Video generation API

Make, render and publish video by API

The Draft API opens the same video features the web editor and MCP use over REST. No render servers, TTS, stock or upload integrations to wire up yourself.

Why a torn QR code still scans · Made in Claude Code with the Draft MCP server

What you'd integrate building video yourself

  • Render servers
  • TTS API
  • Stock and generation APIs
  • Caption alignment
  • Social publishing APIs
  • Job queue and retries

In Draft, one place

Web editor plus one MCP connection
  • Plan and script
  • AI voice
  • Footage
  • Motion graphics
  • Captions
  • Rendering
  • Upload and scheduling
  • Automation

Why teams pick Draft

Cheap with blocks

Scenes are placed from a block name and data instead of generated as code. Block scenes have no generation cost, and restyling is free.

One API, end to end

Projects, uploads, narration, editing, rendering and social publishing through one API and one credit balance.

An engine that keeps improving

New models and render improvements land on Draft's side, so quality and value rise without changing your integration.

How it works

  1. 1

    Create an API key

    Create a key in the dashboard's API settings. Permissions and credits are scoped to a space (business).

  2. 2

    Read the docs

    Find project, production, render and social publishing endpoints in the OpenAPI document.

  3. 3

    Make and render

    Create a project, place blocks and material, request a render and poll its status.

  4. 4

    Publish or download

    Download the MP4, or post it to connected channels now or on a schedule.

What changes with Draft

Instead of finding and wiring tools together, finish in one service that keeps getting better.

No more tool hunting

Script, AI voice, stock and generated footage, editing, rendering, publishing and automation in one account with one credit balance. No sign-ups and API keys per service.

Nothing to install, always improving

No skills or open-source projects to download and configure. As new models and techniques arrive, Draft keeps raising quality and value.

Blocks save tokens and cost

Our own block registry (34 blocks, 3 styles) means scenes aren't coded from scratch. Send a block name and copy, the server draws it, and block scenes and restyling are free.

Same benefits however you work

MCP from Claude, ChatGPT or Codex, the REST API, automations or the web editor all use the same blocks, rendering and publishing.

Coding each scene yourself

Dozens to hundreds of lines of HTML, CSS and GSAP per scene

With a Draft block

Just a block name and copy. The server handles drawing, animation, captions and transitions

{
  "type": "number",
  "from": "208 billion switches",
  "data": {
    "numbers": [
      { "value": "208", "unit": "B", "label": "One B200 chip" }
    ]
  }
}

Which one should you pick?

When another approach is better

  • You must run the render engine on your own servers: an open-source framework
  • You need video structured as React components frame by frame: a framework like Remotion
  • You need an API for avatar presenter video: an avatar specialist

When the Draft API fits

  • You want to add video generation to your product quickly
  • You'd rather not integrate TTS, stock, rendering and publishing separately
  • You want lower scene cost at volume
  • You also want the same flow as a no-code automation
Start free

FAQ

How far does the API go?

Project creation, uploads, editing and block placement, cost estimates, production runs, rendering and export, and social publishing: the features the editor and MCP use, under the same contract. The full list is in the OpenAPI document.

How much do blocks save?

Generating a scene in HTML, CSS and GSAP takes model calls and long code; a block takes short data like a name and copy, and the server draws it. Block scenes themselves have no generation cost, leaving only narration and material.

How do the API, MCP and automations differ?

They're different ways to call the same features: the API from code, MCP from Claude, ChatGPT or Codex, automations on a schedule. Results all land in the same projects and editor.

How is usage billed?

Credits are deducted from the API key's space for the work you run. Check estimated credits with the estimate endpoint before producing.

Can we show renders in our own product?

Yes. When a render finishes, take the MP4 into your own storage or player.

What people with the same question read next.

Make your first video now

Start free with no subscription and top up credits only when you need them.

Start free