Documentation

Call Claude from your own apps.

SubToAPI gives your team one HTTPS API in front of your Claude access: application keys, normalized responses, streaming, tool use and usage metadata. No SDK required — plain JSON over HTTPS from any language.

Quickstart Start free trial

Your first request

curl -X POST \
  "https://api.subtoapi.app/v1/messages" \
  -H "Authorization: Bearer $SUBTOAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "system": "You are a helpful assistant.",
    "messages": [
      { "role": "user", "content": "Hello" }
    ],
    "model": "balanced"
  }'

Getting started

QuickstartCreate a SubToAPI account, connect Claude once, create an application API key and send your first request to https://api.subtoapi.app/v1/messages — with curl, TypeScript, Python or PHP.AuthenticationHow SubToAPI application keys work: the Authorization header, creating, rotating and revoking sub_live keys, and how to keep them safe.ModelsSubToAPI exposes three stable Claude tiers instead of raw model ids. What each alias is for, how to pick one, and how to list them with GET /v1/models.

API reference

MessagesReference for the SubToAPI messages endpoint: request fields, model aliases, JSON mode, thinking budget and the normalized response with token usage.ConversationSend the whole thread each time: alternating user and assistant turns, content blocks, temperature and tools. Reference and examples for the SubToAPI conversation endpoint.StreamingPOST /v1/conversation/stream returns server-sent events: response.started, response.completed and error. How to consume the stream from curl, Node and the browser.Status & monitoringCheck entitlement, plan and Claude connection state from your own monitoring with a single authenticated request. Response format and alerting tips.Errors & limitsEvery SubToAPI error is JSON with error, message and request_id. Status codes, what they mean, request size limits, per-plan rate limits and a retry strategy with backoff.

Guides

Tool useDefine tools with a JSON Schema, let the model decide when to call them, and send tool results back in the next turn. Complete round-trip example over the SubToAPI conversation endpoint.