KEY
Requests without a valid active key are rejected.

​
Models

Use virtual model IDs for most requests:
  • astrolabe/auto
  • astrolabe/coding
  • astrolabe/research
  • astrolabe/vision
  • astrolabe/strict-json
  • astrolabe/cheap
  • astrolabe/safe
Use GET /v1/models to inspect the public catalog.

​
Stack metadata

Select a stack and set per-request preferences with metadata.astrolabe: 
{
  "metadata": {
    "astrolabe": {
      "stack": "code-agent",
      "quality": "high",
      "cost_mode": "balanced",
      "latency_slo_ms": 3000,
      "verification": "strict",
      "fallback": true,
      "explain": true
    }
  }
}

​
Response headers

Important headers:
  • x-astrolabe-cloud-request-id
  • x-astrolabe-cloud-workspace
  • x-astrolabe-cloud-key
  • x-ratelimit-limit
  • x-ratelimit-remaining
  • x-ratelimit-reset
  • x-astrolabe-model
  • x-astrolabe-stack-id
  • x-astrolabe-stack-name
  • x-astrolabe-stack-version
  • x-astrolabe-policy-status
  • x-astrolabe-decision
  • x-astrolabe-selected-provider
  • x-astrolabe-selected-model

​
Error shape

Errors use: 
{
  "error": {
    "code": "payment_required",
    "message": "Workspace balance must stay above 1.00 USD before routing.",
    "requestId": "req_..."
  }
}
Log the request id for support and route trace lookup.