Get started

Introduction

The RenderThink API transforms interior images: photorealistic rendering, editing, decoration and upscaling. You call it from your server, with an API key, and it responds in JSON.
https://api.renderthink.com/v1
RouteDescription
POST /v1/rendersPhotorealistic render of a draft image.
POST /v1/editsEdit by mask or by written instruction.
POST /v1/decorationsDecoration of a furnished room, with accessories.
POST /v1/upscales×2 or ×4 upscaling, without changing the content of the image.
GET /v1/generationsLists the organization's generations.
GET /v1/generations/{id}Retrieves a generation's state and result.
POST /v1/generations/{id}/cancelCancels a generation that has not started.
GET /v1/stylesLists the decoration styles.
GET /v1/edit-materialsLists the materials for mask editing.
GET /v1/edit-material-categoriesLists the categories of these materials.
  • Request and response bodies are JSON (UTF-8), with field names in snake_case.
  • Every object states its type in the object field: list, generation, style, edit_material, edit_material_category.
  • Dates are Unix timestamps, in seconds (UTC).
  • A creation request responds when the generation ends, or as soon as it is accepted with "background": true (see Generations).
  • Every response has a request-id header, and every error a stable code (see Errors).
  • The API is called from a server only: it does not accept CORS requests, and a key must never be exposed in a browser (see Authentication).
  • Quickstart: a first render, free of charge, with a test key.
  • Pricing: prices per image, charged to a prepaid balance.
  • Resolutions: the 1k, 2k, 4k and 8k output resolutions.
  • Full reference: every route, generated from the OpenAPI document.
  • MCP server: the API from Claude Code, Codex, Cursor, VS Code, Gemini CLI or any other MCP-compatible agent.